Ubuntu Webcam Server C Language Rules and Assignments USB

Transcription

Why Specifications Matter, p. 14 | X10 Controller Design, p. 28 | Hexapod Robot Development, p. 44
THE
MAGAZINE
FOR
COMPUTER
A P P L I C AT I O N S
July 2010
Issue 240
Ubuntu Webcam Server
C Language Rules
and Assignments
PLUS
The Professional Method:
Research, Design, Write,
Then Repeat
USB and Standard
Device Descriptors
A Look at Powerline
Communication Chips
Wireless Data
Transmission
$7.50 U.S. ($8.50 Canada)
www.circuitcellar.com
Personal Download for [email protected] I © Circuit Cellar
SSH Encrypted
SERIAL TO ETHERNET SOLUTIONS
Instantly network-enable
any serial device
Works out of the box no programming is required
Device P/N: SB70LC-100CR
Kit P/N: NNDK-SB70LC-KIT
$47
Qty. 1000
Customize to suit any application
with low-cost development kit
SB70LC
256-bit encryption protects data
from unauthorized monitoring
2-port serial-to-Ethernet server
Features:
10/100 Ethernet
TCP/UDP/SSH/SSL modes
DHCP/Static IP Support
Data rates up to 921.6kbps
Web-based configuration
Device P/N: SB700-EX-100CR
Kit P/N: NNDK-SB700EX-KIT
SB700EX
with RS-232 & RS-485/422 support
$129
Qty. 1000
Need a custom solution?
NetBurner Serial to Ethernet
Development Kits are available to
customize any aspect of operation
including web pages, data filtering, or
custom network applications. All kits
include platform hardware, ANSI C/C++
compiler, TCP/IP stack, web server, email protocols, RTOS, flash file system,
Eclipse IDE, debugger, cables and power
supply. The NetBurner Security Suite
option includes SSH v1 & v2 support.
Device P/N: CB34-EX-100IR
Kit P/N: NNDK-CB34EX-KIT
$149
Qty. 1000
CB34EX
industrial temperature grade
Information and Sales | [email protected]
Web | www.netburner.com
Telephone | 1-800-695-6828
with RS-232 & RS-485/422 support
and terminal block connector
T
ASK
MANAGER
July 2010 – Issue 240
I
4
THE MAGAZINE FOR COMPUTER APPLICATIONS
Circuit Cellar: The Next Iteration
FOUNDER/EDITORIAL DIRECTOR
Steve Ciarcia
PUBLISHER
Hugo Vanhaecke
EDITOR-IN-CHIEF
C. J. Abate
have noted here before that Circuit Cellar’s on-going mission is to educate professional engineers and academics so they
may design the next generation of embedded applications and
systems. This month marks an important step forward in our
endeavor to provide exciting new content and need-to-know
information. As you’ll see, we’re using all of our resources to
deliver readers a new monthly column, an informative interview section, easier access to essential content, and more.
Let’s move in chronological order.
I don’t usually reference ads here, but check out page 7 for an
example of a new product we developed after reviewing feedback from readers. It’s an 8-GB waterproof flash drive with every
issue of Circuit Cellar and all the associated code. We call the
flash drive “CC Gold.” It’s a great example of delivering today’s
engineers what they require—in this case, a portable storage
drive holding literally “everything Circuit Cellar” in one small
package. And with gigs of space to spare, it’s a great place to
store future issues and personal project files.
Turn to page 8 to read our new “Questions & Answers” section. There you’ll find my recent interview with Jan Axelson,
author of the often-referenced USB Complete and seven
Circuit Cellar articles. After meeting Jan at the 2010 Embedded
Systems Conference in San Jose, CA, it occurred to me that she
would be the perfect first interviewee for this new section,
which is dedicated to giving readers insight into how talented
professionals achieved their success, cultivated their skills, and
contributed to the computer applications industry. In future
issues, we’ll present interviews with other notable engineers.
After learning about Jan’s background and considering her
ideas regarding the future of USB technology, check out George
Novacek’s new column titled “The Consummate Engineer” (p.
