Laser PCB Milling Machine - Department of Electrical Engineering

Laser PCB Milling Machine
Group 18
Nathan Bodnar
David Dowdle
Ryan Maticka
Spring 2010
May 3, 2010
Table of Contents:
SECTION 1: Executive Summary ...................................................................................... 1 SECTION 2: Motivation ..................................................................................................... 2 SECTION 3: Technical Objectives and Goals.................................................................... 3 SECTION 4: Specifications and Requirements .................................................................. 4 SECTION 5: Research and Investigations .......................................................................... 7 5.1 – Burn Test ............................................................................................................................. 7 5.2 – Laser Safety ......................................................................................................................... 9 5.3 – Air Scrubbing ..................................................................................................................... 11 5.4 – Similar products ................................................................................................................ 12 SECTION 6: Laser System ............................................................................................... 13 6.1 – Types of Lasers .................................................................................................................. 13 6.1.1 – Previous Design ............................................................................................................. 14 6.1.2 – Current Design ............................................................................................................... 21 6.2 – Prototype/Test .................................................................................................................. 21 6.3 – Build .................................................................................................................................. 23 6.4 – Evaluation plan ................................................................................................................. 23 SECTION 7: Laser Power Supply Unit ............................................................................ 25 7.1 – Previous Design................................................................................................................. 25 7.1.1 – Rectification ................................................................................................................... 25 7.1.2 – DC to DC Converter ........................................................................................................ 26 7.1.3 – Switching ........................................................................................................................ 26 7.1.4 – LC filter ........................................................................................................................... 28 7.1.5 – Controller ....................................................................................................................... 30 7.2 – Current Design .................................................................................................................. 31 7.2.1 – DC to DC Converter ........................................................................................................ 31 7.2.2 – Switching ........................................................................................................................ 32 7.2.3 – Controller ....................................................................................................................... 33 7.3 – Prototype/Test .................................................................................................................. 33 7.4 – Build .................................................................................................................................. 34 7.5 – Evaluation Plan ................................................................................................................. 34 SECTION 8: TEC Power Supply Unit ............................................................................. 35 8.1 – Design ............................................................................................................................... 35 i
8.1.1 – Rectification and DC to DC Conversion .......................................................................... 35 8.1.2 – Cooling Control .............................................................................................................. 35 8.2 – Prototype/Test .................................................................................................................. 36 8.3 – Build .................................................................................................................................. 36 8.4 – Evaluation Plan ................................................................................................................. 37 SECTION 9: Q Switch...................................................................................................... 38 9.1 – Types of Q Switches .......................................................................................................... 38 9.2 – Previous Design................................................................................................................. 40 9.2.1 – Q Switch High Voltage Power Supply ............................................................................ 41 9.2.2 – Q switch Fast FETs ......................................................................................................... 47 9.3 – Current Design .................................................................................................................. 49 9.4 – Prototype/Test .................................................................................................................. 49 9.5 – Build .................................................................................................................................. 51 9.6 – Evaluation Plan ................................................................................................................. 51 SECTION 10: Motors and Motor Controller Unit ............................................................ 52 10.1 – Design ............................................................................................................................. 52 10.2 – Prototyping and Testing ................................................................................................. 56 10.3 – Building ........................................................................................................................... 56 10.4 – Evaluation Plan ............................................................................................................... 56 SECTION 11: XY Table ................................................................................................... 58 11.1 – Design ............................................................................................................................. 58 11.2 – Prototype/Test ................................................................................................................ 60 11.3 – Evaluation Plan ............................................................................................................... 61 SECTION 12: Microcontroller ......................................................................................... 62 12.1 – Design ............................................................................................................................. 62 12.2 – Prototype/Test ................................................................................................................ 66 12.3 – Build ................................................................................................................................ 68 SECTION 13: Software .................................................................................................... 69 13.1 – Design ............................................................................................................................. 69 13.2 – Prototype/Test ................................................................................................................ 91 13.3 – Evaluation Plan ............................................................................................................... 93 SECTION 14: Security ..................................................................................................... 95 SECTION 15: Personnel ................................................................................................... 98 15.1 – Nathan Bodnar ................................................................................................................ 98 ii
15.2 – David Dowdle .................................................................................................................. 98 15.3 – Ryan Maticka .................................................................................................................. 98 SECTION 16: Bibliography ............................................................................................. 99 SECTION 17: Facilities and Consultants ....................................................................... 102 SECTION 19: Budget and Financing ............................................................................. 103 SECTION 20: Milestone................................................................................................. 104 SECTION 21: Project Summary ..................................................................................... 105 APPENDIX A: Copyright Notices ................................................................................. 108 A.1 – jPicUsb License ............................................................................................................... 108 A.2 – gerb2tiff License ............................................................................................................. 108 A.3 – ImageMagick License ...................................................................................................... 114 APPENDIX B: Example DRL File................................................................................. 116 APPENDIX C: Password Generation Code ................................................................... 117 APPENDIX D: jPicUsb Documentation ........................................................................ 118 iii
SECTION 1: Executive Summary
This goal of this project was to replace the current milling machine used by
senior design students. The main reasons that motivated us for replacing the
current milling machine are the cost factor for the replacement of the milling bits
and its safety. These cost factors, along with the amount of user errors need to
be improved. Therefore, this project utilizes a new design using a high powered
Nd:YAG laser. The laser light is used to mill the printed circuit boards allowing for
tool-less machining. This reduced the cost load of maintaining a rapid printed
circuit board prototyping machine. This process will also avoid some of the safety
issues involved with the old milling machine, which included having fiberglass
debris flying everywhere and the openness of the rotary blades cutting the
copper clad boards.
This project consists of building the high powered Q Switched second harmonic
Nd:YAG laser and the control units to power it. The project also includes the
designing and building of a XY table to move the laser light to the target copper
substrate. The designs for this venture also incorporate safety throughout the
electronics. This will prevent the users from causing damage to the machine and
to protect the user. The software is engineered to allow the user to engrave away
the copper using either a Gerber file or a bitmap Portable Network Graphics
(PNG) file. This will allow the senior design student to have the opportunity to
utilize any design software that they choose and be able to mill their circuit board.
This is something that the current milling machine cannot do. The framing of the
machine incorporates a safety viewing window that will allow the senior design
student to view their work as the laser mills the board out.
The cost of the parts for this project has been funded by the group. Attempts
were made to try to acquire donations for the parts required to build this machine.
The budget for this adventure was around three to eight thousand dollars. The
budget was estimated from what the parts would cost if they were bought new
from the manufacture. The team worked to acquire the materials through used
and surpluses stores and through the online provider of eBay. These materials
greatly reduced the cost of this project.
1
SECTION 2: Motivation
The motivation for this laser PCB etcher is the constant down time of the TTech
Quick Circuit 5000 PCB milling machine in the senior design lab. As two of our
group are ARC members, and are tasked with keeping the machine repaired and
running, we have found this to be an increasingly difficult task. In just the last
year we have had to replace the lead screw, and are currently in the process of
replacing the main spindle motor. These problems resulted from two areas:
ignorance and fatigue. The lead screw only needed to be replaced because
someone thought it would be a good idea to oil it. As the very large stickers warn,
lead screws are not meant to be oiled. If the plastic shavings that were
everywhere was not an indication that something was wrong, the horrible
screeching noise would have. This case of ignorance on the part of a senior
design student caused several months of down time as we needed to first search
for the part, find funding for the part, and have that part replaced. The spindle
motor has simply worn out because of fatigue. These things happen, but this
particular part is very expensive and has been particularly difficult to find funding
for. Our project intends to take care of both of these problems. It will be left with
the amateur radio club when we are finished to be used by future senior design
students. Ignorance has been solved through security and training. The current
milling machine is in a locked room, inside another locked room, but this was not
enough to keep it from being damaged. Our project will take this further in that
our project is completely contained and monitored. There is no interaction with
the machine without first having enrolled in senior design, and being certified by
an ARC member. The certification will only come at the end of an intensive
training session that we will develop as part of the project. At this time the user is
granted a username and password that will determine their level of access to the
machine. We are taking a hierarchical approach to security in this regard. No one
will have access to a part of the machine that they do not need access to, and
have not been trained in the proper use and maintenance of. For the standard
senior design student this will equate to being able to open the main lid, and
nothing else. The compartments storing the lasers, most of the optics and all of
the electronics is sealed. Fatigue is solved through less moving parts. In fact, the
only moving part of our project is the XY table to do the etching. The XY table,
unlike the current system, is driven by stepper motors attached to a belt system.
The motors were purchased from an online retailer for a reasonable price. In this
project, the main thing that can break is the laser parts. Spare parts were
purchased that will be required to keep our project in operation for many years.
2
SECTION 3: Technical Objectives and Goals
Our goals for this project were that the finished machine is safe, reliable, and
timely. This was accomplished through the extensive use of sensors for
everything from an overheated diode to a user trying to force the machine open
at the wrong time. Mediation schemes are in place for as many scenarios that we
can think of. When higher level reasoning is required, the information is passed
off to the computer based software portion of our project and the decision
relayed. This is only used for the less immediate concerns, however. All of the
lower level decisions are computed on chip in the project itself. Any part that
goes out of specification or if a sensor requires immediate attention, the entire
project will enter a programmed shutdown that will attempt to reduce the negative
effects on the hardware. This is the default case if an unexpected error is
reached. We attempted to meet the goal of reliability through the use of high
quality components and through the reduction of moving parts. The use of a laser
will further meet the goals of reliability and timeliness. By the very nature of our
design, we do not have any need to change bits. This allows us to reach our
reliability goal in that very few people will be able to access the laser portion of
the project. As with any system, the less people that have access to it the less
likely someone is going to try and tamper with it. In the current setup, everyone
has access to the bits and often load them wrong, or chose the wrong one for the
current job. While it is true that our project did take a little over double the amount
of time as the current machine to remove all the copper from a standard sized
board, we have the advantage of not requiring a user’s constant input. It will be
entirely possible for a user to start the etch process in the morning, go to all of his
or her classes and come back at a later time to find one entire side of their board
finished. All they will need to do is flip the board over at the appropriate time.
The function of this project is to replace the current milling machine used by
senior design students. This is accomplished through the use of a focused high
powered green laser. Furthermore, our project can not only be able to etch the
copper off of the printed circuit boards, but can also cut the fiber glass substrate
of the boards. This avoids the safety issues of using a shear or rotary saw to cut
prototypes out of the master board and allow holes for through hole components
to be made during the main etch process.
3
SECTION 4: Specifications and Requirements
The specifications outlined below were defined in large part to satisfy the desires
of our sponsor. The milling machine has features appropriate for use in an
undergraduate setting for senior design students.
The software of this project is quite significant. It is able to handle communicating
with microcontrollers, large input files, and running on a wide variety of hardware.
It is also be responsible for providing users and technicians with an interface to
work with or on the machine.
• Software
o XY Table
ƒ Allow positioning of board
ƒ Import images
ƒ Import Gerber
ƒ Store previous boards
ƒ Remember previous board cuts
ƒ Checks safety measures
o GUI for Laser
ƒ Standard Users (Senior Design Students)
• Allows the milling of FR-4 boards only
• Allow only two boards to be milled
• Cannot add boards
• Must follow all safety measures
• Cannot add any user
ƒ Advanced Users (non-Senior Design Users)
• Users may have privileges from some but not all of
advanced users settings
• Can be created from Administrators or Experienced
Users
• Cannot add any users
ƒ Experienced Users (Arc Trainers)
• Allow the addition of boards
• Allow the modification of unique sized boards
• Allow control of laser power
• Allow the selection of different PCB materials
• Must follow all safety measures
• Can add Standard and Advanced Users
• Can remove any Standard or Advanced User
• Can change user access to all but Administrator and
Experienced
ƒ Administrator (Laser Experienced Service Technicians)
• All safety measures can be over written
4
•
•
Can create all users
Can remove a user or change access level
The XY table is the only main moving part of this project. This table is
responsible for moving the mirrors and lenses around, collectively called the
milling head, and is able to reach the entire board.
• XY Table
o Belt driven
o Stepper motors with encoders for positions
o Capable of a maximum board size of 12 in by 12 in
The laser is the heart of this project. Copper is removed from the printed circuit
boards through the power of this laser and the supporting structures alone.
Therefore, a laser is used that is powerful, relatively cheap, and easy to find to
allow for backups to be reasonably attainable.
• Laser
o 5 W @ 532 nm
o 15 W @ 1064 nm
o Q-switched
o Solid State Diode 808 nm, 40 W
o Nd:YAG
o KTP (KD*P)
• Optics
o Allow beam waste to be 1 mil
o Allow Rz to be 0.25 in
o Directing mirrors to be 532 nm HR 45º, 1064 nm HR 45º
o Automatic focusing
o Automatic board positioning and aligning
o Automatic removal checker
o Pixel burning @ 1 kHz
o 1 to 2 mm pixel size
• Laser Power Supply
o Input: 120 V AC, 60 Hz
o Output: 730 V DC
o Current controlled
o Monitoring current
o Monitoring laser output
• Q-switch Power Supply
o Pockel Cell
ƒ 2.5 kV DC
ƒ -2.5 kV DC
ƒ Pulsed with microcontroller
• Stepper Motor Power Supply
o 12 V DC
5
o H-bridge configuration
• Power Management Circuit
o Turn on specific devices
o Controlled by microcontroller
o Emergency power shutdown
6
SECTION 5: Research and Investigations
5.1 – Burn Test
Estimating the amount of power required to generate the plasma required to
vaporize the copper substrate of a printed circuit board will help in designing the
laser for this project. An equation was created to estimate the amount of energy
to vaporize the copper substrate. This was created by calculating the volume of
the copper target. The next step would be to multiply this value with the density.
This value was then multiplied by the molar mass and the heat of vaporization
constant for copper. This value was then used to estimate the total energy
required to vaporize the copper. This value was then multiplied with the
absorption spectrum and the graph below shows the amount of energy to
vaporize the copper substrate depending on the wavelength (Figure 5.1).
Energy to Vaporize Copper
35
30
Power (mJ)
25
20
15
10
5
0
1048
848
648
448
248
W avelength (nm)
Figure 5.1: Graph of energy required to vaporize copper for given wavelength
A testing laser was used to verify the graph. The laser used was a second
harmonic Nd:YAG laser. The laser was flash tube based with a pulse duration of
around 10 ns. The pulse energy ranged from 0.7 mJ to around 2 mJ. The laser
was the YG580 series Active Q switched Nd:YAG laser from Quantel. The pulse
rate was 10 pulses per second.
The target that was used was a printed circuit board that was mounted in a stage
to allow minimal movement. This would allow multiple shots to hit the same
location. The focusing lens that was used was a convex lens with a focal point of
around 100 millimeters. The laser diameter from leaving the laser cavity was
about one centimeter in diameter. The target was placed at the focal point and
7
the laser the laser pulses were counted until the laser light exited through the
copper target. The number of pulses that were required to vaporize the copper
target was about 40 pulses for the 0.7 millijoules and around 30 for the two
millijoules. The following photo is the profile of the pulses of the laser hitting the
copper target. The profile goes form a single pulse at the right of the image to
five pulses, then to ten pulses and continuing this iteration to up to 25 pulses
(Figure 5.2).
Figure 5.2: Burn testing with second harmonics of Nd:YAG
Looking at a close view of one of the shots, it seems that the laser pulse
generated copper whiskers. The width of one of the widest whisker measured to
be around one micron. The comparison of this size to the size of the hole is
around 25 times its thickest width. This would be apparent that these whiskers
should not be a problem for the milling process of the printed circuit boards. The
following image is the crater that was generated from having the laser pulse
hitting the printed circuit board (Figure 5.3). It can also be seen that for one of the
firings that there was a misalignment and another pulse was shot below the
original spot.
Figure 5.3: Close up view of the burn test
8
5.2 – Laser Safety
The laser system that this project will employ is a class four laser. When the laser
is enclosed in a sealed environment that has interlocks and other preventive
measures, the laser class will drop to a class one. When the laser is being
serviced the laser will change its class back to a four. This is what is
characterized by the Federal Laser Product Performance Standard and the ANSI
Z136.1. There are numerous signs that will need to be placed along the project.
The first sign will indicate that the project is a Class one laser. This is a yellow
and black caution sign. The following image will show the layout of this sign
(Figure 5.4).
! CAUTION
LASER RADIATION
DO NOT STARE INTO BEAM
Enclosed Nd:YAG, 532nm, 10mJ, 40ns
CLASS 1 LASER
Figure 5.4: Laser Safety Label for exterior case of project
The next sign that is placed is in the material processing area. This will indicate
that a class 2A laser is in progress. This is because when aligning up the printed
circuit board a low powered 730 nanometer diode laser is present. This will help
in indicating the header position. The following image will show the sign that is
placed in the material processing area (Figure 5.5).
! CAUTION
LASER RADIATION
DO NOT STARE INTO BEAM
LASER DIODE, 730nm, 100mW
CLASS 2 LASER
Figure 5.5: Laser Safety Label for Material Processing area
The next signs that will need to be placed in the project are in the high powered
laser section. Within this section the laser class is a class four. This section will
9
need to have a sign that indicates the laser wavelength output and the power
output. There will also need to be a laser sign for the output of the oscillator. The
following images are the signs for this section of the laser (Figure 5.6, 5.7).
! DANGER
LASER RADIATION
WEAR PROTECTIVE EYEWEAR
Nd:YAG, 532nm, 10mJ, 40ns
LASER DIODE, 808nm, 80W
CLASS 4 LASER
Figure 5.6: Laser Safety Label for High Power Laser
Figure 5.7: Label at exit of High Powered Laser
There will have to be safety precautions that will need to be followed as given by
the ANSI Z136.1. These safety precautions are necessary to allow the high
powered laser to be run as a class one. The following procedures are to run the
system as a class one laser. There has to be a protective housing, interlocks on
the protective housing, a service access panel, and equipment labels. To run the
laser as a class four, the following control measures must be followed. A
protective housing, interlocks, service access panel, key control, removable
interlock connector, beam stop or attenuator, Activation warning system, laser
controlled area, equipment labels, and laser area warning signs.
One of the precautions in working with the high powered laser is to wear laser
safety glasses. The rating that the glasses that need to be worn are the following
ratings that were generated from the ANSI Z136.1 Standard. The OD rating is
five OD for 532 nanometers, five OD for 1064 nanometers and greater than four
optical densities for 808 nanometers.
10
Since there is high voltage for the Q switch there will have to be warning signs for
the high voltage terminal of the Q switch and the power supply that drives the Q
switch. The warning sign is as followed (Figure 5.8).
DANGER
HIGH
VOLTAGE
Figure 5.8: Warning Sign for the High Voltages need for the Q switch
5.3 – Air Scrubbing
Since the process in generating vaporized copper plasma creates particulates
that can be considered as a hazard to the health of the user, the vapors from this
process must be processed in a manor that will protect the user. There are two
types of scrubbing schemes used to purify the pollutants from the system. The
fist type of system is illustrated below; this is a wet scrubber (Figure 5.9).
Figure 5.9: Wet scrubbing of polluted air (Permission granted Wikipedia.com)
11
This process works by injecting the contaminated air with small water droplets.
The droplets generate a static charge and pull debris out of the air. The biggest
problem with this type of system is that the partials that can be collected range
for 20 microns down to two microns [1]. This process will also generate a lot of
waste water and give off a great deal of moisture. This process would work great
for systems were the output of the system exits to the outside of the building. For
this project this process would not work. The next option would be to use dry
scrubbers. The main compound for collecting the pollutants generated through
this system would be active carbon. The only downside of using active carbon is
that the larger particles will clog the scrubber and reduce the efficiency. The best
method would be to include a pre-filter that would collect the larger debris. For
the initial design of this project there is no filtration during this stage. During the
second revision then this system may be employed.
5.4 – Similar products
While researching this topic there were some manufactures that currently use
lasers to mill printed circuit boards. The leading manufacture for this is LPKF [2].
There system uses a 1064 nanometer laser to etch the boards. The system that
they use also using gravimeters to move the laser light to the target on the
copper clad board. Another alternative to etching the copper substrate from the
board is to coat the boards with an opaque polymer and using a CO2 laser to
remove the polymer. The newly prepared boards are then chemically etched.
This process is something that this group currently employs in creating circuit
boards. Another method of creating the printed circuit boards is to use
mechanical methods to remove the copper substrate. This is the method that is
currently employed for the senior design students. In previous senior design
groups a similar project was created [3]. There project consisted of taking a one
watt 808 nanometer laser diode and attaching it to an XY table. When etching
occurred the XY table would move the laser and its optics to the location and the
laser would turn on. This project compared to the previous project is that the
pulse energy is much greater. The power of the laser diode that is being used for
this projects laser system compared to that of the previous system is about 80
times greater. This will allow this laser system to be able to mill away the copper
substrate.
12
SECTION 6: Laser System
6.1 – Types of Lasers
The two common lasers that are used in industry for material processing are the
CO2 laser and the Nd:YAG laser. The CO2 laser works by exciting a mixture of
gas that contains CO2 with a high voltage or with microwaves. The output of the
CO2 laser is around ten microns. For copper this wavelength is over 98 percent
reflective. Using a CO2 laser would require very high power design to allow the
two percent that is absorbed to vaporize the copper target. The Nd:YAG laser is
created by exciting a neodymium doped rod of the crystal yttrium aluminum
garnet. The main output frequency that this laser generates is 1064 nanometers.
This too, is highly reflective for copper with a reflectivity of 96 percent. The big
advantage to the Nd:YAG lasers is that there are nonlinear optics that can
generate harmonics of the fundamental wavelength of this laser. For the second
harmonic of 532 nanometers, the reflectivity of copper drops down to 45 percent.
This will allow 55 percent of the laser pulse to be absorbed and will allow the
laser to run at much lower powers than if just the fundamental wavelength would
be used. The conversion to generate the second harmonic is not perfect and the
best efficiency in generating the harmonic is 45 percent of the fundamental
wavelength. If the third harmonic of the fundamental would be used the
reflectivity drops down to 35 percent. The only problem with generating this third
harmonic is that the conversion efficiency drops to at best 30 percent of the
fundamental wavelength. For third harmonic generation, the process requires
multiple nonlinear optics to generate it. The reason for multiple optics is that the
first optic generates the second harmonic and the second nonlinear optic takes
the second harmonic and the fundamental and generates the third harmonic.
This process becomes even less efficient because it is almost imposable to
generate in that laser cavity. So to first generate the fundamental the laser cavity
will need to be pumped over 90 percent higher than if the third harmonic would
be built in the laser cavity, this is because the standard output coupler is 90
percent reflective. Since the second harmonic can be created in the cavity, it is
therefore more efficient to generate the second harmonic in the cavity at a higher
power than to user the third harmonic. Nd:YAG lasers can also generate even
higher order harmonics, but these conversions waist a numerous amount of the
fundamental wavelength energy and are not used for material processing.
A new development is being implemented in generating high powered fiber
lasers. This design consist of fiber that is doped with some rare earth metals that
when excited by a pump laser and placed in a cavity they can generate high
powered lasers. The most common fiber doped laser is the Ti:Sapphire. This
laser has an output wavelength of around 790 nanometers. The biggest problem
with Ti:Sapphire laser systems is that the fiber is hydroscopic and over a short
time will need to be replaced regularly. Another type of fiber laser schemes is to
use thulium doped fiber. This fiber system generates a wavelength of around two
13
microns. The biggest drawback to this fiber system is that the cost of the fiber
being used costs around a thousand dollars per foot. The second drawback is
that the laser is in the infrared wavelength that cannot be detectable using
standard IR viewers or standard IR fluorescents cards.