14). The purpose of the column is to provide readers with tips on
becoming more well-rounded practitioners of their trade. Sure,
all of our readers are soldering and coding pros. But what separates a workaday engineer from a go-to guru is an understanding
of “robust design.” In this column, George will touch on topics
such as proper spec development, system hazard analysis, reliability analysis, and more. With a background in engineering
and management, George is an excellent resource for readers
looking to take their designs and careers to the next level.
The rest of the issue touches on some of the most intriguing
subjects in the field of computer applications. Examples
include server development (p. 16), mobile robotics (p. 44), and
data transmission (Bluetooth on p. 36 and powerline communication on p. 66).
Lastly, keep in mind that we are constantly updating and
upgrading our digital magazine and website in an effort to meet
our readers’ rapidly changing needs. Be assured: if you want it,
we’ll deliver. Stay tuned.
[email protected]
ASSOCIATE PUBLISHER
Shannon Barraclough
WEST COAST EDITOR
Tom Cantrell
CUSTOMER SERVICE
Debbie Lavoie
CONTRIBUTING EDITORS
Jeff Bachiochi
Robert Lacoste
George Martin
Ed Nisley
CONTROLLER
Jeff Yanco
ART DIRECTOR
KC Prescott
GRAPHIC DESIGNERS
Grace Chen
Carey Penney
NEW PRODUCTS EDITOR
John Gorsky
PROJECT EDITORS
Ken Davidson
David Tweed
STAFF ENGINEER
John Gorsky
ADVERTISING
800.454.3741 • 978.281.7708 • www.circuitcellar.com/advertise
ADVERTISING REPRESENTATIVE
Peter Wostrel
Strategic Media Marketing, Inc.
1187 Washington St., Gloucester, MA 01930 USA
800.454.3741 • 978.281.7708
[email protected] • www.smmarketing.us
Fax: 978.281.7706
ADVERTISING COORDINATOR
Valerie Luster
E-mail: [email protected]
Cover photography by Chris Rakoczy—Rakoczy Photography
www.rakoczyphoto.com
PRINTED IN THE UNITED STATES
CONTACTS
SUBSCRIPTIONS
Information: www.circuitcellar.com/subscribe, E-mail: [email protected]
Subscribe: 800.269.6301, www.circuitcellar.com/subscribe, Circuit Cellar Subscriptions, P.O. Box 5650,
Hanover, NH 03755-5650
Address Changes/Problems: E-mail: [email protected]
GENERAL INFORMATION
860.875.2199, Fax: 860.871.0411, E-mail: [email protected]
Editorial Office: Editor, Circuit Cellar, 4 Park St., Vernon, CT 06066, E-mail: [email protected]
New Products: New Products, Circuit Cellar, 4 Park St., Vernon, CT 06066, E-mail: [email protected]
AUTHORIZED REPRINTS INFORMATION
860.875.2199, E-mail: [email protected]
AUTHORS
Authors’ e-mail addresses (when available) are included at the end of each article.
CIRCUIT CELLAR®, THE MAGAZINE FOR COMPUTER APPLICATIONS (ISSN 1528-0608) is published monthly by Circuit Cellar
Incorporated, 4 Park Street, Vernon, CT 06066. Periodical rates paid at Vernon, CT and additional offices. One-year (12 issues)
subscription rate USA and possessions $45, Canada/Mexico $60, all other countries $63. Two-year (24 issues) subscription
rate USA and possessions $80, Canada/Mexico $110, all other countries $116. All subscription orders payable in U.S. funds
only via Visa, MasterCard, international postal money order, or check drawn on U.S. bank. Direct subscription orders and subscription-related questions to Circuit Cellar Subscriptions, P.O. Box 5650, Hanover, NH 03755-5650 or call 800.269.6301.
Postmaster: Send address changes to Circuit Cellar, Circulation Dept., P.O. Box 5650, Hanover, NH 03755-5650.
Circuit Cellar® makes no warranties and assumes no responsibility or liability of any kind for errors in these programs or schematics or for the
consequences of any such errors. Furthermore, because of possible variation in the quality and condition of materials and workmanship of reader-assembled projects, Circuit Cellar® disclaims any responsibility for the safe and proper function of reader-assembled projects based upon or
from plans, descriptions, or information published by Circuit Cellar®.