There is a new type of laser system that is called thin disk laser system. The
leading developer in this laser system technology is Trumpf. Thin disk laser
system works by generating the doping material to be a 100 micron thick disk.
On one side of the disk is a very precise temperature controller. If used, this
would generate a higher conversion efficiency. The reason for this higher
conversion efficiency is that it reduces the thermal lensing and controls the
crystal’s temperature to the ideal range for the conversion of the pump light to the
laser’s output light. The doping material that is used for this system is Yb:YAG.
This crystal structure generates a wavelength of 1060 nanometers. This is very
close the Nd:YAG laser wavelength and this would allow many of the same
optics to be used. This would also include the second harmonic crystals. The
biggest downside of this system is that because this is a new process, the cost of
the thin crystal is very high and the means of accurately controlling the
temperature of the disk would be costly as well.
The last alternative in laser system design would be to use the laser diode as the
fundamental laser and generating the second harmonic wavelength by using
another second harmonic generation crystal. The downside of using this schema
is that the energy output will reproduce the same as if the laser was running in
continues wave mode. Comparing this design in continues wave configuration,
the conversion efficiency is very high compared to Nd:YAG. The difference
comes when the system is pulsed. Unlike Nd:YAG, this system cannot be Q
switched. This will prevent the crystal from dumping all of its energy at once. So
to compare the laser output energy, if 80 watts of laser diodes would be used the
output energy is around 3.2 microjoules. Comparing this to a Nd:YAG laser the
output would be about a thousand time greater.
6.1.1 – Previous Design
The laser design used for this project will consist of a second harmonic Nd:YAG
Q-switched laser. The second harmonic generation is generated in the cavity.
The following layout will show the design of the laser system (Figure 6.1).
14
HR:1064
HT:808
Nd:YAG
HR:1064
HT:808
Diode
Diode
Polarizer
HT:1064
HR:532
Q-switch
¼ Wave
HR:1064
KTP
HR:1064
HR:532
Figure 6.1: Layout of Laser system
To generate the fundamental wavelength, a Nd:YAG rod is end pumped with one
to two 808 nanometer laser diodes. The diodes chosen for this project are fiber
coupled diodes. The output of each diode is rated for 40 watts of laser light
output. The manufacture of the diodes is Coherent. The reason that these diodes
were used was because they are commonly available on eBay. If the diodes
were bought from the manufacture, they would have cost over ten thousand
dollars each. Also the advantage of using fiber coupled diodes is that the optics
required to generate a round beam is much easer than if a non fibered coupled
diode is used. The reason is that in order to generate the high power output the
diode must be multiple diodes stacked on top of one another. The shape of the
output of the stacked diodes reproduces exactly how the diodes were stacked.
This would require many optical elements to fix the shape of the beam and the
addition of each optic would degrade the power output. The design is to first use
only one diode, and if the power output is too low then a second diode is added
to increase the pump power. When the two diodes are setup, it is important to
prevent the diodes form having the pump laser light from one diode enter the
other diode. So when the system is setup, the focal point of the diodes is off
center of the crystal rod. This will prevent a large amount of the pumping light
that exits the crystal from hitting the other diode.
The Nd:YAG rod that is used for this laser was collected from a old range finder
that was used on a M1 tank. The rod was bought also from eBay. An alternative
crystal that could have been used for this system is the Nd:YVO4. This crystal
has the great advantage to have an acceptance pump wavelength that is much
wider than Nd:YAG. The disadvantage to this crystal is the pricing for just a small
crystal us about four times the cost for a larger Nd:YAG rod. Another
disadvantage to this crystal is that the crystal is polarization dependent. The
common practice for using Nd:YVO4 is to use it as a low power oscillator and to
15
then use a Nd:YAG rod for an amplifier. This would increase the complexity of
this projects design, therefore it is better to stick with using one Nd:YAG than the
Nd:YVO4 crystal. The doping proportion of the neodymium is unknown, but
comparing it to other rods, it looks to be around 0.7 percent. This is the standard
doping used for diode pumping. The Nd:YAG rod will also need to be cooled
when it is being pumped this is because the conversion efficiency from the
pumping laser light to the laser’s fundamental light is around 45 percent. The
crystal is mounted in an aluminum block that will water cooled. As a safety
precaution the aluminum heat sink’s temperature will also be monitored. This
aluminum heat sink will need to also be about a quarter of an inch longer than
the Nd:YAG rod. The reason for the increase in length is that the crystal will start
to generate thermal lensing at the ends of the rod. If the heat sink was not longer
than the rod the crystal will destroy itself by fracturing at the ends.
To generate the second harmonic wavelength a nonlinear crystal is used. There
are about four different crystals that are used for this generation. The easiest
nonlinear crystal to grow is the Potassium Dideuterium Phosphate (KDP). This
crystal is the cheapest second harmonic generation crystal. The downside of
using this type of crystal is that it is hydroscopic and the conversion efficiency in
generating the second harmonic is temperature dependent. The lasers that use
this type of crystal normally need to be continually purged with desiccated
nitrogen, or the crystal is placed in a small box that has anti-reflective windows
and filled with expensive index matching fluid. The crystal would also need to
have a peltier on it to regulate the temperature. If the humidity of the crystal is not
low the crystal will start to fog over and the crystal could easily be destroyed by
putting minimal pressure on it. This pressure would be the pressure of just
mounting the crystal in a heat sink. The crystal also has the issue of aging. When
the KDP crystal ages, it darkens and reduces the power output [4].
Another second harmonic generation crystal is the Lithium Triborate (LBO). It
does not have the setback of KDP when it comes to being hydroscopic. The LBO
crystal also has a conversion efficiency of about three times greater than KDP
[5]. The Beta Barium Borate (BBO) second harmonic crystal also is not
hydroscopic and has a conversion efficiency of about double that of LBO and six
time that of the KDP [6]. BBO crystals also do not need to be temperature
controlled like the LBO and KDP because it has a temperate bandwidth of 55
degrees centigrade.
Lastly the commonly used second harmonic crystal used today for the green
laser pointers is the Potassium Titanyl Phosphate (KTP). The reason that this
crystal is used is that it has a conversion efficiency of 15 times greater than the
KDP [7]. It also is not hydroscopic and has a large working temperature range
from 25 degrees centigrade to 80 degrees centigrade. The biggest problem with
KTP crystals is that the cost of generating large crystals is much greater than
KDP. So the general rule is that if a long crystal is needed than a KDP crystal is
used, otherwise the more efficient KTP crystal is used. There was a more
16
recently development method in generating Anti-gray Tracking KTP crystals.
These crystals have about double the damage threshold than the standard KTP
crystal [8]. Without the anti-gray tracking the aging of the crystal over time
generates gray tracks within the crystal, the new crystals are engineered to
reduce this problem and therefore increase the lifetime of the crystal.
For this project a KTP with the anti-gray tracking crystal is used. The crystal
dimensions are three and a half millimeters by six millimeters. The crystal was
also acquired from eBay. The KTP crystal was also coated with anti-reflective
coatings for 1064 and 532 nanometers. To prevent overheating of the crystal it is
placed in an aluminum heat sink. The heat sink will also be water cooled. The
temperature of the heat sink will also be monitored. If the crystal is misaligned
the temperature of the crystal will increase, measuring the temperature of the
heat sink will help diagnose that there is a misalignment. This will also save the
laser cavity because if the crystal heats up past a hundred degrees Fahrenheit
the crystal will fracture.
To protect the diodes from the 1064 nanometer laser light the cavity is folded.
The folding mirrors are coated to reflect the YAG fundamental light and to be
transparent to the laser diodes wavelength. So what happens is that 99.8 percent
of the fundamental light is reflected and the 0.2 percent hits the diode. The
amount of light that hits the diode is a much less than 160 mircojoules. This
amount of energy would not be enough to damage the diode.
The design of the cavity is to allow the second harmonic to be generated in the
cavity. The cavity is folded to allow the pump diodes to be placed for end
pumping the Nd:YAG rod. For the second harmonic generation crystal, it is
placed as close to the second focal point in the cavity. The efficiency of the cavity
is dependent to how much the majority of the YAG rod is in the cavity. To prevent
the cavity from continuing to expand and preventing the second harmonic crystal
from being completely in the beam, a lens was placed in the cavity. The lens is
placed in a location that generates an image of the center of the YAG rod. At the
focal point of the lens a highly reflective mirror is placed. The mirror will need to
be designed for high power laser light for the fundamental wavelength of Nd:YAG
and the second harmonic. The reason is that the cavity is converging to a very
small spot size on the mirror. The mirror’s coating will need to be rated for a
power of greater than 120 joules per centimeter.
The section that is in between the lens and the mirror is the location for the
second harmonic crystal. This will allow the second harmonic crystal to be fully
exposed without letting the edges of the crystal be exposed. The output coupler
for this laser is a special coated mirror that allows the fundamental Nd:YAG laser
light to pass through and reflect the second harmonic light. The second harmonic
laser light that exits the cavity will need to be collimated. This is done with adding
a convex lens on the output sections of the laser cavity. The lens reduces the
beam from diverging. This will help in maintaining the same spot size when the
17
laser is engraving. Since the output power of the laser pulse is very, high the lens
will need to be coated with an antireflective coating. The focal point of the lens
that is used for collimating the output of the laser will have to be determined after
the laser is built. The measurements of the divergence will determine the lens
that is used.
To generate pulsing of the laser a Q switch is implemented. The additional parts
required to generate pulsing of the laser are the polarized beam splitter and a
quarter wave plate. The pulse generation works by first having the Q switch
change the polarity of the laser light by a quarter-wave. Then when the laser light
passes through the quarter wave plate it rotates by another quarter wave. The
light is then bounced back from the cavity mirror and passes back through the
quarter wave plate and the Q switch then changes the polarity back to the
original polarity and passes right through the beam splitter. When the Q switch is
turned off and when the laser light is passed through the system the polarity is
horizontal and when it goes through the beam splitter it is sent out of the cavity.
This prevents the laser from oscillating and the laser is turned off. The two
methods of preventing the polarization from passing are to use a cubic beam
splitter or to use a polarizing filter. The reason that a cubic beam splitter was
chosen was that polarizing filters absorb the entire laser light that is not the
correct polarization. Since the design of this laser is for high power, this would
burn and destroy the polarizing filter very quickly. Using a cubic beam splitter
diverts the horizontally polarized light out of the cavity and allows the vertically
polarized laser light to pass through. Once the light exits from the beam splitter it
can be collected in a water-cooled beam dump. This will prevent damaging the
polarizing optic. The polarizing beam cube that was chosen was designed for
Nd:YAG fundamental wavelength. The manufacture of the polarizing beam
splitter is CVI Lasers. The beam splitter was also bought on eBay at a heavily
reduced price compared to buying it directly form the manufacture. The quarter
wave plate that is used for the project is a Spectra Physics multi order wave plate
with the center frequency of 1064 nanometers. This quarter wave plate was
chosen because it was also available from eBay and the company is very
reliable. The down side of using a multi order wave plate is that the acceptance
wavelength is around one percent. If a zero order plate was used the acceptance
bandwidth is around two percent. The cost of a zero order wave plate is about
double the cost of a single wave plate, therefore a multi order wave plate was
bought.
The cavity mirror that is going to be used for the laser is a highly reflective
concave mirror. The focus point of the mirror is 500 millimeters. The lens that is
used for relaying the center image of the Nd:YAG rod will have a focal point of
800 millimeters. These values were chosen because a model was created using
a computer simulator named LASCAD. This software is designed to take in the
crystal parameters and the different lenses and will generate the cavity stability.
The software will also estimate the output power of the laser cavity. Using the
18
Nd:YAG rod that was specified in the above test, a simulation of the entire cavity
was designed. The following figure is the design of the cavity (Figure 6.2).
Figure 6.2: Cavity deigned in LASCAD
This cavity design was also simulated to be pumped with two end pumping lasers
with a total pump power of 80 watts. The software design was not created for
frequency doubled cavities, but it does follow the same rules for stability. The 11
elements in the software represent the objects of the cavity. The first element
starting at zero is the highly reflective mirror. The second and third elements are
the interfaces of the KTP crystal. The fourth element is the mirror that would
separate the second harmonic from the fundamental wavelength. The fifth and
sixth section is the Nd:YAG crystal. The seventh section is the beam splitter and
the eighth element is the quarter wave plate. The ninth and tenth elements are
the interfaces for the Q switch. The last element is the concave mirror. In
simulating the laser cavity, the concave mirror was chosen to be the output
coupler. The data collected from the simulations shows that this laser cavity will
oscillate and that the output power is mostly exponential by the dependence of
the input pumping power. This can be illustrated in the graph of the input pump
power to the output lasing power (Figure 6.3). This data can be used to estimate
the laser power for the second harmonic generation.
Figure 6.3: Power output Vs Input estimations using LASCAD
19
The software will also simulate the ability to pulse the laser with a Q switch. The
design criteria that were inputted were that the laser would be pulsed for 41
nanoseconds and that the laser pulsing would have a duty cycle at one kilohertz.
The data collected showed that the output power for this laser cavity would
generate pulses of over nine millijoules. If the efficiency of converting the pulse
form the fundamental frequency to the second harmonic is 45 percent then the
output power of the second harmonic pulse should be around four millijoules.
This power output is about six times higher than the minimum amount of energy
required in generating the plasma to vaporize the copper off the circuit boards.
As previous experiments showed that the laser power output of the software is
concretively lower than the experimental outputs. This indicated that this laser
design will generate a laser system that should be able to generate the plasma
needed to vaporize the copper substrate.
The control elements that are required to run the laser correctly need to be
synced together in order to allow the proper operation of the laser. When some of
the elements fail the entire system must act accordingly in order not to damage
the laser system. The following is a flow chart that illustrates how these different
elements interact with one another (Figure 6.4).
Main
Computer
USB
USB
USB
USB
AC Power
AC Power
Thermal Electric Cooler
Laser
Power Supply
Laser
Power Supply
Thermal Electric Cooler
Cooling Lines
Wire
Thermistor
Wire
Thermistor
Cooling
Flow
Cooling
Flow
Laser Diode #1
Laser Diode #1
Wire
Cooling Lines
Cooling Lines
Figure 6.4: Flowchart of Laser system
20
Wire
Cooling Lines
6.1.2 – Current Design
Unfortunately we were incapable of using the design outlined in the previous
section. This is due to the fact that the laser 808 nm laser diode that we
purchased from eBay was out of the listed specifications for it. The diode was
running at 800 nm instead of the 808 nm that it was supposed to run at. As the
Nd:YAG crystal that is part of our cavity design has an acceptance region of 0.6
nm, we would have only been capable of making this design work by actually
heating the diode to a level that is dangerously out of the diodes safe operating
range.
The current design of our laser is a flash tube based system. The main drawback
of using this type of laser system instead of a diode system is that they are at
best 2% efficient, have a low duty cycle, and have a shorter mean time to failure
than a diode system. The flash tube system that we will be using will have a
maximum duty cycle of 100 pulses per second, with a more realistic goal of 10
pulses per second. These pulses consist of containing four nanosecond pulse
durations. The advantage of using a flash tube laser system is that we will have a
higher output power from the oscillator. This will allow us to take fewer shots to
achieve a burn through, but is mainly because we need the higher power
because we are doing the frequency doubling outside of the cavity and our
conversion efficiency is now much lower.
The theoretical power output of the current design will be 15 J per pulse. This will
be obtained because we will have 100 mJ of 1064 nm light per pulse out of the
flash tube cavity. This is our main laser, which will be referenced as the laser
oscillator from now on. We will then use two amplifiers that are both flash tube
based amplifiers. The amplifier’s régime will consist of a single pass amplifier
meaning that the light will enter one port and exit from a different one by only
making a single pass through the excited gain median. The light will then enter a
focusing lens and near the focal point the laser light will enter a KTP crystal. At
the focal point will be a highly reflective mirror that will reflect the beam back
through the KTP for a second pass. After having the light pass through the
nonlinear KTP crystal creating a double pass, the laser light is now converted
into 532 nm green light which is the final light that we will be using for the mill
operation. The light will hit a mirror that is highly reflective for 532 nm and
transmissive for 1064 nm to separate the 532 nm light out of the laser amplifier
system. The 1064 nm light will stay in the laser section and the 532 nm light will
be transmitted to the XY stage.
6.2 – Prototype/Test
The first task in prototyping and testing this design is to first test the diodes. The
diodes are designed to give out a wide range of wavelengths from 804
nanometers to 810 nanometers. The center frequency is determined by the
temperature of the die. The first test would be to determine what temperatures
21
are required to generate an output wavelength of 808 nanometers. The reason
that this is important is because the efficiency for Nd:YAG conversion to the
fundamental wavelength is best preformed with a pump wavelength of 808
nanometers. The test will also be used to determine what power outputs are
generated for the current that is used by the diode. This data is helpful when
building the laser power supply. Another measurement that was preformed is the
calculation of the diodes laser power output to its internal power meter. This was
used to determine if the diode is working correctly.
Once the laser diodes are fully tested, the next step is to build a simple laser
cavity that would consist of the laser diode, the highly reflective concave mirror
and a 99 percent output coupler. The next step would be to add in the imaging
lens. Once the lens is placed back in, the cavity length will have to be changed to
regenerate a stable cavity. Once the cavity is stable, the next element that would
be added would be the mirror that is reflective for the second harmonic and
transparent for the fundamental wavelength. The laser mirrors will then be
tweaked to generate the highest power output. The next object that would be
added would be the KTP crystal. The crystal rotation and position was adjusted
until the highest output power was generated through the output of the reflective
mirror for the second harmonic. The next step is to change out the 99 percent
output coupler with a highly reflective mirror for the fundamental and second
harmonic of Nd:YAG. The next element that would be placed in the cavity would
be the beam splitter. Since the KTP crystal is polarity dependent, both the cubic
beam splitter and the second harmonic generation crystal’s rotation will have to
be adjusted until the highest output could be achieved. The next element that
was added is the Q switch. The Q switch will need to be adjusted until the output
power was the maximum. The last element that would be added would be the
quarter wave plate. This was adjusted so that the cavity will not lase. The next
step would be to test the Q switch to see if generates a pulse laser pulse. The
power was measured and all components were adjusted to improve the output
power of the laser. The next step would be to calibrate the power output
compared to the input pump power. The next step would be to see if the power
output can reach the minimum required energy to generate the plasma that
vaporizes the copper substrate.
If the power output is below the required energy output then the second diode will
have to be added to the setup. This will consist of first removing the Q switch, the
quarter wave plate and the cubic beam splitter. The next step would be to add
another highly reflective mirror for the fundamental Nd:YAG and transparent to
the pump diode’s wavelength. Then the cavity was tweaked to produce the
maximum energy with just one of the diodes. Once this is done the next step was
to turn on both diodes at half power and adjust the cavity until maximum power is
achieved in the output. Now that the maximum is found the next step will consist
of placing back in the cubic beam splitter. After tweaking the beam splitter, the
next step was to add the Q switch. As before, the next element that was added is
22
the quarter wave plate. The final step was to adjust all the elements to generate
the maximum output power.
After the adjustments, the laser system needs to be calibrated to understand
what characteristics of diode power were needed to vaporize the printed circuit
board material. The final test was to measure the divergence of the output beam.
This data will help in choosing the correct focal point in the output lens to
collimate the beam. Characteristic such as the mode and the M2 will also need to
be measured to determine what the spot size was when the beam is being
focused on the copper substrate. This data will also be used to determine what
type of focusing lens was used. The next testing that will need to be done is to
redo the burn testing that was done with the quantaray laser.
6.3 – Build
Once the burn tests were finished, the next step was to look at the layout of the
prototype laser and redesign a new bread board that was the final placement of
the elements. When the design of the bread board was designed, the next step
was to prepare the stock aluminum by first flattening it and then drilling the holes
to mount the elements. The next step was to attach the elements on the new
bread board. Each element was placed on the bread board in the same order as
when they were prototyped. The next step was to adjust the elements until the
power outputs match the outputs that the prototype gave. When this process is
finished the next step was to cover the laser. This will prevent dust form entering
the cavity and degrading the power output. The exit portal for the laser is an
antireflective coated window. This will allow the entire laser to be sealed.
For diagnostic and safety purposes temperature sensors and humidity sensors
were added to the laser cavity. For the laser cavity mirror a photo diode will be
placed in its location. This is used to measure the output of the laser cavity
without using a high powered power meter. This can be used to detect faults and
warn the computer that a fault has occurred and to shut down the system. The
same could be said for the humidity sensors and the temperature sensors. The
diodes are also equipped with photo detectors and with the calibrated data from
the testing section this data could be used with the software to determine when a
diode is going bad.
6.4 – Evaluation plan
The first step in evaluating the laser section of this project is to first test a simple
cavity. The next step was to add additional elements to the laser cavity. This
process allows the ability to diagnose problems before adding additional
complexity. Once all the elements are added to the laser cavity. The next step
was to test it to see if the power output is within specification. If the power is
below specifications, then a second laser diode was added to the cavity. The
next step was to determine the divergence of the laser beam and the modes. The
23
beam will then be corrected to generate a near collimated divergence. The next
step was to place the laser on a bread board that would be dedicated to this laser
design. Lastly, the laser was sealed up and the laser beam was ready to be
routed to the XY table.
24
SECTION 7: Laser Power Supply Unit
7.1 – Previous Design
The previous power supply for the diode laser system requires a DC output with
low voltage and high current. The power supply was to have an input of 120 V
AC at 60 Hz from the AC mains and deliver an output of 0 to 5 V DC at 60 A to
the laser diode. The method used to power the laser diode had to be reliable,
precise, and have heat dissipation comparable to similar power supplies. The
laser diode needed a constant source of power that had a precise value, which
requires the power output of the conversion to have a small voltage and current
ripple. If the change in input power is too large, it will not allow the laser diode to
function and may damage the laser diode. The output voltage ripple must be less
than 1 mV. It was current controlled, and it had current and temperature
monitoring. The laser power supply was designed and simulated using LTspice.
It was used because it is commonly used in engineering courses and uses
equations that closely simulate actual circuit operation.
7.1.1 – Rectification
The AC mains was converted to a DC voltage before entering the converter. The
first part of the rectifying stage filters out high frequency AC noise using a lowpass filter. A metal oxide varistor (MOV) has a high resistance at low voltages
and a low resistance at high voltages, and it is used to protect the power supply
circuit from power surges. After this, a 60 Hz transformer provides electrical
isolation from the AC mains. The output of the transformer enters a rectifier
bridge with a filtering capacitor on the output to change the single phase AC to a
DC waveform. This DC waveform enters the DC to DC converter, which gives a
regulated DC output that can be changed in magnitude and corrected by a
control circuit if it differs from the desired output. There are different options to
consider when creating the DC to DC conversion circuit. The following schematic
is the rectifier used for converting AC to DC (Figure 7.1).
Figure 7.1: The filtering and rectification stage.