The information provided by Circuit Cellar® is for educational purposes. Circuit Cellar® makes no claims or warrants that readers have a right to
build things based upon these ideas under patent or other relevant intellectual property law in their jurisdiction, or that readers have a right to
construct or operate any of the devices described herein under the relevant patent or other intellectual property law of the reader’s jurisdiction.
The reader assumes any risk of infringement liability for constructing or operating such devices.
Entire contents copyright © 2010 by Circuit Cellar, Incorporated. All rights reserved. Circuit Cellar is a registered trademark of Circuit Cellar, Inc.
Reproduction of this publication in whole or in part without written consent from Circuit Cellar Inc. is prohibited.
CIRCUIT CELLAR®
•
Let your geek shine.
Meet Leah Buechley, developer of LilyPad—a
sew-able microcontroller—and fellow geek. Leah
used SparkFun products and services while she
developed her LilyPad prototype.
The tools are out there, from LEDs to conductive
thread, tutorials to affordable PCB fabrication,
and of course Leah’s LilyPad. Find the resources
you need to let your geek shine too.
»Sharing Ingenuity
SP A R K F U N.C OM
©2010 SparkFun Electronics, Inc. All rights reserved.
240
INSIDE ISSUE
July 2010
16
28
36
44
•
Internet & Connectivity
Denny Lin
Webcam Server,
p. 16
Build an X10 Controller (Part 1)
Circuit Design
Devlin Gualtieri
Wireless Data Exchange
Build a 2,700-lb. Bluetooth Headset
Tom Struzik
Hexapod Kinematics
Motor Communication, Motion Planning,
and Image Processing
Matt Bunting & John Rogers
X10 Control,
p. 28
Hexapod Robot,
p. 44
14
54
58
6
66
THE CONSUMMATE ENGINEER
It All Begins With a Spec
George Novacek
LESSONS FROM THE TRENCHES
Put C Language to the Test (Part 3)
Rules and Assignments
George Martin
FROM THE BENCH
Application Communication with USB (Part 2)
The Importance of Descriptors
Jeff Bachiochi
SILICON UPDATE
Wall Whisperer
A Fresh Take on Powerline Communication
Tom Cantrell
TASK MANAGER
Circuit Cellar: The Next Iteration
C. J. Abate
QUESTIONS & ANSWERS
Research, Design, Write,
Then Repeat
An Interview With Jan Axelson
C. J. Abate
4
8
NEW PRODUCT NEWS
10
TEST YOUR EQ
13
CROSSWORD
74
INDEX OF ADVERTISERS
August Preview
PRIORITY INTERRUPT
72 Flavors and More
Steve Ciarcia
CIRCUIT CELLAR®
•
79
80
QUESTIONS & ANSWERS
Research, Design, Write, Then Repeat
An Interview With Jan Axelson
Jan Axelson is the author of USB Complete (Lakeview Research, Fourth Edition, 2009), Serial Port
Complete (Lakeview Research, Second Edition, 2007), and many other books and articles about
hardware interfacing. Between 1997 and 2008, she wrote seven articles for Circuit Cellar. I met
Jan Axelson at the April 2010 Embedded Systems Conference in San Jose, California. In a May
2010 interview, I asked Jan about topics ranging from her earliest design experiences to the future
of USB technology. — C. J. Abate, Editor-in-Chief
CJ: Tell us about yourself. What’s your current occupation?
JAN: I live in Wisconsin. When I’m not doing computer projects, I like to spend time outdoors. My favorite gadgets for
outdoors are a GPS for navigating and an iPod with BirdJam
software for identifying bird songs. I’m working on the
design of a hand-held data logger that records the date, time,
and GPS coordinates of sightings of target birds.
With my books, articles, and website, I help developers
use USB and other computer interfaces in their projects.
CJ: How long have you been reading Circuit Cellar?
JAN: I’ve subscribed to Circuit Cellar from the beginning.