25
7.1.2 – DC to DC Converter
A linear regulator is one option to convert the DC voltage. It can deliver a lower
output voltage compared to the input, but it is not efficient enough. The regulating
component is an adjustable voltage divider network that maintains a constant
output voltage. The divided voltage that is not used as the desired voltage value
is dissipated as heat. This can make the efficiency drop to as low as 40% [9].
They are more inefficient as the voltage drop from input to output increases, and
they dissipate heat power equal to the product of the output current and the
voltage drop [9].
A better option is the switched-mode DC to DC converter or switched-mode
power supply. Instead of constantly dissipating unused power as heat, it uses
power semiconductor devices to allow the input power to provide power at
certain times, charging the power storing inductor and capacitor. The duty cycle,
D, is the ratio of on-time of the semiconductor to the switching period. Changing
the duty cycle allows for this converter to use the minimum input power
necessary to achieve the desired output power. Today, switching devices have
very high switching speeds and very high power handling capabilities, which
allows switched-mode power supplies to operate with an efficiency greater than
90% with low cost and relatively small size and weight [9].
Another reason the switched-mode converter was used is because its complexity
allows for the most control of the output voltage and current, compared to a linear
regulator. It can be adjusted by changing the duty cycle of the PWM on the
microcontroller. Changing the output power is more difficult on a linear regulator
because in many cases of linear regulators more difficult hardware adjustments
must be made.
The two switched-mode converter configurations that can be used in the design
to decrease the voltage and increase the current are the buck, which can only
have a voltage gain less than one, and the buck-boost, which can have a voltage
gain less than, equal to, or greater than one. A gain below one is the only
requirement for the laser power supply, so it was used instead of the buck-boost.
7.1.3 – Switching
There are several options for the semiconductor to be used for switching. The
desired characteristics of the semiconductor are high forward current carrying
capability and fast switching.
The BJT, IGBT, and the power MOSFET are the best options for achieving these
requirements. The BJT has a high power rating with high current carrying
capabilities in the on state, but if it becomes too hot, it is prone to malfunction
26
from second breakdown [9]. Also, it requires more power to drive than the
MOSFET, which gives the MOSFET an advantage. The IGBT is made of a BJT
and a MOSFET so it has a high current, but its turn-off time is larger than a BJT,
and can only operate up to 20 kHz in medium power applications, such as this
power supply [9]. The laser power supply cannot be operated at this low of a
frequency, because a much higher frequency must be used to obtain a voltage
ripple smaller than 1 mV. The power MOSFET has the highest switching speeds
compared to the other two, reaching more than 1 MHz, and a power rating of up
to around 600 V with a current rating of 40 A [9]. This power and current rating is
sufficient for the power supply, and the high switching speeds are good, so the
MOSFET was used.
In Figure 7.2 the diode that blocks when the MOSFET is on is also an important
part of the efficiency and successfully switching operation of the power supply.
The diode dissipates power as heat equal to its forward voltage drop multiplied
by the current through it. This heat requires the converter to be cooled to prevent
heat from damaging the circuit, and the diode current increases exponentially
with the voltage across it [9].
Figure 7.2: (a) A buck converter with a switch and diode (b) A buck converter
with two switches for synchronous switching.
A low power design uses a diode to block power from reaching the ground node
of the circuit. The high power being delivered to the output puts a high stress on
the diode. Diodes are not usually used for high power applications such as this.
This is a major disadvantage of this circuit, because the diode can be prone to
breakdown, and a power diode must be used to successfully block the high
power being delivered to the output.
The switched-mode power supply was made more efficient by replacing the
diode with a MOSFET, as seen in Figure 7.2.b. This is more efficient than using
any type of diode. This method is called synchronous switching [10]. To replace
the diode, M2 will turn off when M1 is on, and then M2 will turn on only for a short
time when M1 is off, which will drain ground the negative voltage on the inductor.
It is important that the two MOSFETs are not on at the same time, because some
of the output power from M1 will drain to ground, and the desired amount of
power will not reach the output of the converter.
27
The three MOSFETs put in parallel in place of M1 in Figure 7.2 provide the 60 A
of current needed for the output of the converter. Initial designs had current
output lower than 60 A, so adjustments had to be made to the circuit. The
requirements of a low voltage and high current output can be achieved by using
more than one switching MOSFET in parallel instead of a single switching
MOSFET. This will increase the output current while having little effect on the
output voltage. Because the majority of the heat dissipation in the converter is
proportional to the conduction time of the MOSFETs, having more MOSFETs
allows the heat to be more spread out on the heatsink, compared to using a
smaller number of switches.
7.1.4 – LC filter
The LC filter at the output of the buck converter serves to reduce the output
voltage and current ripple. If the inductor is too large, it will require too much time
to get to the desired steady state value. A larger inductor can decrease the
voltage ripple by keeping the converter in continuous conduction mode (CCM),
but because synchronous switching is used, the converter was operating in
discontinuous conduction mode (DCM). Therefore, inductor was small to save
space, have a lower resistance, and provide the desired inductor current in a
reasonable time.
Increasing the output filtering capacitor was a method used to lower the voltage
ripple. The capacitor was derived from
ΔVC
1− D
=
Vo
8LCf 2
where ∆VC is the voltage ripple, Vo is the output voltage, D is the duty cycle
created with PWM for the three MOSFETs in parallel, L is the inductor value, and
f is the switching frequency of the three MOSFETs in parallel, as can be seen
below in Figure 7.3.
28
Figure 7.3: Basic buck converter schematic using parallel MOSFETS.
The output voltage ripple must be smaller than 1 mV, which was achieved by
adjusting the output filtering capacitor of the converter or changing the switching
frequency of the switching MOSFETs. The ripple size decreases as the capacitor
increases. The ripple can also be decreased when the switching frequency
increases, but this also lowers the power delivered through the switching
MOSFET due to the switching limitations of the MOSFET. The buck converter
was simulated and the output values for voltage and current are given in the
following figure (Figure 7.4).
Figure 7.4: Buck converter simulation measuring output voltage and current.
29
In this converter, a relatively moderate switching frequency is used, and a
relatively large capacitor is used to achieve the desired output voltage ripple. The
ripple measures approximately 0.7 mV, as seen in Figure 7.5 below.
Figure 7.5: Buck converter simulation measuring output voltage ripple.
7.1.5 – Controller
The first design used a microcontroller to control the output. A difference
amplifier would measure the output current and provide a voltage that
corresponded to the current. This voltage would be the signal sent to the
microcontroller to adjust the duty cycle of switching MOSFETs. This design was
not used because it would be simpler if one less microcontroller were used, and if
a controlling chip is used, it could provide more features and the power supply
can more separate from the rest of the systems.
The design was then changed to use the LT1339 to control the buck converter. It
is a buck/boost controller for high power converters. Its capability of giving the
converter a high efficiency is a major reason for using it. It also has anti-shootthrough circuitry for converters dealing with high power. The LT1339 controls the
drivers of the two sets of power MOSFETs, which are BJTs that are powered by
a 12 V source. The high currents involved generate substantial heat, and the
high efficiency from the controller and the use of synchronous switching allows
the heat dissipation to be at levels comparable to power supplies that have
similar power conversions. The average current limit function is used to control
the output current. An op-amp circuit controlled by a digital potentiometer sets
the average current limit. The use of a digital potentiometer allows the laser to
operate at different energy levels for testing or troubleshooting and be easily
controlled using software. The output voltage is programmed with a resistor
feedback network. When the input voltage goes above 30 V, the four MOSFETs
in parallel exhibit phantom turn-on. This is solved by adding a negative 3 V offset
30
to the gate driver for these MOSFETs. If the input voltage unexpectedly drops, it
can cause the system to become locked in an undervoltage state. A shutdown
function is used to shut down the system when this occurs. The complete buck
converter with control circuit is depicted in Figure 7.5.
Figure 7.5: Buck converter with controller and current controlling circuit.
7.2 – Current Design
The previous laser design used laser diodes and required a converter to deliver
power with low voltage and high current. The current laser design uses flash
tubes and requires high voltage and a low current. This was achieved using a
boost converter. The rectification for the current converter design is the same as
the previous design. The flash tube system is not susceptible to moderate
voltage and current ripple in the output, so the output voltage and current ripple
do not need to be lowered much compared to the output for the laser diode
design.
7.2.1 – DC to DC Converter
The best option for the DC to DC conversion is the switched-mode DC to DC
converter to supply the laser flash tube system with 730 V. The boost converter
31
was used to output a voltage gain larger than one. A double boost converter
configuration was used to generate a faster current response to the output and to
help spread out the heat dissipated by using more switches. A current
transformer is used to lower the current through the main IGBTs. The boost
converter is depicted in Figure 7.6.
Figure 7.6: Buck converter with controller and current controlling circuit.
7.2.2 – Switching
The BJT, IGBT, and the power MOSFET are the possible switching devices
considered. The BJT is designed for high current, but the current used in this
converter will be handled fine with the other two switches, and BJT requires a
higher driving power than the other two devices. The power MOSFET has the
highest switching speeds compared to the other two, reaching more than 1 MHz,
and a power rating of up to around 600 V with a current rating of 40 A [9]. The
boost converter requires a higher voltage than 600 V and does not require
switching over approximately 100 kHz, so the MOSFET was not a good choice
for the switching device.
The IGBT is made of a BJT and a MOSFET so it can handle high current and
can be controlled easily because a voltage is used to control the gate instead of
current, but its turn-off time is larger than a BJT, and can only operate up to 20
kHz in medium power applications, such as this power supply [9]. The
32
disadvantage of the IGBT’s turn-off time is small compared to the disadvantages
of the other two devices, and the converter will be operating at 17 kHz, so the
maximum frequency limit of the IGBT is not a problem. Also, the IGBT is capable
of high voltage blocking, which is needed for the high voltages in the boost
converter. Compared to the MOSFET, the IGBT is generally more robust at
maximum operating loads and are more efficient at high temperatures [25]. The
IGBT is the best option for the boost converter.
7.2.3 – Controller
For the current design, the regulating pulse width modulator, UC3526, was used
to control the boost converter. Its operating temperature range is 0° C to 70° C,
allowing for proper function when the power supply is running at full power, which
generates substantial heat. The related controllers from Unitrode did not have a
high enough operating temperature range for this converter. Both of the IGBTs
for the boost converter are controlled using this controller with added op-amps
circuits. The soft-start feature protects transistors and diodes from high currents
when the converter is turned on. If the input voltage unexpectedly drops below a
certain voltage, it can cause the system to become locked in an undervoltage
state. If this occurs, the output is disabled until the system comes out of the
undervoltage state and can operate normally. Current limiting is controlled
digitally, allowing the user to control the boost converter output via software.
7.3 – Prototype/Test
The parts were acquired via online distributors. All of the testing of the laser
power supply was done in the UCF senior design lab. The control circuit and
other circuits that do not have high voltages or high currents were constructed on
a breadboard. They were constructed using the parts and wire that was acquired.
The parts of the power supply using high voltages or high currents were soldered
onto a printed circuit board (PCB), because a breadboard would have been
damaged by the high currents, or high voltages would have caused a short
between the small gaps in a breadboard’s circuits. This circuit was constructed
on a PCB and tested with the circuits on the breadboards before all circuits were
created on PCBs. The high voltage circuits must not have sharp points on the
solder, because these can ionized the air and cause the power supply to fail.
Oscilloscopes were used to probe the circuit and check to make sure there were
no short circuits and the voltages and currents were at the designed values at all
nodes in the circuit. The laser power supply was then connected to a GFCI outlet
and the node values were checked. Then it was connected to a standard wall
outlet and the node values were checked. A dummy load was the first load to be
tested. This was a resistor rated for 50 W. The nodes were checked, especially
the output current, which must not vary much from 730 V. Next, the power supply
was connected to the flash tube laser system and the nodes were checked. It will
33
also be turned on for continuous operation for a length of time to see if the heat is
dissipated without damaging the power supply and the output stays constant.
7.4 – Build
All components were bought from online distributors, making sure they arrived on
time and then the soldering and other building were done in the senior design
lab. Some inductors were bought professionally wound, and other larger
inductors were wound by hand in the senior design lab. The laser power supply
is housed in a container to eliminate anyone from damaging it, or anyone being
injured by the high power in the converter.
7.5 – Evaluation Plan
The laser power supply must meet the requirements of a voltage output of 730 V.
Once all the circuits are combined on PCBs and tested, the final version is tested
with all other hardware to see if the desired results had been achieved. If not,
they would be redesigned either on breadboards or PCBs and tested again. The
power supply is air cooled and it was made sure that it did not overheat and
require more cooling. It is important that the design for the high power circuits
were accurate, because they were not prototyped on a breadboard.
34
SECTION 8: TEC Power Supply Unit
8.1 – Design
Thermoelectric cooling was designed for the previous laser diode system;
however, the current laser flash tube system does not require it. Air cooling is
sufficient for the flash tube laser operating at a few pulses per second. If the
pulse rate is increased, a water cooling system would be used to cool the flash
tube system. In the future, a laser diode that outputs the correct wavelength
could be acquired and the TEC system could be used. This will probably happen,
because the laser diode can etch boards much faster than the flash tube system,
and the diode would probably be worth the investment.
This section explains the power supply used to power the thermoelectric cooler
(TEC) for the laser diode system. The power supply must be reliable and precise
to keep the laser diodes operating in a safe range of temperatures. The TEC
power supply must have an input of 120 V AC at 60 Hz from the AC mains and
deliver an output of 60 V DC at 3 A. The temperature is sensed to prevent
damage from overheating.
8.1.1 – Rectification and DC to DC Conversion
The AC mains is converted to a DC voltage before it can be used to power the
TEC. This rectification was similar to that of the laser power supply, except the
transformer was utilized to step-down the voltage.
The first part of the rectifying stage filters out high frequency AC noise using a
low-pass filter. A fuse was used on the incoming line to prevent any current
surges from reaching the power supply. A metal oxide varistor (MOV) has a high
resistance at low voltages and a low resistance at high voltages, and it is used to
protect the power supply circuit from power surges. If there is a surge, it will send
it to ground instead of the power supply. After this, a 60 Hz transformer provides
electrical isolation from the AC mains and it is a step-down transformer. The
winding ratio is two to one to transform 120 V AC into 60 V AC. The output of the
transformer enters a rectifier bridge with a filtering capacitor on the output to
change the single phase AC to a DC waveform.
8.1.2 – Cooling Control
The DC waveform then enters a buck converter is used to control the power sent
to the peltier (Figure 8.1), the thermoelectric cooling device. The amount of
cooling is proportional the power sent to the peltier [11]. A thermistor circuit is
used to sense the temperature of the laser diode and create a signal to be sent
to the microcontroller. The microcontroller adjusts the duty cycle from the PWM
that controls the switching MOSFET of the buck converter, giving it a larger duty
35
cycle if more cooling is needed. The ripple is not an important factor for the
converter, because the voltage has already been stepped down with the
transformer to 60 V. Since four peltiers are being used, they require nearly 60 V
and it will not damage them if 60 V is applied. As long as the ripple is not more
than about 10% to 15%, and the frequency of the switching MOSFET is above
about 1 kHz to 2 kHz, the peltiers should operate well [11]. A high current is not
required, so only one switching MOSFET is required to provide the 3 A output.
Figure 8.1: A peltier that has been acquired.
8.2 – Prototype/Test
The prototyping and testing was similar to the laser power supply testing,
including the parts acquisition, testing in the senior design lab, and putting only
low power circuits on a breadboard. Oscilloscopes were used to probe the circuit
and check to make sure there are no short circuits and the voltages and currents
are correct at all nodes in the circuit. The precision of the control circuit and the
PWM was verified for proper cooling of the diodes. The laser power supply was
then connected to a GFCI outlet and all node values were checked. Then it was
connected to a standard wall outlet and all node values were checked. An
ohmmeter cannot be used to measure the peltier, so a resistance test was done
with an LCR meter, which measures the resistance using an AC voltage [11].
A dummy load was the first load to be tested. This needed to be a heat source,
and the heat source we used were the hot sides of spare peltiers. The peltiers
were run at a certain temperature, and the temperatures of the heating and
cooling peltiers were measured. The data sheets for the peltiers were checked to
see if they are operating at the expected efficiency. Next, the TEC was to be
tested on the high power laser as the final test.
8.3 – Build
Similar to the laser power supply build, the TEC power supply would have had all
components bought from online distributors and then the soldering and other
building was to be done in the senior design lab. The inductors would have been
bought already wound, instead of being wound manually, so that they are
professionally wound. It would be housed in a container with the laser power
36
supply to reduce the chance of anyone damaging it, or anyone being injured by
the high power in the converter.
8.4 – Evaluation Plan
The TEC power supply must meet the requirements of a conversion of 120 V AC
at 60 Hz to 60 V DC at 3 A. The voltage ripple must be less than about 10% to
15% of the output. Sufficient testing sessions at durations comparable to that of a
PCB being etched that inspect every voltage, current and temperature would
determine if corrections needed to be made to the power supply. After iterations
of corrections are completed, the final build would occur and be placed in the
final housing for the etcher.
37
SECTION 9: Q Switch
9.1 – Types of Q Switches
A Q switch is an optical device that changes the quality (Q) factor of the laser
cavity. One of the main reasons to implement this technique is that it allows the
laser to generate a high intensity pulse instead of a continuous wave. To
generate the high energy laser pulse there are two main principles. First when
the Q switch is off the laser cavity has a low Q. In this state all the laser cavity is
being dumped out of the cavity and preventing the cavity from oscillating. This
also allows the gain medium to become saturated. The second state is when the
Q switch is turned on. At this state the laser cavity generates a high Q factor and
oscillation begins. Since most of the gain medium was becoming saturated in the
first state, when the cavity starts oscillating the gain medium generates a
population inversion and most of the medium dumps all of its energy into one
pulse.
The first use of this technique was used on a ruby laser [12]. When the first tests
were ran the ruby crystals fractured because of the immense pulse energy. The
reason for this was that the lasers were pumping out many megawatts of laser
light.
There are four main devices used for Q switching. The easiest to use are
mechanical shutters or choppers. These work by physically placing an object into
the cavity to reduce the Q factor. The main drawback to this design is that the
standard pulse duration is less that 200ns and the mechanical device would have
to move fast enough to open and close the shutter in this time period, or the
output energy would be reduced. The following is a common chopper that would
normally be used for CO2 laser systems (Figure 9.1).
Figure 9.1: Mechanical chopper for Q switching
38
Another technique is to use a Saturable Absorber. This technique employs a
crystal such as Cr:YAG or bleachable die. The material reduces the Q by
absorbing all the light until it becomes saturated, once that happens it becomes
transparent and the cavity begins to oscillate. The main drawback to this design
is that the crystal generates a lot of heat when it is absorbing the light; this
prevents the laser from operating at high repetition rates. The crystals used in
this technique over time will start to darken and reduce the power output.
The third design is an acousto-optic modulator (AOM). The AOM works by
generating a pressure wave in a crystal medium. This pressure wave bends the
lights path. When the crystal is off, the laser light passes straight through the
crystal. In high powered laser designs the cavity is setup to run on the first order
path that is bent when the crystal is on. The reason for this is that with high
power the little light that doesn’t become bent is enough to start lasing. The
following is the standard output beams of the acousto-optic modulator (Figure
9.2).
Figure 9.2: AOM diagram (Permission granted Wikipedia.com)
The last design is to use a pockel cell. There are two common types of crystals
used for this cell, they are BBO and KD*T. These cells work by rotating the
polarity of the optical beam and using a polarized beam splitter to remove one of
the polarities. When the crystal is off the beam is not rotated and the beam exits
through the beam splitter. When high voltage is applied to the cell the crystal
rotates the beam’s polarity by a quarter turn and when it bounces through it again
39
it rotates another quarter turn. The Beam then passes through the beam splitter,
and oscillations begin. The following diagram shows how the polarized wave
reacts after leaving the pockel cell (Figure 9.3).
Figure 9.3: Pockel cell diagram (Permission pending)
For this project it was chosen that we use a pockel cell. The reason is that it will
allow easer alignment than an AOM and will allow higher power pulses than the
other two régimes. The crystal type that we chose was a KD*P crystal. The
reason for choosing this type of crystal is that it would fit in the budget. The
drawback in this crystal compared to BBO is that the crystal cannot be modulated
at higher frequencies. Since our goal is to modulate around one kilohertz this
crystal will work well.
9.2 – Previous Design
In order to drive this Q switch, it will need to have a differential of around three to
five kilovolts. The switching speed of this will need to be around 10 ns. To
accomplish this there are three common methods, they are planar triode,
avalanche transistors, and fast FETs.
The main reason that a planar triode was not chosen was that it is hard to find in
common electronic stores and the government highly regulates the selling of
them. The reason that the Q switch driver was not designed using avalanche
transistors is that basing each transistor would generate many issues and create
a larger part list. Therefore the best solution would be to use fast FETs.
The driver works by generating a high voltage power supply and driving the
voltage to ground using the fast FETs. The power supply was broken into nine
sections that would allow easer for easer testing. The following block diagram will
help explain the breakdown of the high voltage power supply (Figure 9.4).
40
+12V
Microcontroller
Voltage
Regulator
40kHz
Oscillator
1:98
Transformer
Darlington
Transistor
Voltage
Multiplier
Filter
Caps
Fast
FETs
Voltage
Divider
Figure 9.4: High Voltage Power Supply Block Diagram
9.2.1 – Q Switch High Voltage Power Supply
The design of the oscillator for the high voltage power supply is a square wave
generator using two operation amplifiers. The reason that a square wave was
chosen was because the darlington transistors used generate a non linear gain.
The non-linear gain would modify a sine wave, if it was used and the
amplification would generate more of a square wave. It seemed that using less
parts and loosing the efficiency of not using a sign wave would be the design of
this part.
The transformer used for this high voltage power supply is a CCFL transformer.
The reason for choosing this transformer is that it is commonly available from
electronics stores and the cost associated with it is minimal compared to having
a transformer custom made. The transformer has a primary tap that is split this
will allow bidirectional pulling. This is why two operation amplifiers are used. The
first one generates the square wave and the output drives the one of the
darlington transistors, the output also drives the second operational amplifier,
who is set up and an inverter. This output will then be connected to the second
darlington amplifier. The scheme will allow the transformer to have a push pull
operation and generate a more efficient system compared to just modulating one
side of the transformer. The transformer also has a feedback coil that would
commonly be used to generate the oscillations, but for our project it was
determined that allowing the device to freely oscillate may generate instability
when the circuit is connected to unknown devices, therefore the outputs from the
feedback coil are left floating, and the main coil is driven at a set frequency. The
frequency of the oscillator is controlled by the RC constant of the circuit, using
the following equation the values of R and C are rounded to the nearest common
value. The R was set to one kiloohm and the C was set to 4.7 nanofarad. This
combination generates a frequency of around 50 kilohertz. The following is the
schematic of the oscillator (Figure 9.5).
41
Figure 9.5: Schematic of oscillator circuit
The specification sheet for the transformer gives a range of usable frequency to
run it, they range from 40 to 80 kHz, and therefore the given RC values are within
specifications of the datasheet. The following is the simulated output of the
oscillator circuit (Figure 9.6).
Figure 9.6: Simulated outputs of oscillator
The voltage regulator for this circuit consists of a transistor that limits the current
that enters the transformer. The transistor is controlled by an operational
amplifier that compares the voltage that the voltage divider circuit gives and the
voltage that the microcontroller gives and if the voltage is lower than the
microcontroller, the operational amplifier give a high state and the transistor
permits more current to enter the transformer. When the voltage is above the
microcontroller’s, the opamp gives a negative state and reduces the current that
the transistor is sending to the transformer. This design does generate a voltage
swing between the two states, but this voltage change would be low enough to
allow the Q Switch to operate correctly. This current limiting design was
presented in “The variable High-Voltage Power Supply 0-3kV [13].”
The Voltage multiplier section of the High Voltage power supply consists of a
Cockcroft–Walton multiplier. The reason that this section is needed is that the
maximum voltage that the CCFL transformer can deliver is 1.34 kilovolts. To get
the five kilovolts needed to run the Q switch the output of the transformer goes
through a two stage voltage multiplier. This will generate a four times
42
multiplication. Since the cost of high voltage capacitors that are needed to run
the voltage multiplier are exponentially expensive verses surge voltage, it is more
cost effective to run the capacitors in series. Running the capacitors in series will
allow higher voltage handling while maintain a low cost for materials. To prevent
one of the capacitors form generating a higher voltage then the others, in its
series, each capacitor will have a ten megaohm resistor. This will allow the
capacitors to discharge when not in use and prevent one of the capacitors from
generating a voltage above its specification. For the diodes used in the multiplier
it is also cost effective to use multiple diodes in series than to use just one. The
diodes used are the HER108. The diode is rated to have a 1000 volt reverse
voltage and is designed for fast switching with a recovery time of 75
nanoseconds. The reason that this diode was chosen was that it was a cheap
diode that is readily stocked and allows a faster recovery rate compared to the
standard 1N4007. The following is the schematic of the voltage multiplier (Figure
9.7).
Figure 9.7: Schematic of Voltage multiplier
The left terminals are connected to the CCFL transformer and the right side
terminals are connected to the filter caps. The filter caps were added to the
circuit to regulate the voltage and to create a storage tank to prevent voltage
drops when the FETs are switch on. Like the voltage multiplier, it is more cost
efficient to use multiple capacitors in series than to use one capacitor rated for
the five kilovolts. The capacitance chosen was not too critical because the
voltage multiplier smoothes out the voltage as it passes through the circuit. The
capacitors chosen was a 0.15 microfarad capacitor rated for 3000 volts and eight
milliohms for ESR. This is the same capacitor used for the voltage multiplier. The
following is the schematic of the filter cap design (Figure 9.8).
43
Figure 9.8: Schematic of the Filter Cap layout
To allow the circuit to regulate the voltage for this power supply, a voltage divider
will be used the divide ratio will be 999:1. This ratio will allow the voltage to be
read at safe levels of three to five volts. The high voltage resistors used are the
8.2 megaohm metal film resistor, and one 100 kiloohm resistor and a 6.8 kiloohm
resistor. The current rating for this circuit is 0.25 watts and the rated values for
the resistors are 0.5 Watts. To protect the comparers that the voltage regulator
uses there will be a Zener diode on the output that is rated for a breakdown
voltage of 5.1 volts and a power rating of 1 Watt. This circuit will prevent voltage
spikes form destroying the comparers.
The voltage regulator circuit is a circuit that will take the voltage from the voltage
divider circuit and compare it to the voltage from the digital to analog chip that the
microcontroller set. The voltage comparer circuit checks to see if the voltage form
the divider is lower than the digital to analog converter. If it is lower the output is
railed to high. This output is sent to an N type transistor. The current is regulated
by a 330 ohm resistor. The transistor used is the 2N3055, this transistor was
chosen because it has a voltage rating of 60 volts, which is well above the 12
volts that the circuit uses, and it was used in the reference high voltage power
supply [14]. The comparer used for this portion of the high voltage power supply
is the standard LM358 comparer microchip. This was chosen because of its
reliability and it is commonly available. The following schematic illustrate the
voltage divider and the voltage regulator (Figure 9.9).
44
Figure 9.9: Schematic of the Voltage Divider and Voltage Regulator
The output from the voltage regulator is sent to the CCFL transformer. The 12
volts that this power supply uses is from an ATX power supply. The reason that
this type of power supply was chosen was that it is cheap and more stable than a
wall version. Using an ATX power supply will also allow this circuit to be smaller
because there will need not be a reason to have a voltage regulator regulate the
12 volts for the circuit. This will also reduce the space that this power supply
occupies.
For the microcontroller section of the high voltage power supply, The Digital to
Analog chip that was used is Microchip’s MCP4251. The reason that this chip
was used was that it has a low cost and had 256 divisions [15]. This will allow the
computer to have step control of 20 Volts per step. There will also be override
pins that will be connected to a multi turn potentiometer. The reason for the
override scheme is that it will allow easer testing of the circuit, and give an
alternative to have the voltage permanently set if the microcontroller fails. The
microchip is controlled by SPI serial. The microcontroller will read the preset
values from its EPROM and send it to the Digital to Analog potentiometer. If the
computer requires changing the voltage, the microcontroller will get the packet
from the computer and resend the data to the Digital to analog potentiometer.
The microcontroller will also sample the voltage form the voltage divider and
when the voltage does not reach the correct voltage after a preset number of
cycles the microcontroller will send to the computer a fault and shutdown the
power supply.
The Microcontroller will have seven main functions that will be personalized for
the Q switch power supply and driver. In the first section it will read the EPROM,
this is the data that the microcontroller will send to the high voltage power supply
when the chip turns on. The second section will be computer change data, this
will read the USB data packet and will determine to change the value of the
Digital to Analog potentiometer. The third section is the Change EPROM, this will
45
read the USB data packet and change the EPROM data that relates to the
Voltage that the potentiometer sets. This section will also call the fourth section
to update the potentiometer. The fourth section of code will read from the
EPROM and send the data to the potentiometer to update the voltage. The fifth
section will read the onboard Analog to Digital circuits and determine if the
voltage is within the given range that the potentiometer set. The sixth section will
determine if the voltage is below the specification for over one second the
microcontroller will shutdown the power supply. Shutting down the power supply
is cutting the 12 volts power supply. The cutting of the 12 volt power supply is
done by shorting a relay coil that cuts the voltage. This section will also send a
USB data packet to the computer, with the data that a fault has occurred and at
what voltage the chip read. The last function that the controller uses is the pulse
function. This section will toggle a pin from low to high and then back to low. This
will trigger the fast FETs to switch on. This function will be called when a USB
packet is sent by the software with the command to call this function.
When the device shuts down, to create isolation from the circuit and the ATX
power supply, a relay will be switched on. To switch the relay on and to keep it
on until the device is reset, a SCR is used. The SCR used for this section is the
2N5061. This SCR was used because of its pricing and that it allows the on state
hold current is within the specifications of the relay that will be used. The relay
that will be used for this section is the G6RL-1-DC12. This relay was chosen
because it has a coil voltage of 12 volts and a current usage of 18.3 milliamps
[16]. The reason for the low current design is to be a little more efficient in power
management. This size of this circuit is a little bigger than if solid state
components are used, but it is a little more reliable than the solid state versions
when it comes to electrostatic discharge. The reason is that many times when
the solid state relays fail, they are normally shorting the switch on. Since the
design of this section is geared for safety, that scenario would be defeating the
purpose of this circuit. So a relay was chosen because it is more resistant to
electrostatic discharge. The only part of this circuit that is sensitive to ESD is the
SCR and if the SCR would be destroyed, it will turn on the relay coil and cause
the circuit to shutdown. The setup of the relay is that the switch is normally
closed. This will allow the relay to only be used in a shutdown situation. This will
increase the lifetime of the relay because it is not switching every time the power
supply is turned on. The following schematic will show how this section is
connected to the high voltage power supply (Figure 9.10).
46
Figure 9.10: Schematic of the Shutdown Circuit
9.2.2 – Q switch Fast FETs
The main design of the fast FETs section is that there will be two banks that will
be charged by the high voltage power supply. When the laser needs to pulse,
one of the banks will be discharged through the fast FETs. This will create a
differential between the two banks and will switch the Q switch on. When the
pulse is over the circuit will fire both banks, so that they will be shorted to ground.
This will then generate a differential of zero between the two banks and turn the
Q switch off. The RC component of the bank will then start to refill the banks and
the voltage on both of the banks will be back to the high voltage power supply’s
voltage. There are six components to the fast FETs circuit. The flow diagram
below shows how they are linked together (Figure 9.11).
Figure 9.11: Schematic design of the fast FET section
The microcontroller will send a pulse to the pulse generator. This section was
covered in the above literature for the High Voltage power supply. The
47
microcontroller will just toggle the states from low to high. The pulse length of this
toggle is about two clock cycles. Since the clock of the microcontroller will be
running about 48 MHz, the pulse duration will therefore be 41.6 nanoseconds.
The next section that the microcontroller passes the toggled pulse to is the pulse
generator. This section will shorten the pulse and generate the second pulse to
turn off the Q switch. The pulse generator will consist of a LC delay network that
will be connected to the input of the circuit. The output from the delay circuit will
then be compared with an exclusive OR gate. This will generate an output of two
pulses. To remove the falling edge pulse the output will be compared with the
original input signal, using an AND gate. The output will then be the pulse that
will be sent to the first bank to turn on the Q switch. This pulse is also connected
to another passive LC delay network and generates a delay of around 20
nanoseconds. This section will be allowed to be changed because when
tweaking the pulse energy of the laser this delay time will greatly change the
efficiency of the laser pulse. The output of this section is connected to the pulse
driver of the second bank and when fired it will turn off the Q switch.
The pulse driver is used to drive the main inductor coil. The pulse driver will be a
120 volt line driver that will amplify the five volt pulse that the pulse generator
created and amplify the voltage to 120 volts. This is done by first generating 120
volts DC. This is created by taking the 120 volts from the AC power line and
passing it through a bridge rectifier. The output of the bridge rectifier is then
filtered with a 10000 microfarad capacitor. The bridge rectifier is just four 1N4001
diodes. The reason that they were used is because the current needed for this
section is minimal so buying a premade bridge rectifier would increase the cost of
the design. The output of the filtered DC source is then passed through a one
kiloohm resistor. The resistor is to limit the current that will be generated when
driving the pulse driver. The driving mechanism is four MOSFETs that are
connected in parallel. This will generate a faster turn on for the pulse than just
having a single MOSFET. The MOSFET used for this section is the
STF5NK100Z. This MOSFET was used because it will also be used for the FETs
section of this project.
The inductive coil that is used to bias the MOSFETs in the FETs sections, consist
of a Teflon wire that is also covered with an additional Teflon sleeve. This will
generate the isolation that is needed for the circuit. The wire is then placed in the
center of six ferrite beads. Each bead will have a loop of Teflon wire that will be
connected to the base of the FETs and the gate of the FETs. When the pulse
enters the inductive coil, a voltage will be generated on the gate and will turn on
the FETs.
The FETs section will consist of six series of MOSFETs. Within each series there
will be two MOSFETs connected in parallel. The reason for the series design is
that high voltage MOSFETs are not commonly stocked through electrical
distributers and the cost of a high voltage MOSFET is significantly higher than
48
the cost of multiple transistors in series. To prevent some of the transistor from
generating voltages high than their specifications, each series connection will
have a ten megaohm resistor connected to the source and drain. This was the
same idea that was used for the high voltage filter capacitors. At the end of the
FETs section the output of the last series is connected to the ground plane of the
board. The beginning section is then connected to the RC bank. This section and
the pulse driver sections are duplicated for both the ON Q switch bank and the
OFF Q switch bank.
The last section is the RC bank that is used to connect the high voltage power
supply, the FETs and the pockel cell together. The resistor that is used is a high
voltage resistor that is rated at 750 kiloohms. The voltage rating of the resistor is
5 kilovolts. The capacitor that is used is a high voltage ceramic capacitor. The
voltage rating of the capacitor is six kilovolts and the capacitance rating is 47
picofarads. The connection to the pockel cell will consist of a high voltage screw
terminals. The pockel cell will be connected to the terminals using high voltage
Teflon wire. The voltage rating for the wire must be greater than five kilovolts.
The following schematic is the pulse driver, the inductive coil, the FETs, and the
RC bank (Figure 9.12).
Figure 9.12: Schematic of the Pulse generating circuit
9.3 – Current Design
Since the old design’s laser cavity was large enough to add in a pockel cell, this
was what was going to be used for Q switching. Now that the laser cavity is less
than 10 cm long, a pockel cell could not be added to the cavity. In order to still
have a Q switch design, the alternative is to add a saturable absorber. This
device only takes up 3 mm to the cavity.
9.4 – Prototype/Test
To prototype the electronics for the Q switch, the electronics will be broken up
into two main sections. The first section consists of the high voltage power supply
49
and the second section will be the fast FETs. The reason for the splitting of the
electronics in the two sections allows ease for testing them. Once the power
supply and the fast FETs electronics are verified to be working the boards will be
joined together with a high voltage Teflon wire.
To prototype the high voltage power supply as a whole, a circuit board will be
generated on the computer and the board will be etched out. The reason for not
building the supply on a bread board is because it will be generating high voltage
that would be able to jump the traces in the bread board and short its self out.
Some of the sections that were explained in the design section can be made on
the bread board. The sections that could be made on bread boards are the shut
down circuit, the 40 kilohertz oscillator, the voltage regulator and the
microcontroller. Once those parts are verified to be working the board will be
designed and when the board is made, the parts will be soldered in place and
verified that the connections are in place. Sections that need special care are the
high voltage sections. The high voltage sections must be checked to see if there
are any sharp solder points. These points will start to ionize the air and could
generate a failure of the high voltage power supply. When the board passes its
visual inspection, the board will then be connected to a GFCI outlet and tested
with high voltage probes. The board will also then be connected to the
microcontroller. The high voltage power supply will then be regulated by the
microcontrollers set values. The results will also be verified by testing it with the
high voltage probes. The microcontroller will also simulate a power supply will
then be safely biased incorrectly. This will test to see if the microcontroller will
shut down the high voltage power supply. The next step will see if the computer
will be able to control the voltage of the high voltage power supply. The computer
will send the USB data packet to change the voltage to two kilovolts. Then if the
power supply works correctly it will be changed to other voltages and verified as
well. The next step would be to see if the other functions that were programmed
will also work.
Since the complexity of the fast FETs electronics, by using mostly surface mount
parts, it is unreasonable to prototype this section on a bread board. The first step
in prototyping this section, is to break it up into the six sections that were broken
up in the design section. This will allow the ability to test the circuits to see what
characteristics they have and how to modify the delay networks in order to
improve on the switching speeds. Once the sectional boards are verified to work
correctly they will be designed to be on one board. To prototype the fast FETs
electronics, they will also be drawn up on the computer and a circuit board will be
made. To test the circuit, a lower voltage of around one hundred volts will be
applied to the high voltage power supply section and when the pulse is
generated, the high voltage probes will check to see the fall rate of the circuit. It
will also tell if the circuit is working correctly without destroying the majority of the
components. Then the board will be tested at the five kilovolts to see if it will
switch and at the rate specified for the pockel cell. Then if all goes well, the high
voltage power supply board and the fast FETs board will be connected and
50
retested to see if they will work together and that the system will be within
specifications to run the pockel cell.
9.5 – Build
Once prototyping and testing is finished the boards will be remade on one board
and mounted in an enclosure. Like before in the prototyping section care must be
taken in checking to see for sharp point on the high voltage side. Because of the
high voltage characteristics of this power supply, the board will be elevated a
quarter of an inch above the enclosure using fiberglass spacers. This will
generate an isolation of five kilovolts. The enclosure will also be labeled “Caution
High Voltage.” The inputs to the box will be a USB port and AC power jack and a
12 volt terminal. The outputs will be two high voltage Teflon coated wires with
enough length to connect to the pockel cell.
9.6 – Evaluation Plan
The evaluation plan for the Q switch electronics is to first layout the sections of
circuits that could be tested using bread boards. Then after fully testing the
circuits that were made on the bread boards the parts that could not be tested
this way would be designed and prototyped on printed circuit boards. Once all
the sections are tested and working correctly the main design two sections will be
designed and created in printed circuit board material. The Boards will be built
and tested. If the boards work correctly they will be joined together and retested,
otherwise they will be redesigned and redo the testing. After the boards pass the
testing a new printed circuit board will be constructed, this one would include the
electronics for the high voltage power supply and the fast FETs. Once built, it will
be tested. The board will then be packaged in an enclosure and will be ready for
use with the pockel cell.
51
SECTION 10: Motors and Motor Controller Unit
10.1 – Design
To move the mirrors into the location for milling away the copper substrate, a
motor was used. The delivery mechanism for moving the mirrors is through a
belt. There are two main types of motors to choose from. The first is a DC motor.
The biggest problem with a DC motor is that the reliability that the motor will stop
in the same locations time and time again is very minimal. This could be
improved by attaching a rotary encoder. This device would allow the
microcontroller to know the location of the big steps. The down side of this is that
if the motor needed to make smaller steps than what the encoder measures, then
it is more of a guessing game. The reason is that if during the step the motor
experiences inconsistent load changes, then the position location would not be
the same location every time. The second type of motor that could be used is a
stepper motor. The stepper motor allows high reliability and accuracy compared
to a standard DC motor. The stepper motor also has the advantage of allowing
microsteps. These steps would be divisions of the major steps and would still be
more reliable compared to the standard DC motor. The biggest problem with
stepper motors is that the controlling electronics to move the stepper motor is
more complex than a standard DC motor. For an example, a standard DC motor
would only require an H-bridge to control the movement of the motor. For a
stepper motor, the electronics can vary from using two H-Bridges. This scheme
would only allow the motor to move in the major steps. The second type of
electronics that could be used for a stepper motor, that would allow the
microstepping scheme, is a full bridge configuration.
There are three different kinds of wire configurations that can be used to run the
stepper motor. The more common variety is the unipolar configuration. This
configuration consists of two main coils per phase. Each coil also contains a
center tap in the coils. This configuration my also come with two of the ends of
the coils connected. This would reduce the amount of wire used from six wires to
five. The five wire scheme also allows easer design for the electronics. If a six
wire stepper motor was reconfigured, it can be used to generate the second most
common type of stepper motor. The second type is called a bipolar stepper
motor. The bipolar stepper motor is the same as the unipolar stepper motor in
that it contains two coil windings per phase. The difference between the two is
that there is no center tap for the bipolar scheme. This allows the wire count to
be down to only four wires. As said before, this design can be created using a
unipolar motor scheme by just ignoring the center tap wires. The last variety of
stepper motor is the eight lead stepper motor. This is a hybrid version of the two
varieties given above. The reason for the hybrid design is that the stepper motor
can be designed for high torque operations and then switch to high speed
operations with low current consumption. An example of this form of operation is
for a lathe CNC machine. The reason is that when the bits are not cutting the
52
material they can be moved quickly to the next position and when the bit would
need to be cutting the material the circuits would switch to a high torque portion
of the motor. This design would not be needed for this project because there will
not be changes in the torque while the machine is in operation. This leads the
choice of using a bipolar motor or a unipolar stepper motor. After looking at the
current availability of the stepper motors available through our current
distributers, the stepper motor that was chosen was a unipolar stepper motor.
The motor is from Jamco. The reason that this motor was chosen was that it had
a step angle of 0.9 degrees. This is half the stepping angle of standard stepper
motors. The holding torque for this stepper motor is 30 oz per inch. Since the
gears that will be driving the belts for this project will be around a quarter of an
inch, the holding weight that the motor will hold will be seven and a half pounds.
This is far higher than what would be needed to run the XY portion of the XY
table.
The motor is run as a bipolar motor because the electronics to run a bipolar
motor are commonly built in a single package design. The reason that we chose
to use the electronics that are in a single package design, compared to just using
discrete components, is that the part count is far lower, and the need to
customize the electronics to match the stepper motors by using discrete
components is not needed. The reason is that this project is not running on a
battery or with other unique design criteria that would prevent the use of an all-inone stepper motor driver. There are many different types of stepper motor driver
chip that are commonly used by many hobbyists. The chips range from using
multiple H-bridges to dual full bridge modules. The most common one is the
L298. This is a full bridge MOSFET driver. This chip was designed for running
stepper motors with a driving current of one amp. The chip also has a continuous
drain current of four amps. This was chosen because there are many coded
examples for driving this chip for microstepping. Even Microchip has an article for
using this chip with the PIC18C452 to drive a bipolar stepper motor [17]. The
model that is used for this project is the vertical mounted L298N. This chip was
made from STMicroelectronics. The reason that the vertical mounting scheme
was chosen was that the chip will be generating a great deal of heat when the
chip is running microsteps. There must therefore need to be a heat sink
connected to the chip to prevent them from overheating. Since the
microcontroller that is used for this project is the PIC18F2550 series, the
microcontroller only has two PWMs. To run the circuit for microsteps the chip
must have four pulse width modulators. One way that was noted from Microchip
was to use logic chip, such as two inverters and four AND gates. This scheme is
used to modulate the PWM to make four PWMs from two. The following
schematic is what is used for the microcontroller and the full bridge driver for the
stepper motor (Figure 10.1).
53
Figure 10.1: Schematic of microcontroller and full bridge driver
To prevent the feedback that is generated from switch the stepper motor, a
snubber circuit was added to the terminals of the stepper motor. The snubber
circuit will be primarily two diodes that will be connected to the 12 volt power
supply and ground. The diode that STMicrochips recommends for this type of
application is the 1N4001. This diode is the common diode that is used in other
designs. For this project the diode that was chosen was the 1N4001. The main
reason for choosing this diode was because of the recommendations and
because it is the most reliable design since the popular use of it. This would help
54
in debugging this section of the circuit because of the review that others have
generated.
The power supply that is used for this portion of the project is one of the 12 volt
leads from an ATX power supply. The ATX power supply is able to drive the
circuit at a much greater current that what the circuit uses. It is therefore a
reasonable choice to use the ATX power supply than to build a voltage regulated
power supply that would need to drive more than four amps per stepper motor.
This design that was created would only be able to drive one of the stepper
motors. Since this design will be the same for both the X and Y axis the circuits
will be duplicated for each.
When the circuit first starts up, the stepper motors must move to the home
position. The reason for this is that when the machine is turned off the motors
can be moved to any position without having the microcontroller knowing where it
is. When the machine is turned on the microcontroller will move the motor
backward until it hit the home position. Once the motor reaches the home
position the stepper motor will stop and will be ready to receive the commands
from the computer. The stepper motor knows when it reaches the home position
because there will be a limit switch located at the end of the rails. When the axis
hit the limit switch the switch will send a low to the microcontroller and the
microcontroller will stop the stepper motor. The circuit for the limit switches must
contain low noise. The reason is that if the noise level is too high then a false
limit will be hit and the zeroing of the axis will not be correct. The method that will
be used to prevent the device form generating high levels of noise is to connect a
pull up resistor to the pin that is used for the input. The value that was chosen
was a one kiloohm resistor. A capacitor was also be added to the circuit to further
reduce the noise on the lines. The capacitor that will be used is a multilayer
capacitor. The value for this capacitor is 0.1 microfarads. This limits much of the
noise that is generated from the stepper motor and the other electronics. There is
also a limit switch that is placed on the extreme locations of the axis. The reason
is that in case the commands that are sent to the microcontroller are incorrect or
that there was a false homing position, then the stepper motors will not destroy
the machine. The microcontroller will also send this data to the computer when
this fault occurs. The computer also can have the ability to send to the
microcontroller the command to go to home. This would rerun the same code
that would run when the microcontroller turns on. The microcontroller is able to
maintain the position of the axis. This is important if the computer would send an
absolute position rather than a relative position. The microcontroller is also able
to move the stepper motor in full steps as well as the microsteps. The stepping
space of the microsteps only needs to be about two microsteps per major step to
get the proper resolution, but more microsteps will be used. This value is
calculated by taking that each major step for the stepper motor is 0.9 degrees,
and that the gear that drives the belt is a quarter of an inch, so for each major
step the movement is around two thousandths of an inch. So to get the resolution
of one mil, the resulting value was divided by one mil. This value then needed to
55
be rounded to the nearest decimal place to allow ease in calculating the pulse
widths in order to create the microsteps. The initial design of this system was to
allow the laser to have the accuracy of one mil, but to allow the laser to achieve
this accuracy the stepper motors must be able to move with an accuracy of more
than one mil; therefore, the microstep size was increased by a factor of four. This
allows the motor axes to be moved with a step size of 0.00025 per inch, even
though this accuracy was not possible through slippage through the gear and the
belts used to drive the axes. This helped in debugging the errors generated
through the machine.