CJ: What was the first technology that really caught your
attention and led you into the field of engineering?
JAN: I’ve long been interested in what’s outside the box—how
PCs and embedded systems can talk to the world outside
themselves. Early on, the options were parallel and serial
ports, including RS-485 networks. From there, I expanded
my horizons to USB and Ethernet.
8
JAN: I have a degree in electronics technology. What led me
to writing was my frustration in trying to find useful information about the components and circuits I was interested
in. To write about something, you have to understand it
inside out, so writing is a good excuse to dig deep.
CJ: Before writing books, what was your profession?
JAN: In one job, I installed and maintained long-distance
telephone circuits. Each channel extended anywhere from
across town to across the state. Troubleshooting was a group
effort, with one technician on each end and another in the
middle. I also worked as a test technician for a manufacturer
of computer-controlled scientific equipment. That job was
my introduction to serial and parallel ports.
CJ: You wrote seven articles for Circuit Cellar between
1997 and 2008. The main topics were connection (USB and
parallel/serial ports) and memory (EEPROMs). What interests you about these two subjects?
CJ: How long have you been working with electronics and
“embedded technologies”?
JAN: I like to work on projects where a computer monitors
and controls the world outside itself, and to do that, you
need an interface to provide the link. For me, EEPROMs are
just a reason to explore synchronous serial interfaces such as
SPI and I2C.
JAN: I have over 30 years experience with electronics and
computers.
CJ: Give us a quick review. What are the main differences
between USB from the late 1990s and USB of today?
CJ: What was your first MCU-based design?
JAN: The chips, speeds, and tools have all changed dramatically—in good ways—over the last decade. The number of
silicon providers for USB host and device controllers has
grown from a few to several dozen. You can find everything
from bare-bones interface chips to intelligent controllers
that handle many of the higher-level USB protocols for you.
With the release of USB 3.0 in 2008, USB offers four bus
speeds, though host computers are just beginning to support
USB 3.0. Drives and video devices have been the first to
JAN: My first microcontroller project used an Intel 8052
microcontroller with a BASIC interpreter in ROM. An
alphanumeric LCD displayed the temperature, maximum
and minimum temperatures, and time.
CJ: Describe your training. Are you trained as an electrical
engineer, computer scientist, programmer, other?
CIRCUIT CELLAR®
•
incorporate USB 3.0.
Perhaps the best news for developers is that prices for
debugging tools have plummeted. A bus analyzer captures,
decodes, and displays USB traffic and is essential for debugging. You can buy a low/full-speed analyzer for one-tenth the
price of a similar tool 10 years ago, and high-speed analyzers
are available for under $1,000. (Expect to dig deep for USB
3.0 tools, however.)
International is another good source for modules.
CJ: Where is USB technology headed? Is there something
beyond USB?
CJ: More advice. What would you say to a talented university
student who asked you to suggest a topic to research for a
dissertation?
CJ: Many Circuit Cellar authors and readers consider your
book, USB Complete, a “go-to” reference. What about you?
What books do you recommend on the subject?
JAN: For anyone using the new Vinculum-II embedded host
chip from FTDI, I recommend Embedded USB Design by
Example by John Hyde, a free e-book from www.usb-byexample.com. The .NET Micro Framework is an underappreciated and now mostly open-source programming platform
for small systems. A good programming guide is Expert
.NET Micro Framework (Second Edition) by Jens Kühner.
Programming 16-Bit PIC Microcontrollers in C: Learning to
Fly the PIC 24 and Programming 32-Bit PIC Microcontrollers in C: Exploring the PIC32, both by Lucio Di Jasio,
are great sources for useful code snippets and tips.
CJ: I enjoyed meeting you at the Embedded Systems Conference 2010 in San Jose. How do you keep up with the latest
technologies? By attending conferences?
JAN: The expo at ESC is a great way to find out what’s new.
There’s something about wandering the physical space and
talking with real people that’s completely different from
seeking out the same information on the web.
I find web forums useful for finding out what technologies
are catching the interest of developers (or causing them
headaches). I’ve added a PORTS forum to my website
(www.lvr.com) in part to replace the discontinued USB
developer’s forum at usb.org. Of course, Circuit Cellar provides inspiration for projects.