10.2 – Prototyping and Testing
For the motor controller, the first step was to prototype the microcontroller. The
testing of the microcontroller consisted of looking at what the waveforms would
be when the microcontroller is moving the stepper motor to the home position.
The next task was to see if the waveforms stop when the limit switch is initiated.
The next step was to see if the microcontroller works with the computer by
sending an error message when the other limit switch is activated. If all the above
tests worked, the next step was to build the board with both the microcontroller
and the full bridge driver. This was then connected to one of the stepper motors.
The testing of this stage would consist of seeing if the stepper motor moves in
the correct direction and to verify that the stepper motor microsteps correctly.
The last task would be to verify that everything is working correctly.
10.3 – Building
For building the stepper motor power supply and controller unit, the first step was
to redesign the board that was used for prototyping. The newly designed board
was then etched out of the copper substrate. After the components are soldered
onto the board, the next step was to test the board. Testing the board would
consist of checking that the limits work correctly and that the emergency
shutdown works. The next step was to verify that the motors are moving the
correct way. The last step was to connect it to the XY table and check to see if
everything worked correctly.
10.4 – Evaluation Plan
For the design of the motor, this project uses a unipolar stepper motor. The
controller that was designed used the L298. The first step in completing this
portion of the project was to first prototype the microcontroller. The next step was
to design and prototype a board that contained the microcontroller and the full
bridge driver. The next step after building the prototypes was to test the designs.
The last step was to build the final board. If a problem occured, the basic process
of first locating the error and to trying to recreate the problems on the prototyped
boards was the first step. The reason to try to recreate the issues is to see if the
result was an initial design flaw or a problem with the current board. The last step
56
was to connect the stepper motor control board to the XY table and test to see if
all worked correctly.
57
SECTION 11: XY Table
11.1 – Design
For the XY table, we needed a way to access all of the parts of the board to be
milled. There are two options on how to do this. We first discussed using a
system (Figure 11.1) where the laser is redirected above the PCB, and the PCB
is actually moved.
Figure 11.1: Alternative XY Table Design
This would have the benefit of having less moving parts in the laser system.
However, the seeming simplicity of this system is its own downfall. In this system,
we would have had to include a method for users to attach their boards to the
moving XY table. Furthermore, we would have needed to have the capability of
handling the different weights of the boards. Even a small change in either a
different size of a board, or a new board type would change the dynamics of the
system. This could add slip, reducing the accuracy of our milling process.
Because of these problems we decided on moving the optics assembly above a
stationary PCB. This is shown below is Figure 11.2.
58
Figure 11.2: CO2 laser engraver XY table
This eliminated the problem of needing to strap the board down, as all that needs
to be done is correctly align the board. This also solved the problem of varying
weights; the optics will always weigh the same. We may still incur slip, but at
least in this system we have a constant amount of slip over time, and can correct
for it in software. We also have identified the need to change the distance from
the board being milled and the XY head if different boards were used. This
necessitates the addition of a Z axis. We at first considered implementing a fully
motorized Z axis. This would have used a lead screw threaded inside a
stationary nut on the platform holding the PCB. This would have been useful as
we would be able to vary the height of the board through our main program for
various thicknesses of boards. This would be a necessity if standard users were
capable of using boards besides FR4, and a wide range of board thicknesses
were used as the focal point of the laser needs to precisely located at the surface
of the PCB. However, after some consideration we found this to not be the issues
and have disallowed the use of anything besides FR4 boards by standard users.
We had several design decisions left once we eliminated the motor controlled Z
axis. The first of those decisions was how to carry the mirrors and focusing lens
that will be moving above the PCB. We discussed the merits of using either
precision milled rods or open rail system similar to desk draw sliders. We decided
on using the precision milled rods due to the fact that the rail system necessitates
the use of exposed bearings. The open bearings would have accumulated too
much dust for this project, and would have required cleaning. It is possible to
construct this system using only two rods and a caster on the end that is not
connected to a motor. However, this allows for the possibility of the entire
assembly to torque and the machine to lose alignment with the board and the
laser. To counter this possibility, we have designed a system that uses four
precision milled rods, two for each axis. We then had to make a decision on
using either bushings or linear bearings. This was a simple matter of cost, the
bushing are cheaper to purchase than the linear bearings. This is important as
this is one of the most likely components of the project to wear out, and we will
need spares. To move the assembly we had the choice of using either Acme
threaded rods or a belt system. For similar reasons that we did not use the open
59
rail system, the Acme threaded rods have a tendency to get clogged.
Furthermore, we have observed that a belt system allows for faster movement
than a threaded rod system. To move this whole assembly we used two stepper
motors. For more information on these motors, please see Section 10 of this
report. The motor to move the Y axis of the table is stationary and has its shaft
horizontally oriented. Since we are using two precision milled rods that are some
distance apart for the Y movement of the XY table, we had to attach a long shaft
to the spindle of this stepper motor. This shaft had to be very accurately aligned
with the stepper motor spindle to reduce any possibility of wobble. We did this
through the use of adhesives and a stout collar mechanism. Then, we attached
gears to this shaft at the locations where the carriage is aligned. These gears
attach to the belts that drive the actual Y movement of the table. Since the two
belts for the Y movement of the table are synchronized by the common shaft,
there is no difference in the movement of each side of the machine. This is
critical to our design. The other end of the belt is attached to a stationary stand
that holds a freewheeling gear of the same type and size that is attached to the
timing rod. The X movement of the machine is slightly more complicated than the
movement in the Y direction. This is due to the fact that the entire X axis
assembly has to move in the Y direction on a carriage system. The stepper motor
for the X axis is attached to this carriage with its spindle oriented in the positive Z
direction. This spindle will be directly couple to a gear that will interface with the
belt driving the X movement of the XY table. The other end of this belt will be
attached to a stationary stand that holds a freewheeling gear on the other end of
the carriage. This carriage also holds the optics that are responsible for the
transfer of the final laser to the PCB surface. The laser will be shot down the left
Y axis of the XY table, hit a 90 degree mirror, move along the X axis of the
carriage, hit another 90 degree mirror, turn down, and then finally hit a focusing
mirror before hitting the PCB surface.
11.2 – Prototype/Test
We accomplished this with the attachment of a laser pointer directly to the head
of the XY table. Then, using time lapse photography we ran through multiple
iterations of moving the head around the board. Looking at the resulting image,
we were able to verify that the laser is hitting the same location every time it gets
to a certain point in the run. We were able to tell if the machine is running itself
out of square by this method also. After we had gone through this stage of
testing, we needed to align the real laser with the head and do a few test runs
using the same time lapse photography system as before, but with the laser on a
very low power setting. The next test that we performed was one to remove the
time lapse photography system altogether. This was a test with the laser on a
medium power setting with a paper substrate laid down. Over a large number of
iterations of the machine being forced to go to different corners and travel
different paths, we were be able to tell if the machine is going out of square of if
we are accumulating slip anywhere if holes start to appear in new parts of the
paper.
60
11.3 – Evaluation Plan
We needed a working laser system, software system, and power system before
we can do a realistic test. While the tests that we planned were good metrics at
determining the accuracy of the table, the metric that we needed to evaluate is
speed. It took some time to do the mill procedure, and any amount of extra speed
that we could achieve in the movement of the mill head helped greatly. Our main
goal was to replace the current milling machine that resides in the senior design
lab. To be capable of replacing this machine, we needed to be able to approach
it in speed of milling. The capability of the laser to punch through the copper
determines one part of this, but the speed of the movement of the XY head
determines the other. If the PCB that is created by our machine matches the
users input specifications, and is created as specified by the users input Gerber
file then this section was a success. A further evaluation that could be used is the
comparison of a professionally milled PCB with a PCB that our machine creates
using the same Gerber file. If the two matches are identical, then we would
definitely call this section of the project a success. However, even if the PCB’s
are very similar we would have considered this section a success. This machine
is only intended to be used in the prototyping of PCB’s, not production runs or
extremely precise work.
61
SECTION 12: Microcontroller
12.1 – Design
The first step in the design of the microcontroller section of our project was the
decision of what our project requirements were. A Pulse Width Modulation
(PWM) circuit will be needed to allow for microstepping of our motors. The
microcontrollers that we decided to use also needed to be able to interface with
sensors, while being easy to implement, program, and low cost. We are planned
to implement seven types of sensors: contact, temperature, light, current,
voltage, humidity, and flow rate. In total, we planned to have twenty three
sensors inside our project. Due to this large number of peripherals that we will
have to interface with, we know that one microcontroller was not able to handle
all of our needs. The microcontroller that we used must need to be easy to
implement. We have a large range of systems from various disciplines going into
this project, and the last thing that we needed was to incur artificial complexity
due to a poor design decision. For this reason, we have decided to eschew any
microcontrollers that require assembly programming and are only going to look at
microcontrollers that support the C programming language. To help easy
implementation, we also searched for a microcontroller that has a large
repository of code already written for it. It will also need to be easy to reprogram.
We are only human, and it is very likely that we would make a mistake in the
programming of the microcontroller. The harm of this was limited some by having
the main program handling occur on our control computer, but errors were still
most likely occur in our microcontroller code. For this reason, we have decided to
try and find a microcontroller that can be programmed in the final circuit of our
design, without having to be removed. We did not want to have to remove
precision aligned optical equipment to just change a few bits around in a
microprocessor, so this is a very important requirement. This requirement
necessitates that we use a microcontroller that has Universal Serial Bus (USB)
support if we do not want a large number of wires inside our project box. Our final
requirement is that the microcontroller be cheap. This seems slightly counter
intuitive, as we have some very expensive optical equipment in this project.
However, we went through a fair number of these microcontrollers as we move
through multiple revisions of our project, and went through testing of all of the
various components. We could not anticipate all of the problems that might occur
this early in the project, but we weanted to be able to swap out microcontrollers
fairly frequently if we need to. For this reason, we did not use a system such as
the Arduino that has a support system already built in, but for a cost almost an
order of magnitude larger than a bare microcontroller. A major factor that
contributes to the cost of all microcontrollers is their data bus width. We expected
that a 16 bit data bus width would suffice for our needs. However, these
microcontrollers are significantly more expensive than their 8 bit counter parts.
To overcome this obstacle, we have decided that we will search for an 8 bit
microcontroller. The difference is in how we would program this microcontroller.
62
The standard ANSI C programming language allows for an integer to be created
that can be two data entries wide. This allows an 8 bit microcontroller to be used
as a 16 bit microcontroller, with a minimal loss of performance [18].
The lead microcontroller designer in our group was already familiar with two
microcontrollers that fit these needs, the Microchip PIC18F2550 and the
PIC18F4550. These two microcontrollers are nearly identical, except that the
PIC18F4550 has one more “Capture/Compare/PWM Peripheral”, 3 more ADC’s,
8 more output pins, and a streaming port [19], [20]. Very early on we decided to
eliminate the PIC18F2550 from the competition. The PIC18F4550 offers more
functionality in a slightly larger package for just slightly more money.
Furthermore, this microcontroller meets all of the specifications that we have
defined above, while still remaining cost effective. Another plus if we were to use
this microcontroller, would be that we already have the software setup, and the
support hardware (namely a microcontroller programmer) that we would require.
This would allow us to cut down on our initial development time considerably. We
were unsure of where to go from here in finding other microcontrollers to
research for our project. However, we found that the Mouser Electronics website
has a tool that can find similar products. We used this tool on the listing for the
PIC18F4550 and found the Silicon Labs C8051F342-GQ, the Atmel
ATMEGA162-16PU, and the Freescale Semiconductor MC9S08JS8CWJ [21],
[22], [23]. For this research, we decided to compare these microcontroller
offerings to the PIC18F4550 to try and determine if we could find one that
exceeded its qualifications. Through former work in a similar area of one of our
members, we already knew that the PIC18F4550 would work for our purposes
from firsthand experience. At first glance the Silicon Labs C8051F342-GQ
appeared to beat the PIC18F4550 in nearly every field. It has a larger program
memory size (64KB versus 32KB), a larger data ram size (5.25KB versus 2KB)
and the same clock frequency. However, this microcontroller only has twenty five
programmable input/output pins, is in a surface mount package, and is twice the
cost of the PIC18F4550. We still did not know how many sensors our final project
would have, and we did not want to limit ourselves with a microcontroller that
does not have enough input/output pins. Furthermore, we were trying to avoid
the use of surface mount components in the microcontroller section of our
design. We wished to do this to aid in the prototyping process, as it will allow us
to use breadboards without creating custom breakout boards, and to ease the
soldering process for the final product. The Atmel ATMEGA162-16PU is in the
type of package that we wish to use (PDIP-40), and has the same number of
input/output pins as the PIC18F4550, but is seriously lacking in other areas. First
off, it only has a maximum clock frequency of 16MHz. We planned on doing a
large amount of data manipulation and calculation, and did not want to be limited
by a slow clock rate. Also, the ATMEGA162-16PU only has a 16KB program
memory size. We would be loading complex path following and contour routing
algorithms into this microcontrollers memory, and once again we did not want to
be limited by buying a substandard chip. While we could have looked at the next
microcontroller up in Atmel’s offerings, the ATMEGA162-16PU was already over
63
our $5 limit for each microcontroller, and Atmel’s offerings just increased in price
from there. The Freescale Semiconductor MC9S08JS8CWJ microcontroller
really never had a chance of being selected. It only has a 16KB program memory
size, a 256B data dram size, and only one timer to the PIC18F4550’s four timers.
This microcontroller looked appealing, as it is only half the cost of the
PIC18F4550 at the time of this writing, but it just does not have the requirements
that we believe will be necessary to make this project a reality. After some
consideration, we have decided to use the PIC18F4550 as the main
microcontroller in our project. It is simply has more to offer in terms of hardware,
has a larger user community on the software side, and we already have
experience and support hardware in developing for it. The next decision that we
needed to make, was on which topology that we wished to implement.
One method of designing the microcontroller system for our project that we
considered using was to connect the computer to one main microcontroller, and
that microcontroller would then be used to send and received commands and
data from other less complex microcontrollers. These less complex
microcontrollers would be used to do the interfacing with each of the individual
sections of our project. These sections include the power supplies and the motor
controller. One attractive aspect of this design is that these less complex
microcontrollers, we were discussing using PIC16F54’s, would be much cheaper
to purchase than the PIC18F4550 that we have ended up deciding to use.
However, the main reason that we have decided to not use the PIC16F54’s is
that they are not capable of running assembled C code. We would have had to
write all of our code in assembly. After some deliberation, we decided that this
would add too much time to the development cycle for a small monetary reward.
Our project is going to be complex enough that we do not need to be artificially
adding difficulty. In this design, we were going to use a PIC18F4550 to do the
control of the other microcontrollers, and to interface between the machine and
the computer over USB. For our next design, we decided to solely use this chip.
The final design that we have come up with for the microcontroller of our project
is a topology of one to many. This has the computer directly connected to many
microcontrollers. This is possible since every PIC18F4550 that we are using can
be directly interfaced with through a USB connection, and a theoretical limit of
127 devices can be connected to a single USB host controller [24]. We have
nowhere near this many microcontrollers in this project, and we have a dedicated
control computer that has at most three USB devices (mouse, keyboard, and
flash drive) connected to it at any time. The main reason that we have decided
on this design is one of simplicity. These microcontrollers will allow us to do all of
our circuit programming while they are already in the project, with only one pin
(pin RC0) pulled to high and a restart of the board. With this design, our main
controller will be the computer. This allows us to easier record errors, and allows
us to have tighter control of how we implement procedures.
64
Now that we have decided on which microcontroller that we are going to use, and
what topology we are going to use the interface the computer with our multitude
of sensors, we needed to decide on what protocols we are going to use to talk
between the computer and the microcontroller. As the PIC18F4550 is fully USB
2.0 compliant, we have a data bandwidth of 12Mbps and will be sending 64B
packets at a time. This is a standard of the USB 2.0 specification. Following the
standard for most microcontroller applications, the first character, or the first eight
bits, of this packet will be a control code determining what function to run on the
microcontroller. The other 504bits are used for data. This is an entirely arbitrary
setup, but we are able to call 2^8=256 functions with this method, while still
having plenty of space for our data to be passed. The following functions can be
viewed from the aspect of the compute sending the microcontroller commands.
These functions include moving the XY head in the X direction, moving the XY
head in the Y direction, turning on the laser, turning on the Q switch, turning off
the laser, and a verbose senor reading state. Due to the way that we are
planning on designing our software, we are be sending a command every time
we want the XY head to move, and therefore these commands take up the bulk
of what we are sending to the microcontroller. Furthermore, we needed to decide
if the Q switch should be on or off for each location that the XY head is capable
of being located at. For this reason, and due to the nature of the Q switch, the
command to cut the Q switch on and off is a pulsed command that is a sub
routine of the X and Y movement functions. To simplify the move functions, there
is only one move function that reads in specific digits that tell the machine to
move in the X or the Y direction. There is another bit that tells the machine to
pulse the Q switch. The laser is cut on at the beginning of the mill operation, and
left on until the job is finished, as it will not actually be lasing in this state and thus
will not be actually removing any material. The verbose sensor function will be
used to do diagnostics on the machine, and will give priority to sending a large
amount of sensor readings back to the graphical user interface. This will be very
useful when we are fine tuning the laser output power. Finally, we implemented a
shutdown function. This function will read specific bits out of the data regarding
which parts of the machine need to be shutdown, and if the machine needs to go
through a safe and normal shutdown or if it needs to do an emergency shutdown.
An emergency shutdown has a high likelihood of damaging equipment, and
would only be utilized if there is an imminent safety concern. We can also look at
the functions that are solely focused on sending data to the computer. These
functions are mainly sensor readings and log file operations. These files will take
a background existence to the commands coming from the computer, as they are
less important. Since we have a flat topology, we are able to poll and send
sensor readings. They are not real time, but will be only slightly delayed. This is
required as we have a large number of sensors, and each sensor requires the
use of the analog to digital converter for it to be read. For other things, however,
such as the limit switches on the XY table, or the contact switch on the case, we
will be able to read those in real time as they will be coming in as a digital signal.
These digital signals can be read on a per buss basis. This is necessary as we
want to stop the XY head as fast as possible if it has hit one of the limit switches,
65
and we want to shut down the laser system immediately if someone has forced
open the case without entering the administrative override. We are planned on
having two commands go to the machine for every one command that comes
from the machine. This gave us a good data rate for sending commands, and still
gave us a good sensor and log resolution on the computer side of things. We do
not have a fancy method to handle memory bandwidth bottlenecks. We simply
enter a loop until the data that has been requested is received.
12.2 – Prototype/Test
The figure below (Figure 12.1) shows the schematic that we used to create the
prototype board, and Figure 12.2 shows an actual picture of the board that we
have created for testing.
Figure 12.1: Schematic used to create prototype board
66
Figure 12.2: Picture of the actual prototype board
This board was created in house using an etching process. A FR4 board was first
evenly coated in a layer of black spray paint. The portion of the coating over
areas of the PCB where we wanted the copper to be removed was then etched
off using a CO2 laser on a professional laser milling machine. The now exposed
copper was then removed using hydrochloric acid and hydrogen peroxide. This
finished board will be used for the prototyping of each section of our project that
is going to require a microcontroller. There will be a microcontroller for each axis
of the XY table. The main purpose of the prototype board was to determine which
pins we will need to use for that section. This allowed us to make the final boards
with the exact layout that is required.
Extensive testing was be required before we can actually use any of the
microcontrollers in our project. The number one thing that we needed to test for
is logic errors. This included things like the machine reporting an error, and a
result occurs such as the laser power increases instead of the machine shutting
off. This took the longest of all of the tests, as a large number of combinations
needed to be tried. The main thing that we were be looking for, however, was
simple mismatches between event codes and actions. So, while this took the
longest to test it took the shortest amount of time to fix. After we have assured
ourselves that there are no logic errors in any of the microcontroller code, we
tested the connections between the microcontroller and the peripherals
themselves. This involved connecting each individual sensor to the prototype
board, and applying a stimulus to the sensor and verifying that the correct result
is received by our software. After we had verified that each individual sensor
67
worked, and that the microcontroller is reading those sensors and getting the
data back to the computer correctly, we tested the combinations of sensors that
will occur in each section of the machine. This is to make sure that sections that
have to process multiple sensors that use the analog to digital converter, such as
the laser sensor microcontroller, are capable of handling reading all of the
sensors, and reporting any changes to the sensors in a timely fashion.
12.3 – Build
We used the same process that we used to design the prototyping board for the
actual boards that went go into the machine. The difference in the actual boards
and the prototyping boards is that each section’s board was custom made for it.
68
SECTION 13: Software
13.1 – Design
For this section of our project, we first needed to decide which programming
languages we wanted to use. The C programming language was a given for the
embedded portion of the project due to the PIC2550’s support of it. This allowed
us to avoid the tedious task of hand coding assembly. The other reasons for
choosing the PIC2550 have been discussed in Section 13. For the portion of our
project that is running on the control computer we have chosen to use the Java
programming language. The other options that we considered were the C
programming language and Visual Basic 6. The main reason that we chose the
Java programming language was that our main coder for the GUI and
computer/machine interactions has had the most experience in graphical user
interface design in the Java language. The C language does not support the
creation of a GUI as easily as Java does, from a standpoint of object oriented
programming as well as from the standpoint of the Netbeans IDE available for
Java. We did not chose to use VB6 as our language of choice as there are some
issues with it interfacing with a PIC microcontroller, along with our coder not
being familiar with the language. Furthermore, the jPicUsb Java library allowed
us to easily send and receive data from the PIC microcontroller from within Java.
The Java programming language was used to display the GUI to the milling
machine user, make calculations, and send data to the machine itself. The GUI
initially presents the user with a simple interface to load a Gerber file or to load a
PNG file. The Gerber file format is able to be produced through a wide range of
PCB layout software, and is the current standard used by many professional
PCB manufacturers. The two main files that we are concerned with are the *.GTL
and *.GBL which are the layers for the top layer copper trances, and the bottom
layer copper traces respectively. The format supports other layers such as a silk
screening layer (*.GTO) and a solder resist layer (*.GTS), but these are not
implemented in this project. If a project requires those two layers, the user will
need to contact a professional PCB manufacturer to have their project
processed. The processes used to transform the Gerber files into a format that
our machine can handle are discussed later on in this section. To determine the
user’s access level, a user name and password combination is used. This was
initially going to be done with a hardware based user access card that stored the
user access level in an encrypted format. This turned out to be one too many
additions to this project and has been cut for the foreseeable future. The user’s
password will be assigned to the user at the time of their milling machine
certification. The user name is the students NID. The password is a random
string generated from lower case letters and numbers that is between six and
eight characters in length. This access level determines the interactions that a
user is capable of with the machine. These interactions are discussed further in
the Security section of this paper, Section 15. The program then takes the bitmap
69
generated from the Gerber file and allows the user to place it in the valid areas of
the board.
The GUI for this project will consist of several parts. The first part is the window
(Figure 13.1) that will allow a user to log into the system. The main part of the
GUI will contain the rest of the options, including the placement of the image on
the board and the options for the users above the level of Standard User. These
features will be discussed in more detail below.
At this stage the user will enter their user name and password combination in
Figure 13.1 that was assigned to them at the time of their certification. This will
determine the user’s access level so that they are not shown restricted options in
the next step. If the user name and password combination do not exist in the
memory of our program the user will be returned to the dialog asking for their
user name and password until they enter a valid combination or exit the program.
The user name entry will be a simple JTextField, but the Password field will be a
JPasswordField, so that the password is hidden when the user is typing it in.
Figure 13.1: Enter a User Name and Password
In this stage, the Next button will call the method authUser(userName, password)
to determine if the user is a valid user or not. Additionally, this method will return
the user’s access level. If a -1 is returned the entry is not valid, if a 0 is returned
the entry belongs to a Standard user, if a 1 is returned the entry belongs to an
Experienced user, and if a 2 is returned the entry belongs to an Administrator.
This authentication will be accomplished through the use of the SHA-512 hashing
function. The user name and password pairs will be stored in a plain text file in
the same directory as the main program. Each time a new user is certified, their
user name will be entered into the file along with the hashed value of the
password they are assigned, slightly modified to provide the user’s level. Instead
of simply storing the user’s password as a hash, we will compute the hash of a
combination of the user’s password and their assigned access level. For
example, if the user aa123456 is assigned to user access level 0 and they are
assigned the password b1234cd, the string that will be hashed is “0+b1234cd.”
When the user gets to this window, all of the user name and hash key pairs will
70
be loaded into a TreeMap. When a user enters their credentials at this step, the
user name they enter will be used as a key for the TreeMap data structure. This
will return the stored hash key, and this stored hash key will be compared against
the four hash key’s that are possible for this user’s password that was just
entered. The user will be granted their access level depending on which of these
hashes matches the hash on file.
The security applications of this implementation are severe. Since hash functions
are one way, it will be difficult for someone to guess a user’s assigned password.
However, if the user obtains access to the main password file, it would be trivial
for them to change their access level by simply re-entering a hashed entry of a
new access level concatenated with a password. We planned to eliminate this
possibility by dynamically setting the permissions of the user access file and the
user data folders. The user access file would then only be able to be read or
write when the main program needs to read or write it, and the user data folders
will only be able to be read or written when the correct user is logged in. We
thought this would be possible through the use of the windows command line tool
CACLS, however it is not.
This is the main area where the user will interact with the milling machine. Figure
13.2 shows the layout of the full GUI that an Administrator would see. All of the
content in this window is displayed based on the user’s access level. This figure
shows what a user of access level Administrator would see after logging in. If an
Experienced User logged in, they would only see the Standard and Experienced
user boxes. The reason for displaying the options like this is three fold. First, the
GUI is shown in this manner in the spirit of security. There is no need for users to
be able to see the options available to them at a higher access level, and be
tempted into trying to trick the system into granting this higher access level. This
sounds like the principle of security through obscurity, but it is mainly to take
away the temptation for users to break the system. The actual methods that will
be controlling the features contained within the GUI will all be locked based on
the user access level. Second, this is done for reasons of simplicity. The vast
majority of users will be standard users, and will not need to see the
Administrator options, even if they will not work for them. Third it cuts down on
the number of windows that need to be opened for the user to interact with the
machine. If we had spawned a separate window for each access level the
software would have been difficult to work with. The top row of Name, Access
Level, and Status are shown no matter what level the logged in user is. The
userName and accLevel fields are not modifiable, and are included in this
window to allow someone to easily tell who is logged in by glancing at the
screen. The userStatus field will display either OK or ERROR, with OK being
displayed when the machine operating at full capacity and ERROR being
displayed when the machine has any error occur that limits the operation or
safety of the machine and the user. This status is left purposefully vague, as the
Standard level users will not be able to fix any of the errors that are serious
enough to result in an ERROR status.
71
Figure 13.2: Shows how the individual user sections fit into the main GUI
The Standard Users box (Figure 13.3) within the main GUI is where most of the
work will be done, and is accordingly the largest area on the screen. This box will
contain the area where users will load their file, place their image on the available
mill area, reconnect to the machine, and start the mill procedure. Furthermore,
the user will be able to redisplay the console if they have closed it by accident
from
this
pane.
72
Figure 13.3: Standard user input pane
The load file button will produce a GUI that displays the possible locations for an
image to be placed. The area that will be used for the placement of the image on
the mill area will be the most complicated part of this GUI. This area will start off
with a 700x700 pixel square drawn on the Java Canvas class with the paint()
method. This hollow square will be in white, and represents the area that can be
milled on a new FR4 board. Then, a filled in red rectangle was going to be drawn
on top of this square to represent the area that cannot be milled due to a user file
being loaded. If no user file has been loaded, then there will be no red rectangle
drawn. This feature was removed because of time limitations, and currently
user’s simply need to know the extents of their boards and be wary of them. The
user image was going to start off following the computers mouse. This was to be
accomplished through the use of the getX() and getY() methods of the
MouseEvent class. The result of these methods would have been used to move
the rectangle representing the users Project image around the board, based on
the location of the top left corner of the image. This feature was removed in favor
of allowing the user to move the image around with the arrow keys, as this was
easier to implement and less confusing for the user.
The isValidImageLocation() method was to be used to determine if the image
that the user is attempting to add to the board is in a valid location. It was to be
accomplished through two checks. The first check was to be based on the
centroid of the image being placed and the remaining area that can be milled. If
the centroid of the image is within that area, the image will not be allowed to be
placed there. This will be used to eliminate any images that are too large for the
remaining area that can be milled. An example of this case is shown below
(Figure 13.5).
73
Figure 13.5: Invalid image placement locations, condition one
The second check (Figure 13.6) would have been based on the lines that make
up the image, the maximum area that can be milled, and the area that has
already been removed. If the lines belonging to the image that needs to be
placed intersect any of the other lines on the display, a false will be returned.
This would have been accomplished through the use of the
intersects(Rectangle2D p) method in the java.awt.geom.Area class. This method
returns true if the current Area object is intersecting the specified Rectangle2D
object. This will take care of the cases that appear on the following image. Note,
only one image is actually going to be placed on the window at a time, this is just
showing the possible areas of conflict that the intersects(…) method will detect.
Figure 13.6: Invalid image placement locations, condition two
We were unable to implement this check due to time constraints. In the current
design of our machine the user needs to be cognizant of the amount remaining of
their board and where the home position of the machine is to keep from running
out of available PCB material. We were going to implement an automatic
placement algorithm, but time did not permit and without a valid image location
check it would have been impossible. This algorithm would have taken the user
image and attempted to place it anywhere in the image. As of right now, the only
way that we know of doing this would be to start with the two corners of the
image and the maximum area that can be milled and move the image over pixel
by pixel until the right most side of it reaches the right side of the millable area.
Then move down a pixel, and sweep to the left. Every time the image was moved
a pixel, the isValidImageLocation() method would have to be called. In the worst
case scenario of a 1pixelx1pixel image, this would result in 144,000,000 calls to
the isValidImageLocation() method. This is an unacceptably large number, and
74
until a better algorithm can be found for moving the image around this will not be
implemented.
We were going to allow users to create a user file that could store a previous
boards remaining millable area. Clicking of the New User File button would have
performed several tasks. First, it would have saved and closed the current user
file. Every time a user file is created, it is named by the convention of
YYMoMoDD_MiMiHHSS.user where DD is the day in two digit format, MoMo is
the month in two digit format, YY is the year in two digit format, and MiMiHHSS is
the time. This user file would then be displayed onto the GUI in the Millable Area
box. Furthermore, the corresponding entry to the millReady String would be set
to a zero, explained below, as the image has not been properly placed on the
new user file. Even if the image is in a correct location, it must be moved at least
once, even if it is back to the starting location, for the isValidImageLocation()
method to trigger. The Load Previous User File button would have presented the
user with another browse window, much like in the earlier part of the GUI, that
would only have displayed *.user files. This browse window will default to the
user’s directory on the controller computer. The last thirty saved user files would
have been stored in this file, along with the last thirty PNG files that have been
created from user input Gerber files. These features were not implemented due
to time constraints.
The Mill button will do just that, and start the mill procedure. This will first check
with the millSet() method. The millSet() method will check the millReady String to
verify that all entries are a 1. We have designed the system this way so that we
can easily implement additional checks before the mill procedure is ran. While we
have not decided on the final length, or what all will be included in the checks, if
there is even a single 0 in the String readyToMill will stay false and the mill
procedure will not be allowed to move forward. A for loop checking each entry
will be responsible for the verification using the substring() methods and the
length() method of the String class. If the string is found to contain all 1’s, the
readyToMill Boolean variable will be set to true. Once the millSet() method
returns true, the mill() method will be run, and the readyToMill variable will be
returned to false. This is to prevent users from double clicking the Mill button, and
causing two mill procedures to run simultaneously. And finally, the saveUserFile()
method would have been run. This would have saved the user file in use to the
user’s folder, and check if there are more than thirty user files present. If there
are more than thirty, a policy of least recently used will be implemented, and the
oldest files will be deleted until there are only thirty left. Due to this policy, it is
unadvisable for users to store their custom user files in this directory if they wish
for them to remain for more than thirty milling cycles. The mill() method will be
discussed in this section below.
We had planned on including an option for user’s to enter the dimensions of the
remaining millable area of their board. After some consideration, however, we
decided that this was too advanced a feature to include in the GUI of the
75
standard user, but something that standard users might still find useful and thus
something that could not be exclusively part of one of the higher user access
level groups. Therefore, we have decided to just include documentation on the
format of the user files, and allow users to input the data themselves into a file. A
final format for the user files has not been decided on yet, but it will most likely be
pixel coordinates outlining the areas that have already been removed from the
board. A single entry would be sufficient to remove a rectangle from the board,
but any other shapes would require multiple entries, each describing a
rectangular area that cannot be milled on. The conversion factor from inches to
pixels is simply 1000 pixels per inch. Therefore, if you have a 12inx12in board,
and a square that is 3inx3in has been removed from the top right corner, the
entry in the custom user file would be ‘9000 12000 0 3000’. The first two
numbers are the X coordinates that have been removed, and the second two
numbers are the Y coordinates that have been removed. We will use the
convention that the top left corner of the board is 0,0.
Figure 13.7 is the window that the user will see after clicking on the load file
button on our main GUI. The options available to them in this window entail
entering a file name or the more advanced option of entering a PNG file. Both of
these options can be accomplished through the use of a text box and simply
typing in the full path to the file, or using the Browse button to the side of it to
locate the file on disk. The Clear button will just clear the specific text field. This
allows a user to accidentally enter a file name, clear it, and then enter a PNG file
without causing an error. If the PNG file option is used, the user needs to have
cleared the file name box before proceeding with the Next button, and vice versa.
The user will only be allowed to proceed past this stage if no errors occur. Errors
will be caused by having both an entry in the File name, and PNG File fields, if
one of the files that are specified does not exist, or if a required field is left blank.
The only required fields are the File Name or PNG File fields, but not both at
once. If an error occurs, the same window, with the same inputs, will be shown to
the user after an explanation of the error is shown to the user in a popup window.
The user will be given two more chances to fix the error using the same inputs. If
after three errors, the user has not cleared all of the errors, all of the inputs will
be cleared for the user to try again. This step does not ascertain the fitness of the
input files, it just accepts them. After the user presses the Next button, only then
does the system try to determine if the files are well formed.
76
Figure 13.7: Select a File Name and User File
The Browse button functionality in this section is handled by the Java
JFileChooser class. This allows us to show a fully functional browsing window
that we can pull the path name of a file from, and still limit the user to what types
of files they select. We used different instances of the JFileChooser class for
each of the two Browse buttons. This is required as we want each button to filter
to only a specific type of file that can be selected. The File Name browse button
will only be able to select *.GTL and *.GBL files while the PNG File browse
button will of course be only able to select *.PNG files. This would cause some
inconvenience to the user, as each new instance of the JFileChooser class would
default to the logged in users home directory while most of our users will be
either accessing their files off of email or a flash drive. We will solve this problem
by saving the first directory that the user enters with the getCurrentDirectory()
and setCurrentDirectory(File dir) methods of the JFileChooser class. Every time
a browse button is pushed, a check will be run if a directory has been used
besides the default directory of My Documents. If a new directory has been used
to open a file, then that directory will be displayed when the Browse window is
displayed. This will be useful, as most users will have all of their files in either a
single directory or in directories that are close to each other on the file tree. The
result of the file selected in the Browse window will be stored in the text box next
to the specified button with the getSelectedFile() method. Since this is a text box,
the toString() method of the File class will need be called to be able to display
this information. This String will be written to the text box with the setString(String
str) method of the TextBox class. The Clear button will simply use this same
method to set the String in the text box to null.
The Next button will be responsible for several tasks. First of all, it will popup the
next window, but with the Go button disabled. Next, it will perform a final check to
verify that there are no errors in the types of files in each text box, that either the
File Name or PNG File field is filled in, and that the files actually exist at the
specified location. These tasks are accomplished through the use of string
77
comparison, Boolean logic, and by attempting to open the files while checking for
a FileNotFoundException to be thrown. The logistics of these tasks are trivial,
and will not be mentioned in this report nor will tasks such as this be mentioned
again. If there are no errors from these procedures, the next step of determining
if these are well formed files is done. The easiest way to check if the input Gerber
file is a valid Gerber file is to actually run it through the gerb2tiff program, and
check the error messages that are returned, if any. If any error messages are
returned, the first GUI window will be redisplayed with a copy of the error
message shown in a popup window. To check the PNG image we will have to do
a pixel by pixel read of it to determine its fitness. Since this is done in the
conversion of the PNG image to a format that the machine can use, we will
simply run the convertPNG(String str) and issue a flag if a discrepancy is
discovered. A discrepancy would be if the file is composed of anything besides
black and white pixels, or if the file is too large. The board size would be too large
if the PNG file is more than 12,000 pixels in either dimension. This check will also
be completed on the PNG that is automatically created if the user chooses to
input a Gerber file, to keep users from trying to mill projects larger than the
largest available milling area. We arrived at this number as we will support a
maximum of 12inx12in boards, and will be running our XY table at 1000dpi with
one dot equaling one pixel. Again, if a discrepancy is discovered the first GUI
window will be redisplayed explaining to the user that their PNG image is
malformed. If no errors are found, the first GUI window is closed, the Go button is
enabled, and the machine will go through its startup procedure if it was not
already on.
The next conditionally displayed box (Figure 13.7) is the one that would have
belonged to the Advanced user group. We decided to not implement Advanced
users in this implementation of the program. Advanced users would have all of
the permissions of the Standard users and will be able to modify the attributes of
existing boards and modify the available PCB materials. When a user clicks the
‘Select Current Board’ button, the window in Figure 13.8 would have appeared.
Figure 13.7: Advanced user input pane
78
Figure 13.8: Select Current Board
This window would have allowed an advanced user to select which type of board
is going to be mounted in the machine. This will remain the default board for the
machine until this setting is changed.
Figure 13.9 shows the window that would have opened when an Advanced user
clicked the ‘Modify Available PCB Materials’ button. In this window, the user
would have been able to select which types of boards that can be selected in the
Modify Board window discussed below. We have designed it this way to allow us
to be able to easily add and remove boards that our machine can work with, as
we will only be able to tell how well the machine works with a certain type of
board after we have tested that board on the actual machine itself. Any number
of the jRadioButton’s can be selected. We have included options to select or
deselect both individual groups, as well as every entry displayed.
Figure 13.9: Selection of PCB Materials
79
When a user clicks on one of the ‘View Saved Attributes’ buttons, Figure 13.10
figure will be displayed. None of these entries are editable, as this window is for
the sole purpose of allowing a user to see what that particular board’s previously
saved attributes are.
Figure 13.10: Information about PCB Materials
Figure 13.11 shows the window that would have opened when an Advanced user
clicked the ‘Modify Attributes of Existing Board’. Notice, Advanced user’s are not
capable of adding new boards to the system, only modifying boards that are
already in the system. This is accomplished through the use of the jComboBox
that will populate the variable savedBoardName. The entries that will be in this
drop down box will be decided on by the aforementioned ‘Modify Available PCB
Materials’ button. When a previously saved board has been selected, the entries
in this window will all auto populate with the data stored in the data structure
used to store the boards. For more information on how the board data will be
saved, see the section below on adding new board types. There will always be a
default saved for normal FR4 boards that can not be edited by any users under
Administrator level. Furthermore, the four laser entries in this window are only
suggestions. Advanced users are not laser experienced service technicians, and
adjusting these settings could cause harm to the machine. For this reason, when
an Advanced user makes a suggestion on the laser attribute entries, they will
only be able to use this setting in a test mill. A test mill will allow a user to run the
mill procedure, but will only allow the procedure to run for a set amount of time.
We will determine this amount of time after we have built the machine, and tested
its response to various laser settings. This test mill will be sufficient for the
Advanced user to determine if the settings will work for the current type of board.
Advanced users are not allowed to over ride any safety measures, and therefore
they will only be able to vary the laser power and pulse rate slightly. An
Administrator must verify the laser settings from this window before they will be
available for a normal mill procedure.
80
Figure 13.11: Board modification options
The next conditionally displayed box (Figure 13.12) is for the Experienced user
access level. These users will be responsible for the training and certifying of
Standard and Advanced users, and will almost exclusively be members of the
Amateur Radio Club at UCF. For this reason, they will be able to add and modify
users, and to add new board types to the system.
Figure 13.12: Experienced user input pane
Figure 13.13 shows the window that will be displayed when the ‘Add User’ button
is pressed by an Experienced user. The userName should be the students NID.
The other two fields, Access Level and Password, will be conditionally displayed.
Since Experienced users can only add Standard and Experienced users, they will
only be able to select an access level of 0 or 1 for the new user. Furthermore,
experienced users will not be able to override the password restriction and will be
forced to use the ‘Generate Password’ button. Administrators will be able to
select any one of the four access levels, and will be able to manually type in a
password. The ‘Generate Password’ button will generate a password of the
format x####xx, where x is any lower case letter and # is any single digit decimal
81
number. This can be accomplished through an extrapolation of the code snippet
shown below (Figure 13.14).
Figure 13.13: Add a user to the system
Figure 13.14: Example of Password Generation Code
For the full program that will be generating our passwords, please see the code
titled ‘Password Generator’ in Appendix C. The isValidUserName() method will
be called when the ‘Done’ button is pressed to verify that the user name that was
entered is a valid NID, if the user is not an Administrator. The check to see if the
user name is a valid NID is bypassed when a user with the access level of
Administrator is logged in. After this is completed, a folder is created for the user
in the main program directory and its permissions are set. For more information
on user folder permissions, please see Section 14.
Figure 13.15 shows the window that will be displayed when an Experienced user
clicks on the ‘Modify User’ button. The isValidUserName() method will run after
every keystroke into the userName field. This will be done by checking the
KEY_PRESSED field of the java KeyEvent class after every entry into the box.
The isValidUserName() method will return true if the user name is found in the
system. We were originally going to verify that the user name is a valid NID here,
but we have decided to skip that step as Administrators can create user accounts
that do not have an NID for a user name. The user’s current access level will be
read out of the data structure that we are using when isValidUserName() returns
as true. User access levels will be placed in a TreeMap along with the user name
and hashed password pairs when the user access level files is first opened.
Experienced users will only be able to click on the 0 and 1 buttons, as they are
only allowed to modify a user to become a Standard user or an Experienced
user. Furthermore, if the user’s access level is returned as that of Experienced or
82
Administrator, an Experienced user will not be able to edit the access level at all.
Administrator users will be able to change anyone, even other administrators, to
a different access level. Due to the way that we are storing user passwords,
hashes, there is no mathematically possible way to retrieve a user’s password.
To overcome this, we will only be able to re-issue passwords to users. This will
be accomplished through the same method that the Add User window used, and
will only be available to Experienced and Administrative users.
Figure 13.15: Modify user account
Figure 13.16 shows the options that would have been available to an
Experienced user when the ‘Add New Board Type’ button was pressed. It is very
similar to the ‘Modify Attributes of Exiting Board’ button that is available to
Advanced users, but with the important distinctions that Experienced users are
allowed to add new boards and that Experienced users do not need to wait on
the authorization of an Administrator to change the laser power output. The
entries (Name, Substrate Material, and Copper Thickness), under the information
heading are for human information and records keeping only and do not come
into effect at all. The entries under the Data sections are the values that are
actually used by our main program and the machine itself. The width and height
values are used only in the main program, and the laser power settings will be
fed to the microcontroller.
83
Figure 13.16: Add a new board to the system
The Administrator section (Figure 13.17) of our GUI is by and far the largest
section. We used this section to test our laser system, XY table, exhaust system,
and TEC system while we are building the project. Furthermore, this section
helped us to troubleshoot any errors that arose, and adjust laser output power for
new board types. This section can be split into two areas. The top part is mainly
controls that control some aspect of the state of the machine. The bottom part is
mainly sensor readings to determine the health of the machine. It is important to
note also, that there are instantaneous state controls, and controls that only
matter for the next mill operation. The instantaneous state controls are the
movement of the milling head up, down, left, or right, forcing the laser either on or
off, and adjusting the lasers power up or down. The controls for adjusting the
amount of power used to remove either the copper layer or the copper and
substrate layers will only affect the next mill operation. This is possible because
the values that are stored with the board types have already been loaded into a
temporary data area. These controls only change that temporary data area, and
the next time that a user logs in, or a new board type is selected, the temporary
values are overwritten. Furthermore, Administrators have the ability to override
safety measures which entails opening and closing the cover during the
machines operating, turning on and off the TEC, and turning on and off the fume
extraction system. The last two controls that the Administrator has access to is
the ability to verify the laser settings (Figure 13.18) recommended by an
Advanced or Experienced user, and to force the stepper motors to run back to
the origin until they hit their limit switches. This return to origin button is
necessary because the stepper motors will lose their position data if there is a
power outage.
84
Figure 13.17: Administrator user input pane
85
Figure 13.18: Verify Laser Settings
As for the XY table head movements, we went with a stepped system for this
based on time. For the first five seconds that one of the up, down, left or right
buttons is held down, the XY head moves at one tenth of its maximum speed.