CJ: Did you see anything at ESC 2010 that really excited you?
JAN: I got a look at some inexpensive Linux-capable modules for high-end USB host applications. The BeagleBoard
(BeagleBoard.org) is based on a TI OMAP3530 processor
and has a large and enthusiastic user community. Emac,
Inc.’s iPac9302 board is based on a Cirrus EP9302 processor
and comes with Linux installed and an Eclipse IDE. Digi
•
JAN: My advice is to develop writing skills. Most jobs
involve some writing, and the ability to produce useful documents can make you stand out.
JAN: The iPhone aside, so many devices still have difficult
and frustrating user interfaces. I would like to see more
research on how to improve that situation.
CJ: If you were to research and write about something other
than USB, what would be?
JAN: I’m interested in wireless technologies, especially
inexpensive ways to do remote, real-time monitoring.
CJ: What projects are you currently working on?
JAN: I’ve been looking at ways to add USB host capability to
small systems. Because the USB host manages the bus, implementing a host is a much bigger job than implementing a
device. Microchip and other
vendors help by providing
example code for their microcontrollers with embedded
host-controller hardware.
If you want host capability
without messing with the
low-level details, a variety of
products can help. The USBwiz (see Photo 1) from GHI
Electronics and the Vinculum
from FTDI are host modules
that handle the USB host protocols and communicate via
commands sent over a UART, Photo 1—On this Microchip
Explorer 16 development board,
SPI, or other serial or parallel
a PIC24FJ256GA110 microconinterface. For devices protroller is used with a USBwiz
grammed using the .NET
OEM module to access USB
Micro Framework, GHI Elecdevices.
tronics has modules with
USB host support. Applications that require a major OS can use a module running
Linux or one of the Windows editions for embedded systems.
CJ: Are you planning any other projects for the near future?
JAN: Watch for my book on USB hosts for embedded systems
in 2011. Thank you for this chance to share my thoughts with
the readers of Circuit Cellar! I
CIRCUIT CELLAR®
JAN: USB 3.0 promises data throughputs of up to 400 MBps,
900 mA of bus current per device, and new power-saving
protocols. The interface adds a new 4-wire, 5-Gbps, dual
simplex bus that is similar to PCI Express. Higher-level
details such as the four transfer types and USB classes
remain the same.
CJ: Advice time. Is there one thing that you think every
electronics engineer needs to know? Why?
9
N
S
W
E
N
CT
DU
R
O
EW P
Edited by
John Gorsky
A MINI LINUX PC FOR
UNIVERSAL USE
The PortuxG20 is a miniature Linux
PC that’s energy-saving, compact,
and comes with numerous interfaces.
Its connection possibilities make it
ideally suited for a wide variety of
applications like data logging or computer interfacing. For instance, data
can be recorded over time and periodically read into another device for
additional processing as needed.
In addition to the immediately
available on-board interfaces, additional peripherals also can be connected. Depending on the multiplexing, up to four serial interfaces at TTL
level, two SPIs, one I²C-compatible
bus, an SSC, or up to 32 digital I/O
ports are available.
The SBC is equipped with an Atmel
AT91SAM9G20 processor core running at 400 MHz and includes 64-MB
SDRAM and 128-MB NAND Flash. The
board measurements are 100 mm ×
75 mm × 17 mm. It supports many
common interfaces, including Ethernet, USART, USB 2.0 full speed, Micro
SD card, JTAG, and the Portux Extension Bus (PXB).
Additional interfaces such as SSC,
SPI, TWI, an external SD card, a fourchannel 10-bit ADC, a 16-bit parallel
bus, and up to 64 digital I/O ports are
available via the PXB. Current consumption is 150 to 170 mA at 5 V
(with an active Ethernet interface).
Measurement, control, and regulation technologies are well-suited
applications for the PortuxG20 compact single-board computer. The
Linux operating system, the “U-Boot”
boot loader, and the “JamVM” Java
Virtual Machine with the “GNU Classpath” Java Class Library are all preinstalled.