After five seconds, the head will move at half of its maximum speed, and after ten
seconds the head will move at its maximum speed. This algorithm will only apply
in the central milling area. When the XY head gets close to the edge of the
milling area it will be forced to slow down no matter how long the control has
been held down, just so that no physical damage is done when the head hits the
limit switch. Where applicable, the sensor data that is being displayed primarily in
the bottom half of the Administrator section will have units incorporated with this
display. This allows us to vary the scale on the displays in real time as the values
vary of a large range.
The same main Java program was also used for the backend of the user
interface. This backend was responsible for transforming the input file into a
milling format, performing operations on the created file, transmitting the file to
the machine, and handling errors thrown by the machine hardware.
The first step of the backend was transforming the user input Gerber file into a
format that can be sent to the machine. This is done by transforming each
Gerber file into a monochrome bitmap TIFF file. The conversion process is
handled through the use of the Gerb2tiff program created by Adam Seychell. Our
machine is capable of producing 1000 dots per inch (DPI), and therefore the
gerb2tiff program is passed the argument to set the DPI to 1000. This will help
keep the file sizes down, and will make further image process less time intensive
while not losing any resolution.
86
Since we have to do pixel by pixel operations on this image, and Java does not
natively support TIFF images, we first need to do an image conversion. The PNG
(Portable Network Graphics) image format was chosen for our final image type
because it is natively supported by Java and employs a lossless data
compression scheme. We did not want to use an image format that implements
lossy data compression, such as JPEG, and risk image artifacts ruining the mill
procedure. This conversion process was done through the use of the
ImageMagick software suite, most specifically the convert.exe binary. This
conversion can easily be run programmatically through Java in a very short
amount of time, one to two seconds on the tested TIFF transformed Gerber files.
The reason for the conversion from Gerber format to a tiff image, and finally to a
PNG image, is one of abstraction. This abstraction will allow us to implement
other milling formats in the future, if time permits, as all of our backend
processing of the image will be on a standard format PNG image file. Also, if a
user wishes to implement a nonstandard feature of one of the major PCB layout
programs, they will be able to directly input a formatted PNG file. A feature
allowing users to save converted PNG files from pervious runs will be included,
which will be the easiest way for a user to have a starting PNG to make changes
to that is in the correct format.
The final operation on the PNG file is the actual read. This was accomplished
through the use of the BufferedImage Java class. The getWidth() and getHeight()
methods of this class will allow us to create a double nested for loop that will
traverse every pixel in the PNG image. These pixels are then placed in a two
dimensional integer array of the same size. The locations on the board that will
either have the copper removed, or have the copper and the PCB material
removed, will be indicated by a black area in the image and a 1 in the integer
array, while the areas that will not be touched will be indicated by a white area in
the image and a 0 in the integer array. There are separate layers for areas that
just need copper removed and areas that need copper and PCB material
removed, as the laser will need to be set to a higher power and spend more time
at the location for the second case. This will allow us to do a mill of one side of
the board in two steps, copper removal and copper and PCB material removal.
Currently our laser system is not capable of penetrating the PCB substrate, and
this design for the input to the machine will allow us to at least accomplish the
copper removal successfully. Users could then be able to but the board on a
standard drill press and drill the holes themselves, while still getting a prototype
far cheaper and faster than the professional PCB manufacturers.
The next decision that we faced was to do a raster or a vector mill. Raster is the
obvious choice as our data on where to cut the laser on, where an integer of 1 is
stored, and where to leave the laser off, where an integer of 0 is stored, is
already in a raster format. We were initially hesitant to use this format though. A
raster mill procedure will take significantly longer than a vector mill procedure,
just due to the movement time of the XY head where the laser is not on.
Furthermore, a raster mill will add undue wear and tear to our drive motors and
87
table assembly. A vector mill would locate and mill contiguous objects, instead of
simply just sweeping over the entire accessible area of the XY table. This would
add a large amount of computation time to the computer side of the program.
However, it would reduce the wasted amount of time during the actual mill
process. This would be the most favorable way to run the machine, as it also
provides the minimum amount of movement of the motors, which translates into
the minimum amount of cumulative slip generated and thus provides the highest
tolerances. We have been unable to find an algorithm to successfully do a vector
mill operation on the shapes that would need to be milled in a PCB. Therefore,
we implemented the raster mill procedure. However, as we are implementing this
in an object oriented fashion, we designed the interfaces in such a way that a
vector mill operation will be able to be implemented in the future. This is made
possible due to the way that we are sending data to the milling machine.
Java was also used to interface with the PIC2550 MCU’s on the milling machine.
This was accomplished through the use of the jPicUsb interface created by
Geronimo Oñativia. The main portion of the data that we are sending to the MCU
will be relative position data along with a binary value of whether to cut the laser
on or cut the laser off. Since we will not be sending any complex pictures or data
sets to the microprocessor side of this project, it doesn’t matter how the
directions are provided. In the raster mill procedure that we will be implementing,
the directions are mostly along the lines of ‘Move Right/Left/Up/Down X Steps’,
‘Cut laser on/off’. The difference between this and the vector mill procedure is
that the directions would be telling the XY head to move around the shape of
objects. Either way, the microprocessor is only seeing commands on where to
move and if the laser needs to be on or not once we are in the main portion of
the program. The jPicUsb software only needs three commands to be
instantiated.
These
commands
are
“iface.load();”,
“iface.set_vidpid(“vidString&pidString”);”, and “iface.set_instance(0);”. After this,
data can be sent back and forth from the main Java program to the
microcontroller through the use of the Write(…) and Read(…) methods. Please
see Appendix D for the documentation of these methods.
The mill() method makes up the bulk of what could be considered our core
program. This method is responsible for doing two things: calling the
translateData() method and sending that data to the PIC2550. The
translateData() takes the internal data structures that we are using in Java and
translate them into something that can be used by the PIC2550. We will have to
send commands to turn the laser on and off, commands to turn the Q switch on
and off, move a distance in the X direction, move a distance in the Y direction,
and a go home command. The data on how to move the XY head and whether to
cut the Q switch on or off will come directly out of the array created from the
processing of the PNG file. This will be done line by line in the mill() method
itself. As we will not be sending any complex data structures to the
microcontroller, when the X array index moves over one location, we will send a
command for the XY head to move in the X direction one step. The same goes
88
for the Y direction. If the integer at that index is a 1, the Q switch will turn on, else
the Q switch will stay off. The number of times that the Q switch will pulse before
the head moves on to the index will be set by the type of board that is loaded or
based on what the Administrative over ride has been set to. After this entire run
has been completed, the XY head will move back to the origin. This will be done
through the running of the returnToOrigin() method. All that this will do is move
the XY head back towards the origin until the two limit switches in that corner
trigger. We are doing this to limit the amount of slip in our XY table between the
copper remove and the process that removes copper and PCB substrate
material. We have separated these two processes as we were unable to get
through the substrate material in our burn test. If nothing else, the user will have
a location to drill at and a drill bit width to use in that location if the machine fails
to punch through. After the mill procedure, they will be able to put the board on a
drill press and drill out their hole locations accurately. We will be using the *.drl
file of the same name as the Gerber file to get the data for the location of the
holes to be made. As this is a file intended to be used by a mechanical drill, we
will be ignoring a large portion of the file. The entries that we will mainly be
interested in are the tool sizes, indicated by the T prefix, and the absolute X and
absolute Y position. Please see Appendix B for an example *.drl file. This T code
would look something like “T1C0.025”. This indicates that tool one (T1) will be
making holes that are 0.025inches, or 25mils. The line indicating where this hole
would be placed would look something like “X4.775Y-3.625”, after a “T1” had
been issued. This absolute location would be where the center of the hole made
by T1 would be located. Instead of delivering this information directly to the
milling machine, we will be doing the processing on it in our Java code and
sending the same style of relative position and laser on or off commands that we
sent for the removal of the copper layer. To accomplish this, we were first going
to plot the locations that the exterior of the circle will touch in an array of
12,000x12,000 entries. This will equate one cell in the array to one mil on the
actual board. This would have resulted in an extremely large array that would be
difficult to work with. Instead of this, we have decided to use a multilevel indexing
strategy where we will take the entire 12000x12000 array and divide each
dimension by 8. Then, we will have an offset that will index inside this initial
index, and can select the correct bit within this index. The way that we were
initially going to accomplish this was with the code in Figure 13.19 below.
Figure 13.19: Code snippet of circle tracing algorithm
89
The code that we used in our final project is very similar to this code but with the
index translations described above. In this implementation (Ax, Ay) describes the
locations that the outside of the circle traces, (Bx, By) is the location of the center
of the circle, and D is the diameter of the circle as read from the tool entry of the
*.drl file. We started with a map of all zeros, and each position on the map that
intersects with the trace of the circle will be turned into a one. We will then take
this array, and go line by line cutting the Q switch on, just the same as before,
whenever the array index matches up with the stepper motor position.
Our main program’s backend will also be responsible for handling high level
errors thrown by the machine. Any immediate concerns that could affect user or
machine safety are handled by the onboard PIC microprocessors and be echoed
to the main program for logging purposes only. These errors (Table 13.1) are
things like a laser diode is getting too hot, the cover has been opened, or the XY
table is not responding. The solutions to these sample problems are outlined
below.
Problem
Laser diode overheating
Solution
Increase chilled water flow to that
particular diode OR reduce the power
to the diode OR shut off the diode
Cover has been opened
Stop all activity except for critical cool
down procedures (some exceptions,
discussed in Section 14.
XY table is not responding
Stop sending commands to the
steppers motors to attempt to lessen
the damage. Turn on a warning light
on the outside of the machine.
Table 13.1: Problems handled by the microcontroller
The main program is responsible for handling less concrete errors (Table 13.2),
and general errors that get thrown that we have not programmed specific
resolutions to in the microprocessors. These errors are things like the input data
is malformed, the machine is in an unrecognized state, or there are ambiguous
power fluctuations. The solutions to these problems are outlined below.
Problem
Input data is malformed
Solution
Request user to resend data unless
the user has a level of Administrator.
Then the machine will mill the given
data unless it would damage the
machine itself.
Request help from the user. Log the
actions that lead up the error.
Report
the
problem.
Increase
Machine in unrecognized state
Power fluctuations
90
monitoring of power fluctuations
unless the problem has disappeared.
If problem is determined to be
detrimental to the hardware, a flag
will be issued to shut down the
machine and not allow it to be
restarted without resolving the issue.
Table 13.2: Problems handled by the Java code
13.2 – Prototype/Test
We realized this was going to be a complicated project from the start. We have
lasers, moving parts, high power, and lots of data moving around. Because of
this, we started prototyping the software as soon as we had decided on the
project. The sections of the software that have already been prototyped are
described as follows.
We started a preliminary prototype of the software needed to pass data from the
main Java program and the program running on the PIC microcontroller before
the end of last semester. This attempt was hampered due to some difficulty we
have had in developing the hardware interface between the microcontroller and
the computer.
The software to transform the Gerber file to a TIFF file, and then into a PNG file
had also been prototyped at this time. The software that handles this is written in
Java, and will be incorporated into our main program but relies on two external
programs. These programs are command line oriented, and are thus easy to run,
send commands to, and receive info from, directly from Java. The program that
handles the Gerber to TIFF conversion is gerb2tiff by Adam Seychell. We call
this program with arguments indicating the location of the file that needs to be
transformed, and a parameter setting the DPI of the output image to 1000. The
gerb2tiff program supports the use of a user defined file name to save the TIFF
image as, but as this will just be an intermediary step we will be using the default
mode that simply appends a .tiff to the end of the original file name, extensions
and all. For example, if Project.gtl is input as the Gerber file to be milled, the
intermediary file Project.gtl.tiff will be created. So far in our testing of this
program, we have not run into any problems with the Gerber to TIFF conversion
process. A problem that may arise however is mentioned on Mr. Seychell’s
documentation of the program, the entire RS-274X standard is not implemented.
The gerb2tiff program will abort and report an error to the main program if the
parameters AS, IR, RO, IJ, KO, or SR are contained within the Gerber file. These
parameters are rarely used features of the specification, but if the gerb2tiff
reports this error we will stop the mill procedure. As of right now, we do not have
plans to manual read the Gerber file, and thus the file that the user is trying to
import will fail for our machine. The only recourse in this instance for the user will
be to use a different CAD program for their circuit layout, produce a PNG file of
91
the correct format, or contact a professional PCB manufacturer. After the
conversion from a Gerber file to a TIFF file, the method handling this portion of
the code will call another method to handle the TIFF to PNG. This method will
call the convert.exe program out of the ImageMagick software suite. We chose
this software as it is licensed under the GPL, and thus is free for us to use in our
project and is well supported in both code maintenance and tutorials with several
books having been written on the suite. While we are planning on using the
ImageMagick suite for the foreseeable future, we are not committed to it in any
way. Our project is only using a very small part of this advanced software suite,
and we do not need anywhere near this level of functionality for what we are
doing. Literally, all that we need to do is convert a TIFF image into a PNG image.
We have found that the convert.exe binary can run independently from the actual
ImageMagick suite. However, to the best of our understanding this would violate
the license that the software is published under. Due to this restriction, we will
have to actually install the entire software suite on the machine running our Java
program. While this is not necessarily a problem, as we are already constrained
to a PC running Windows, it does add one extra piece of software that would
need to be installed and configured correctly if our control computer is ever
replaced. If we are able to locate a program, or have time to design one
ourselves, that can convert a TIFF into a PNG image, and can be run as a
standalone binary under the GPL, we will switch our project over to use that
software. This will allow us to decrease the work load on the maintainers of the
project while increasing the portability of the project as well. Currently, we are
planning on simply installing the software on the machine and hardcoding the
location of the binary into the Java program. This will be an adequate solution as
long as the one computer that we setup is the one that always controls the milling
machine. If we have time, we will create a software package that can be installed
on any Windows based machine. This software package will have a text file that
contains the path locations of all of the necessary external programs, such as the
Java virtual machine as well as the ImageMagick suite. The text file will allow
new versions of the programs that we are utilizing to be installed. This way
whoever is the current person servicing the machine will be able to install the
files to any location that they wish. The Java program should run on any
operating system that has a Java virtual machine installed, however our project
will not. This is due to the fact that the way that we are running the external
programs is Windows specific. The programs that we depend on will only run on
a Microsoft Windows PC also. For these reasons, the machine that is controlling
the milling machine will have to be Windows based. It would be possible to run
our project on a different operating system, but substantial changes would need
to be made. The changes that are required for our project to run under a GNU
Linux operating system are outlined below in Table 13.3:
Method currently used
gerb2tiff program
Possible solution
Create or find a program that
does the same conversion
Find a program that can handle
ImageMagick program
92
TIFF to PNG conversion that is
not Windows specific
jPicUsb program
Create
a
custom
interface
between Java and the Microchip
API that is not Windows specific
Table 13.3: Possible solutions to the Windows dependence
The main changes are shown above, but other smaller changes would also need
to be done in the source itself. The changes necessary would mainly be in file
handling sections of the code.
13.3 – Evaluation Plan
We wish to write robust software, and have therefore extensively tested our
software. This was accomplished through the use of both black box and white
box testing procedures.
Black box testing is a method of testing that treats software as a ‘black box’ that
is impossible to view inside. This means that the code was tested with no
knowledge of what internal structures, methods, or external programs are being
used. We accomplished this testing by writing some automated software testing
programs. These programs are responsible for checking to see what exactly can
break our main program, and reporting that result to us. For each version of code
that is accomplished, the testing tools will be run against that code to ensure that
no functionality has been lost. In short, these testing tools will run all possible
ranges values into every possible input of the program. This may be somewhat
extreme, but it is necessary as we have a wide variety of technical competence
among the users of the machine. To accommodate these testing tools, we
implemented a command line version of our program that can accomplish the
same tasks as the GUI version. Currently, we do not plan on this being a
supported option for the end users, but if time is permitting we will write the
documentation and open the ability for them to use our program in this manner.
The testing tools that we ran included a test for multiple instances, a file input
test, a file format test, a PNG test, a user access check, and a user file test.
These tests are contained in one monolithic file that simply cycles through them
The multiple instances test makes sure that the program does not allow multiple
instances of itself to be run. This would cause havoc with the mill procedure if it
was done at the wrong time. The file input test will verify that any range of file
names can be accepted by the program. The file format test will check to see if a
wide range of Gerber files are transformed correctly into TIFF images. We will
collect a sample of Gerber files from various CAD software packages to test. The
TIFF images will need to be inspected manually. The PNG test will verify that a
PNG image that is read in is treated the same as a file that has been transformed
from a Gerber file to a TIFF file to a PNG file. The user access check will run all
commands available to the user on each user access level and verify that the
restricted users can not access the functionality of the higher access levels. The
93
user file test will test an assortment of various premade User files, and ensure
that the user can not place drawings to be milled in that location, and that the
machine can only access the locations specified in the file.
White box testing is a method that relies on knowledge of the internal workings of
the software being tested. This testing uses the same testing tools as the black
box testing, but will purposefully try to break the system with certain input values.
This causes there to be a smaller number of values that need to be tested, and
thus allows us to test faster. Furthermore, this test is more important than the
black box testing, as these are the cases that we need to find and either prohibit
users from accessing or warn the users of the possible consequences and are
the most likely to cause problems. The black box testing is only to find errors in
places that we did not expect to find them, such as a Java method mishandling a
string of a certain length while the white box testing is to verify that we don’t have
problems with something such as reading the first value, the last value, and not
the value after the last value of an array.
94
SECTION 14: Security
The main focus of the security aspect of this project is the support of the safety of
the user, and to a secondary extent the safety of the machine. We accomplished
this through the use of a multi tiered security paradigm.
The first level of this security scheme was to be a user access level that is read
off of the user’s identification card. Since the identification card was to be a
proprietary format that we were going to design for this project it would be difficult
for a user to spoof a higher access level than they have been assigned through
the use of standard card reader hardware. However, we have since decided to
use a user name and password combination for access to the system. The user
names will be students NID’s, and passwords will be of the format x####xx
where x is any letter and # is any single digit decimal number. The user
passwords will be hashed and stored in a file in the directory under the main
program. When a user enters their password, this password will be hashed and
their user name will be used as a key into a TreeMap. If the value of that location
in the TreeMap is returned as the same hash, the user is allowed access to the
system and is assigned the access level that is listed in the same file on the
same line. To protect this file, we will be implementing a system of dynamic
permissions, so that no user can simply open the file and change their access
level. Furthermore, we will be running the machine on a copy of Windows XP that
has been severely crippled to allow the user to only run our program. The
possible user access levels are Standard User, Advanced User, Experienced
User, and Administrator. Each increase in access level provides the user with
more privileges on the machine, as well as more responsibility. The logs of the
machine will contain which user was logged in at the time the event occurred.
These logs will be kept on the control computer until the log file folder reaches
100 MB, and then the oldest log files will begin to be deleted. Each mill
procedure or machine access will generate an individual log file. The name of the
log files will in the format YYMoMoDD_MiMiHHSS.log where D represents day,
Mo represents month, Y represents year, Mi represents minute, H represents
hour, and S represents seconds. The logs will be kept for such a long time so
that we can hopefully determine who was using the machine at the time that a
user preventable, or user caused, error occurred. If that user has an access level
that indicates that they should have had the ability to deal with the error, or
known better than to cause the error, they will either be demoted in access level
or permanently banned from further use of the machine depending on the
seriousness of the infraction.
The Standard User access level is what is assigned to certified Senior Design
students. This access level will allow the users to open the machine when it is
not running and attach their boards to the static portion of the XY table. These
users will only be allowed to mill FR-4 boards, and will not be allowed to add
additional board types to the machine. We will not be checking the type of the
95
board that the user enters, but the height of the XY table will need to be adjusted
for different board thicknesses. This is done to insure that the surface of the
board is at the focal point of the milling laser, and if a different board is added the
machine will still run with the FR-4 board height setting. This will most likely result
in an unusable board, or at best a board that requires multiple runs through the
machine. Additionally, Standard users will not be able to add other users and
must strictly follow all safety measures.
The next level after the Standard User access level is the Advanced user. These
users will not be senior design students, but rather other students that have gone
through the milling machine training sometime in the past. These users will have
needed to prove themselves as responsible Standard level users, and will almost
exclusively be past senior design students that are now working on their
graduate degrees. These users will have all of the privileges of the Standard user
access level as well as being allowed to modify unique sizes of boards and being
allowed to change the selection of different PCB materials. Additionally,
Advanced users will not be able to add other users and must strictly follow all
safety measures.
Experienced users will the users that are responsible for the day to day operation
and upkeep of the machine. These users will be ARC members that have agreed
to take care of the machine, as well as train new users. They will be able to add
and modify Standard and Advanced users, but will not be able to add or modify
permissions for other Experienced users. These users will also have the
permissions of the Standard and Advanced user access levels but will also be
able to add new types of boards to the machine, These users must follow all
safety procedures, but they will be able to suggest the laser output power when
adding new types of boards. An Administrator must verify and allow this
suggestion before it is available for an Experienced user to choose from. The
laser power will be held in check by the safety guidelines however, so these
users do not accidentally violate them.
The last, and highest, user access level is that of Administrator. This access level
will only be used by laser experienced service technicians. This is due to the fact
that users in this access level can ignore all safety measures. This could
potentially cause harm to the machine and to the user, so it will only be issued to
people that have demonstrated knowledge and experience of working with high
power laser systems and all of the risks that are involved in those activities.
Furthermore, all users at this access level will be required to have a liability
waver on file with the Amateur Radio Club. The only thing that a single
Administrator cannot do is run the machine with one of the covers, explained
later, open. This will require for two administrators to log in consecutively. This is
to ensure that at least one responsible user is nearby incase the Administrator
doing the work is incapacitated in anyway. This could be caused by the fumes
coming out of the machine or interaction with the laser itself.
96
The physical security of this project was accomplished through the use of the
external case of the machine. The external enclosure will be designed this way to
ensure that the enclosure can removed for easy access to the internals of the
machine. We initially planned to have a hierarchical security scheme. This was
going to be comprised of the external enclosure, along with enclosures for the
laser subsystem, the XY table, the power systems, and the microcontroller
section. The idea behind this scheme was to allow users to access portions of
the internals of the machine, but not others. We have precision optics in this
project, along with dangerous power systems, and if something needs to be
repaired we did not want that technician to mess with other sections of the
project. This plan fell through due to the complexity of securing the boxes inside
the main box, and the fact that we decided that this project will stay a student
endeavor. The only people that will be working on it will be other students that we
trust to not mess with something that they have not been properly trained on. The
compromise that we have come upon is that there is only one main box.
97
SECTION 15: Personnel
15.1 – Nathan Bodnar
Nathan Bodnar is an undergraduate student working on his B.S. in Electrical
Engineering at the University of Central Florida. In 2008, Nathan joined Dr.
Richardson’s Laser Plasma Laboratories EUV group, there he works as an
undergraduate researcher. In the summer of 2009, he took part in the
International REU program with the main emphasis in droplet formation. Current
projects that he is currently working on includes a time resolving FTIR
spectrometer, building a mass spectrometer, designing motor controllers for the
droplet delivery system, and testing new compounds that generate EUV
emissions. Nathan’s spare time is spent working on projects ranging form
building high-powered lasers to plant tissue culture.
15.2 – David Dowdle
David Dowdle is a senior working on his BS in Computer Engineering. He is
currently employed with the biology department working on amplifiers for
electrostatic disturbances generated by amphibious animals. David previously
worked under Dr. Chew as an undergrad researcher for packaging of the 70 cm
transmitters for high altitude experiments. During his leisure time he diligently
works with the Amateur Radio Club at the University of Central Florida. Here he
works on projects and helps maintain the environment of the senior design lab.
15.3 – Ryan Maticka
Ryan Maticka is an undergraduate Electrical Engineering student at the
University of Central Florida and expects to complete his courses for Electrical
Engineering in the spring of 2010. He also plans to obtain a major in Computer
Science and minor in math after the spring of 2010. After graduation, he would
like to work for several years and then return to school to complete a master’s
degree.
98
SECTION 16: Bibliography
[1] Richards. J. R. "Control of Particulate Emissions," APTI, 1995.
[2] D. J. Meier, S. H. Schmidt, and LPKF Laser, “PCB Laser Technology for Rigid
and Flex HDI – Via Formation, Structuring, Routing,” 2002. [Online].
Available: http://www.lpkfusa.com/articles/hdi/pcb_laser_technology.pdf.
[Accessed: Dec. 13, 2009].
[3] B. Kelly, C. Hallman, D. Arnold, and R. Weiss, "Project ANGEL,"
Undergraduate Senior Design Project, University of Central Florida,
Orlando, 2007.
[4] Red Optics, “KD*P(DKDP) Crystal (Potassium Dihydrogen Phosphate and
Potassium Dideuterium Phosphate),” 2007. [Online]. Available:
http://www.redoptronics.com/KDP-crystal.html. [Accessed: Dec. 13, 2009].
[5] Red Optics, “LBO Crystal (Lithium Triborate),” 2007. [Online]. Available:
http://www.redoptronics.com/LBO-crystal.html. [Accessed: Dec. 13, 2009].
[6] Red Optics, “BBO Crystal (Beta Barium Borate),” 2007. [Online]. Available:
http://www.redoptronics.com/BBO-crystal.html. [Accessed: Dec. 13, 2009].
[7] Red Optics, “KTP Crystal (KTiOPO4),” 2007. [Online]. Available:
http://www.redoptronics.com/KTP-crystal.html. [Accessed: Dec. 13, 2009].
[8] Red Optics, “Anti-Gray Tracking KTP Crystal,” 2007. [Online]. Available:
http://www.redoptronics.com/Anti-Gray-Tracking-KTP-crystal.html.
[Accessed: Dec. 13, 2009].
[9] Batarseh, Issa, “Power Electronic Circuits,” Hoboken: John Wiley & Sons, Inc,
2004.
[10] Maxim Integrated Products. “DC-DC Converter Tutorial – Maxim,” Nov. 29,
2001. [Online]. Available: http://www.maximic.com/appnotes.cfm/an_pk/2031. [Accessed: Dec. 5, 2009].
[11] “Thermoelectric Peltier Device Information (TE Heater / Cooler / Generator
Modules) – General Info,” 2009. [Online]. Available: http://www.peltierinfo.com/info.html. [Accessed: Dec. 12, 2009].
[12] F. McClung and R. Hellwarth, "Giant Optical Pulsations from Ruby," Journal
of Applied Physics, 1962.
99
[13] “The Variable High-Voltage Power Supply 0-3kV,” 2008. [Online]. Available:
http://www.elecfree.com/electronic/the-variable-high-voltage-powersupply-0-3kv. [Accessed: Dec. 13, 2009].
[14] ON Semiconductor, “2N3055(NPN), MJ2955(PNP),” 2005. [Online].
Available: http://www.onsemi.com/pub_link/Collateral/2N3055-D.pdf.
[Accessed: Dec. 13, 2009].
[15] Microchip, “MCP413X/415X/423X/425X, 7/8-Bit Single/Dual SPI Digital POT
with Volatile Memory,” 2007. [Online]. Available:
http://ww1.microchip.com/downloads/en/DeviceDoc/22060a.pdf.
[Accessed: Dec. 13, 2009].
[16] OMRON, “PCB Relay G6RL,” 2008. [Online]. Available:
http://www.omron.com/ecb/products/pdf/en-g6rl.pdf. [Accessed: Dec. 13,
2009].
[17] Microchip, “AN822 Stepper Motor Microstepping with PIC18C452,” 2002.
[Online]. Available:
http://ww1.microchip.com/downloads/en/AppNotes/00822a.pdf.
[Accessed: Dec. 13, 2009].
[18] Microchip, “MPLAB C18 C Compiler Libraries,” 2004. [Online]. Available:
http://ww1.microchip.com/downloads/en/devicedoc/51297c.pdf.
[Accessed: Dec. 12, 2009].
[19] Microchip, “PIC18F2550,” Oct. 27, 2009. [Online]. Available:
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010
280. [Accessed: Dec. 12, 2009].
[20] Microchip, “PIC18F4550,” Oct. 27, 2009. [Online]. Available:
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010
300. [Accessed: Dec. 12, 2009].
[21] “C8051F342-GQ Silicon Labs Microcontrollers,” 2009. [Online]. Available:
http://mouser.com/ProductDetail/Silicon-Labs/C8051F342GQ/?qs=sGAEpiMZZMvu0Nwh4cA1wWtRoPcQBaz3BDp7SQlE0jo%3d.
[Accessed: Dec. 12, 2009].
[22] “ATMEGA162-16PU Atmel Microcontrollers,” 2009. [Online]. Available:
http://mouser.com/ProductDetail/Atmel/ATMEGA16216PU/?qs=I3kMT7EEIOUre5FsjF6g7A%3d%3d. [Accessed: Dec. 12,
2009].
100
[23] “MC9S08JS8CWJ Freescale Semiconductor Microcontrollers,” 2009.
[Online]. Available: http://mouser.com/ProductDetail/FreescaleSemiconductor/MC9S08JS8CWJ/?qs=t6o%252baWFra1eNwN1CL2tWgw
%3d%3d. [Accessed: Dec. 12, 2009].
[24] “USB overview and Plug & Receptacle pinouts,” Jul. 19, 2009. [Online].
Available: http://www.accesscomms.com.au/reference/usb.htm.
[Accessed: Dec. 12, 2009].
[25] “IGBT or MOSFET: Choose Wisely,” Oct. 18, 2000. [Online]. Available:
http://www.irf.com/technical-info/whitepaper/choosewisely.pdf. [Accessed:
Apr. 12, 2010].
101
SECTION 17: Facilities and Consultants
Throughout the project the Laser Plasma Laboratories team in the CREOL
department helped in generating the burn test required for this project. The staff
also helped in simulating the laser cavity and gave thoughts on how to improve
the design of the Nd:YAG laser cavity. For the testing of the laser diodes and the
calibration of the wavelength of the outputs from the diode will also be done with
the LPL team. When first testing the laser diode the laser diode driver will be
supplied as well from the CREOL department. This would allow a high accuracy
for the outputs of the laser than if the laser power supply that this project is
building were to be used. The output power will also be measured with a high
powered light meter that will be barrowed through the LPL team.
102
SECTION 19: Budget and Financing
This laser milling machine will primarily be funded by Nathan Bodnar. Some of
the supplies may be donated from the Laser Plasma Laboratories at CREOL.
The group will also attempt to obtain donations from some of the major optics
supply companies, such as Thorlabs and Melles Griot. Some of the parts that
cannot be acquired through donations may be obtained from eBay, which will
greatly reduce the cost of the parts.
For all other parts that cannot be obtained through the above sources they will be
purchased from optics companies. The following list outlines the general cost of
the parts required for this project.
•
•
•
•
•
•
•
•
•
Software – free
Parts for XYZ table – $200
Laser setup
– Q-switch – $512.95
– 808nm Diodes – $486.99
– Nd:YAG rod – $125
– KTP(KD*P) – $43.22
– Directing mirrors – $27.19
– Lens – $420
– Quarter wave plate – $74
– Polarizer – $49
Parts for Laser Power Supply – $200
Parts for TEC Power Supply – $35
Parts for Q switch Power Supply – $140
Parts for Stepper Power Supply – $30
Parts for Power Management Circuit – $50
Fume controller – $30
Total: $2423.35
103
SECTION 20: Milestone
The following is the milestone of the Laser PCB Milling Machine project:
Research
85%
Bought Materials
Laser Cavity
Laser Power Supply
Q Switch Power Supply
70%
12%
15%
20%
TEC Controller
90%
XY Table
80%
Software
60%
Total
54%
0%
20%
40%
104
60%
80%
100%
SECTION 21: Project Summary
The goal of this project is to replace the current milling machine used by senior
design students. This machine has a history of being bereft of problems, causing
it to almost continually be in a state of disrepair awaiting funding or a particular
component. We aim to eliminate the problems that the machine has suffered in
the past by radically changing the method that is used in the mill procedure.
While the current Quick Circuit 5000 uses a spinning drill bit to remove copper
from copper clad boards, we will use a Nd:YAG laser. While there are health and
safety issues with the light from the laser and the fumes from the copper, we
believe that we can significantly improve the overall safety of the senior design
students, while improving the uptime of this resource. Our solution will achieve
this through the use of strict user access control, both software and physical, and
through the use of an intuitive graphical user interface.
Because we are designing this project to be used in an enclosed environment, in
a building that is not designed to house high powered lasers, we have had to
make some allowances in our design. The most important of these is the air
scrubbing system. This is an absolute must, as this machine will be running in an
enclosed room in close proximity to students. Furthermore, the air scrubbing
system removes the vaporized copper that is being made by the laser. If this is
not removed as soon as it comes off of the board, we would very quickly have
the inside of our enclosure coated in a layer of copper. We wish to make our
project self sufficient, and because the location where this machine will be
located has no high powered laser power supplies we have had to design our
own supplies that we will incorporate into our main project box. Lastly, there is no
chilled building water supply in the Engineering 1 building. We have therefore
had to design our own Thermo Electric Cooling system to provide cooling to
proper cooling for the laser diodes.
We have design a large amount of software based feedback control into this
project. The reason for this is twofold, safety and mean time to failure. It is an
absolute necessity that we provide a safe environment for users to work in.
Therefore, we have designed twenty four sensors into the project that will
continuously feed information back to a graphical user interface. Errors happen,
no matter how well a system is designed. Because of this, we have implemented
a detailed logging system any time there is a system error. These errors may or
may not shut down the machine, and may or may not force the machine to stay
in an off state until the problem can be resolved. This information will help
administrators decide on the best course of action on any errors that might occur.
A safe environment for users to work in is also a safe environment for the
machine to operate in. We are dealing with very expensive equipment, and we
wish for it to last as long as possible. Through the use of judicious safety
practices, we will also accomplish our second goal of reducing the mean time to
failure of the machine.
105
SECTION 22: Project Operation
1) Plug power supply into AC wall outlet.
2) Plug both USB cables into computer.
3) Launch the software by double clicking on the executable Jar file.
Figure 1: executable Jar file
4) Type in the username and password.
Figure 2: Login Screen
5) Click on Load File
Figure 3: Click on Load File
6) Click on either Browse Gerber or Browse png
106
Figure 4: Load Image
7) Click on next.
8) Place Circuit Board in the Laser Milling Machine table positioning the board at
the top left hand corner.
9) Click on Mill.
Figure 5: Click on the Mill Button
10) Wait a very long time.
Time R equired ≈ Size X (mils ) ⋅ SizeY (mils ) ⋅1sec
Equation 1: Timing estimate
Ex: 4”x6” board will take: 0.76 years!
11) Remove the now finished board.
12) Close the software.
13) Turn off the Laser Power Supply.
14) Unplug the USB cables.
15) Unplug the Power supply
16) Go home!
107
APPENDIX A: Copyright Notices
A.1 – jPicUsb License
Permission pending
A.2 – gerb2tiff License
GNU GENERAL PUBLIC LICENSE
Version 2, June 1991
Copyright (C) 1989, 1991 Free Software Foundation, Inc.
59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
Preamble
The licenses for most software are designed to take away your
freedom to share and change it. By contrast, the GNU General Public
License is intended to guarantee your freedom to share and change free
software--to make sure the software is free for all its users. This
General Public License applies to most of the Free Software
108
Foundation's software and to any other program whose authors commit to
using it. (Some other Free Software Foundation software is covered by
the GNU Library General Public License instead.) You can apply it to
your programs, too.
When we speak of free software, we are referring to freedom, not
price. Our General Public Licenses are designed to make sure that you
have the freedom to distribute copies of free software (and charge for
this service if you wish), that you receive source code or can get it
if you want it, that you can change the software or use pieces of it
in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid
anyone to deny you these rights or to ask you to surrender the rights.
These restrictions translate to certain responsibilities for you if you
distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether
gratis or for a fee, you must give the recipients all the rights that
you have. You must make sure that they, too, receive or can get the
source code. And you must show them these terms so they know their
rights.
We protect your rights with two steps: (1) copyright the software, and
(2) offer you this license which gives you legal permission to copy,
distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain
that everyone understands that there is no warranty for this free
software. If the software is modified by someone else and passed on, we
want its recipients to know that what they have is not the original, so
that any problems introduced by others will not reflect on the original
authors' reputations.
Finally, any free program is threatened constantly by software
patents. We wish to avoid the danger that redistributors of a free
program will individually obtain patent licenses, in effect making the
program proprietary. To prevent this, we have made it clear that any
patent must be licensed for everyone's free use or not licensed at all.
The precise terms and conditions for copying, distribution and
modification follow.
109
GNU GENERAL PUBLIC LICENSE
TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION
MODIFICATION
0. This License applies to any program or other work which contains
a notice placed by the copyright holder saying it may be distributed
under the terms of this General Public License. The "Program", below,
refers to any such program or work, and a "work based on the Program"
means either the Program or any derivative work under copyright law:
that is to say, a work containing the Program or a portion of it,
either verbatim or with modifications and/or translated into another
language. (Hereinafter, translation is included without limitation in
the term "modification".) Each licensee is addressed as "you".
Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope. The act of
running the Program is not restricted, and the output from the Program
is covered only if its contents constitute a work based on the
Program (independent of having been made by running the Program).
Whether that is true depends on what the Program does.
1. You may copy and distribute verbatim copies of the Program's
source code as you receive it, in any medium, provided that you
conspicuously and appropriately publish on each copy an appropriate
copyright notice and disclaimer of warranty; keep intact all the
notices that refer to this License and to the absence of any warranty;
and give any other recipients of the Program a copy of this License
along with the Program.
You may charge a fee for the physical act of transferring a copy, and
you may at your option offer warranty protection in exchange for a fee.
2. You may modify your copy or copies of the Program or any portion
of it, thus forming a work based on the Program, and copy and
distribute such modifications or work under the terms of Section 1
above, provided that you also meet all of these conditions:
a) You must cause the modified files to carry prominent notices
stating that you changed the files and the date of any change.
b) You must cause any work that you distribute or publish, that in
whole or in part contains or is derived from the Program or any
part thereof, to be licensed as a whole at no charge to all third
parties under the terms of this License.
c) If the modified program normally reads commands interactively
when run, you must cause it, when started running for such
interactive use in the most ordinary way, to print or display an
announcement including an appropriate copyright notice and a
notice that there is no warranty (or else, saying that you provide
a warranty) and that users may redistribute the program under
these conditions, and telling the user how to view a copy of this
License. (Exception: if the Program itself is interactive but
110
AND
does not normally print such an announcement, your work based on
the Program is not required to print an announcement.)
These requirements apply to the modified work as a whole. If
identifiable sections of that work are not derived from the Program,
and can be reasonably considered independent and separate works in
themselves, then this License, and its terms, do not apply to those
sections when you distribute them as separate works. But when you
distribute the same sections as part of a whole which is a work based
on the Program, the distribution of the whole must be on the terms of
this License, whose permissions for other licensees extend to the
entire whole, and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest
your rights to work written entirely by you; rather, the intent is to
exercise the right to control the distribution of derivative or
collective works based on the Program.
In addition, mere aggregation of another work not based on the Program
with the Program (or with a work based on the Program) on a volume of
a storage or distribution medium does not bring the other work under
the scope of this License.
3. You may copy and distribute the Program (or a work based on it,
under Section 2) in object code or executable form under the terms of
Sections 1 and 2 above provided that you also do one of the following:
a) Accompany it with the complete corresponding machine-readable
source code, which must be distributed under the terms of Sections
1 and 2 above on a medium customarily used for software interchange; or,
b) Accompany it with a written offer, valid for at least three
years, to give any third party, for a charge no more than your
cost of physically performing source distribution, a complete
machine-readable copy of the corresponding source code, to be
distributed under the terms of Sections 1 and 2 above on a medium
customarily used for software interchange; or,
c) Accompany it with the information you received as to the offer
to distribute corresponding source code. (This alternative is
allowed only for noncommercial distribution and only if you
received the program in object code or executable form with such
an offer, in accord with Subsection b above.)
The source code for a work means the preferred form of the work for
making modifications to it. For an executable work, complete source
code means all the source code for all modules it contains, plus any
associated interface definition files, plus the scripts used to
control compilation and installation of the executable. However, as a
special exception, the source code distributed need not include
anything that is normally distributed (in either source or binary
form) with the major components (compiler, kernel, and so on) of the
operating system on which the executable runs, unless that component
itself accompanies the executable.
111
If distribution of executable or object code is made by offering
access to copy from a designated place, then offering equivalent
access to copy the source code from the same place counts as
distribution of the source code, even though third parties are not
compelled to copy the source along with the object code.
4. You may not copy, modify, sublicense, or distribute the Program
except as expressly provided under this License. Any attempt
otherwise to copy, modify, sublicense or distribute the Program is
void, and will automatically terminate your rights under this License.
However, parties who have received copies, or rights, from you under
this License will not have their licenses terminated so long as such
parties remain in full compliance.
5. You are not required to accept this License, since you have not
signed it. However, nothing else grants you permission to modify or
distribute the Program or its derivative works. These actions are
prohibited by law if you do not accept this License. Therefore, by
modifying or distributing the Program (or any work based on the
Program), you indicate your acceptance of this License to do so, and
all its terms and conditions for copying, distributing or modifying
the Program or works based on it.
6. Each time you redistribute the Program (or any work based on the
Program), the recipient automatically receives a license from the
original licensor to copy, distribute or modify the Program subject to
these terms and conditions. You may not impose any further
restrictions on the recipients' exercise of the rights granted herein.
You are not responsible for enforcing compliance by third parties to
this License.
7. If, as a consequence of a court judgment or allegation of patent
infringement or for any other reason (not limited to patent issues),
conditions are imposed on you (whether by court order, agreement or
otherwise) that contradict the conditions of this License, they do not
excuse you from the conditions of this License. If you cannot
distribute so as to satisfy simultaneously your obligations under this
License and any other pertinent obligations, then as a consequence you
may not distribute the Program at all. For example, if a patent
license would not permit royalty-free redistribution of the Program by
all those who receive copies directly or indirectly through you, then
the only way you could satisfy both it and this License would be to
refrain entirely from distribution of the Program.
If any portion of this section is held invalid or unenforceable under
any particular circumstance, the balance of the section is intended to
apply and the section as a whole is intended to apply in other
circumstances.
It is not the purpose of this section to induce you to infringe any
patents or other property right claims or to contest validity of any
such claims; this section has the sole purpose of protecting the
112
integrity of the free software distribution system, which is
implemented by public license practices. Many people have made
generous contributions to the wide range of software distributed
through that system in reliance on consistent application of that
system; it is up to the author/donor to decide if he or she is willing
to distribute software through any other system and a licensee cannot
impose that choice.
This section is intended to make thoroughly clear what is believed to
be a consequence of the rest of this License.
8. If the distribution and/or use of the Program is restricted in
certain countries either by patents or by copyrighted interfaces, the
original copyright holder who places the Program under this License
may add an explicit geographical distribution limitation excluding
those countries, so that distribution is permitted only in or among
countries not thus excluded. In such case, this License incorporates
the limitation as if written in the body of this License.
9. The Free Software Foundation may publish revised and/or new versions
of the General Public License from time to time. Such new versions will
be similar in spirit to the present version, but may differ in detail to
address new problems or concerns.
Each version is given a distinguishing version number. If the Program
specifies a version number of this License which applies to it and "any
later version", you have the option of following the terms and conditions
either of that version or of any later version published by the Free
Software Foundation. If the Program does not specify a version number of
this License, you may choose any version ever published by the Free Software
Foundation.
10. If you wish to incorporate parts of the Program into other free
programs whose distribution conditions are different, write to the author
to ask for permission. For software which is copyrighted by the Free
Software Foundation, write to the Free Software Foundation; we sometimes
make exceptions for this. Our decision will be guided by the two goals
of preserving the free status of all derivatives of our free software and
of promoting the sharing and reuse of software generally.
NO WARRANTY
11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS
NO WARRANTY
FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.
EXCEPT WHEN
OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR
OTHER PARTIES
PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND,
EITHER EXPRESSED
OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF
113
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
ENTIRE RISK AS
TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.
SHOULD THE
PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL
NECESSARY SERVICING,
REPAIR OR CORRECTION.
12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED
TO IN WRITING
WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY
MODIFY AND/OR
REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO
YOU FOR DAMAGES,
INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL
DAMAGES ARISING
OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT
NOT LIMITED
TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES
SUSTAINED BY
YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE
WITH ANY OTHER
PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN
ADVISED OF THE
POSSIBILITY OF SUCH DAMAGES.
END OF TERMS AND CONDITIONS
This General Public License does not permit incorporating your program into
proprietary programs. If your program is a subroutine library, you may
consider it more useful to permit linking proprietary applications with the
library. If this is what you want to do, use the GNU Library General
Public License instead of this License.
$Id: GPL,v 2.0 1998/12/12 08:05:30 roessler Rel $
A.3 – ImageMagick License
Licensed under the ImageMagick License (the "License"); you may not use
this file except in compliance with the License. You may obtain a copy
of the License at
http://www.imagemagick.org/script/license.php
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See
the License for the specific language governing permissions and limitations
under the License.
114
Figure 9.2: Copyright Notice
Figure 5.6: Copyright Notice
115
APPENDIX B: Example DRL File
M48
;DRILL file {PCBNEW (2007-07-09)} date 26/3/2009-21:21:07
;FORMAT={2:4 / absolute / inches / decimal format}
R,T
VER,1
FMAT,2
INCH,TZ
TCST,OFF
ICI,OFF
ATC,ON
T1C0.025
T2C0.032
T3C0.040
%
M47
G05
M72
T1
X4.775Y-3.625
X4.425Y-3.325
M30
116
APPENDIX C: Password Generation Code
import java.lang.Math;
public class passwordGenerator {
public static void main(String[] args) {
String alphabet="abcdefghijklmnopqrstuvwxyz";
for (int i=0; i<100; i++) {
int ch=(int)(Math.random()*26);
String partialPass1=alphabet.substring(ch, ch+1);
ch=(int)(Math.random()*26);
String partialPass2=alphabet.substring(ch, ch+1);
ch=(int)(Math.random()*26);
partialPass2 += alphabet.substring(ch, ch+1);
int num1 = (int)(Math.random()*9);
int num2 = (int)(Math.random()*9);
int num3 = (int)(Math.random()*9);
int num4 = (int)(Math.random()*9);
String numString = num1+""+num2+""+num3+""+num4;
System.out.println(partialPass1 + numString + partialPass2);
}
}
}
117
APPENDIX D: jPicUsb Documentation
118
119