The board itself costs about $225.
Several starter packages are also
available.
10
taskit GmbH
www.Taskit.de
CIRCUIT CELLAR®
•
CC430 PLATFORM FOR
COMPLETE SOLUTIONS FOR
WIRELESS NETWORKING
The CC430F513x MCUs combine
the world’s leading ultra-low-power
MSP430 MCU with the high-performance, sub-1-GHz CC1101 RF transceiver. The CC430F513x devices offer
up to 20-MIPS performance in a small
package (7 mm × 7 mm) and support
security options such as an integrated
AES hardware module. Also available
is the CC430F61xx series of LCD control devices, providing developers with
more options to meet varying design
needs. With the capability to support
multiple protocols and a broad frequency range, the CC430 MCUs, along with
the extensive third-party ecosystem,
spur innovation in applications like
home and building automation, smart
metering, energy harvesting, asset
tracking, and portable medical applications. Developers can also immediately
jumpstart CC430 MCU-based designs
with the EM430F6137RF900 and
eZ430-Chronos wireless development
tools that include all the hardware
design information required to develop
a complete wireless project.
Eight devices offer non-LCD and LCD
options, as well as a range of pin count,
memory and high-performance analog
integration to meet varying design
needs. The devices draw low enough current to enable battery-operated wireless
networking applications that operate without servicing for several years, reducing
maintenance costs and overall BOM.
Production quantities of the
CC430F513x MCUs are immediately
available and pricing starts at $4.15
(1,000 units). Sampling quantities of
the CC430F61xx family with integrated
LCD are also immediately available to
order with pricing starting at $4.35
(1,000 units).
ICs AUTOMATICALLY AND SAFELY DISCHARGE X
CAPACITORS
CAPZero is a family of innovative, two-terminal, automatic X capacitor discharge
ICs that eliminates power losses while allowing power supplies to comply with safety
standards.
CAPZero acts as a smart high-voltage switch when placed in series with discharge
resistors. When AC voltage is applied, CAPZero blocks current flow in the X capacitor
safety discharge resistors, reducing the power wasted in these components to zero
at 230 VAC. When the AC voltage is disconnected, CAPZero automatically and safely
discharges the X capacitor by closing the circuit through the bleed resistors and
directing the energy away from the exposed AC plug. This approach provides engineers with total flexibility in their choice of the X capacitor used to optimize differential-mode EMI filtering without worrying about the effect of the required bleed resistors on system no-load and standby power budget. The innovative design inherently
meets international safety standards for all open and short-circuit fault tests, allowing
CAPZero to be used before or after the system input fuse.
CAPZero is suitable for all AC-DC converters with X capacitors that require very low
standby power. It’s offered with 825- or 1,000-V MOSFETs to support a variety of
power supply design needs. It is ideal for a wide range of applications, including PCs,
servers/workstations, monitors
and TVs, printers and notebooks, and appliances requiring EuP Lot 6 compliance and
adapters requiring ultra-low
no-load consumption.
CAPZero devices are available now in an SO-8 package
at $0.40 each for 10,000piece quantities.
Power Integrations, Inc.
www.powerint.com
Leading Embedded
Development Tools...
Texas Instruments, Inc.
www.ti.com
®
!
"#$
$%
$
!&'"(
)$%
For ARM Application Processors:
*+$%
,$)
$
)
$+$
+
/$26
+%)2
+
.
NPN
•
www.keil.com
1-800-348-8051
CIRCUIT CELLAR®
For Microcontroller:
11
OBD-II INTERPRETER MODULE IN A DIP-24 PACKAGE
The microOBD 200 (#440101) is a complete OBD‐II
interface in a vibration-resistant DIP-24 module. The only
product of its kind, it offers engineers a simple and inexpensive way to add OBD-II support to any embedded project and tap into the wealth of information available on the
vehicle’s communication bus.
The interface is fully compatible with the popular
ELM327 command set and supports all legislated OBD-II
communication protocols, as well as the heavy-duty SAE
J1939. It features automatic protocol detection, a large
memory buffer, a UART interface capable of speeds of up
to 10 Mbps, and a bootloader for easy firmware updates.
The microOBD draws less than 1 mA in Standby mode,
which makes it suitable for permanent in-vehicle installations. The host can force the module to enter the lowpower state by sending it an explicit “sleep” command or
pulling the digital “host present” pin low. The module can
also put itself in Standby automatically on UART inactivity
or by sensing that the engine is off.
Typical applications include diagnostic scan tools, code
readers, data loggers, digital dashboards, fleet management, and vehicle tracking. Product documentation and
volume pricing are available from www.microOBD.com.
The module costs $49 in single-unit quantities.
ScanTool.net LLC
www.scantool.net
NPN
/,-
/,-
/, ( , -)&0 *,. $( '$&&$)( .) -#)1 3)/ 1#.5- ./&&3
#** ($(" /, ), $.
$"$.$4 , "$0 - 3)/ 2 && (. ( .#$- 2'*& 1 -'*& .1) -$(
10 - )! ( 4
'$2 .)" .# , #0 4))' )( %4 *),.$)( $( .# 4
$-*&3 (1$.# ). )/,
2 *.$)(& , -)&/.$)( $(. ,
')/&.$)( $-.),.$)( ( July 2010 – Issue 240
# $,- 12
'! ( $ %##$
#!%!! ! "%#& )$$
)! %$
&$%! & %$
!") "$%
#%!#
!# %# %
!# $"$ $%!#
* $" & !# % (% ## % "&%$
$ #% #$- +/$0& (. $-*&3
#- '2$'/' !, +/ (3
, -)&/.$)( )! *,. $( # $. $"$.$4 , &$'$.- )1 , $(
)''/($.$)(-
(3 )1 , $( )''- -3-. '/- (,,)1 ( -$"(&$(" $( .#
6 #4 !, +/ (3 ,(" )/ ( ")) !, +/ (3
, -)&/.$)( ( $"(& .) )$,.$) .) - 1#. $- ")$(" )(
-$(" & 0 ,-)* 3)/ ( # %
)/. )/*$ (1$.# .#
!! .- )! ()$- ( -1$.#$("
( &)" ,$0 0&/ - -/# ( 0)&." (5. - .# .1) * %- #
-$("& * % )/*$ - %4 )!
$-*&3 (1$.# # $-*&3 *),.$)( $- 4 1$ )/ #0
() $ 1#. $- ")$(" )(
CIRCUIT CELLAR®
•
CIRCUIT CELLAR
Edited by David Tweed
Problem 1—Assuming that VP and VN are connected to voltage sources (negligible source impedance), what is the value of VOUT of the following
circuit? The two RX have the same value, as do the
two RY. Assume the op-amp is “ideal.”
EQ
Test Your
Problem 3—What kind of feedback circuit do you
need in order to construct a switching regulator whose
output is exactly half of its input voltage?
Problem 4—In the C programming language,
name at least two ways in which the following
structure
for (expr1; expr2; expr3) {
...
}
is not precisely equivalent to this structure:
Problem 2—What is the significance of this circuit?
What’s your EQ?—The answers are posted at
expr1;
while (expr2) {
...
expr3;
}
www.circuitcellar.com/eq/
Contributed by David Tweed
You may contact the quizmasters at [email protected]
•
13
CIRCUIT CELLAR®
T
HE CONSUMMATE ENGINEER
by George Novacek (Canada)
It All Begins With a Spec
Your goal is to become a well-rounded, multidisciplinary engineer who can
address hardware and software issues alike, tackle difficult design problems,
and deliver quality work on a consistent basis. But achieving this goal takes
time, effort, and guidance. So let’s start with the basics.
E
ngineering is fun. Where else do we get
to invent, create and play with such
exciting toys—from iPhones to advanced military aircraft—and get paid for it? I know, I know,
we also have to worry about schedules, budgets,
and managers who are not engineers—but in the
end, nothing beats the satisfaction of seeing our
ideas become realities.
The creative fun, however, goes hand in hand
with a great deal of responsibility: to the end
users, to our employers, to our colleagues, to
our companies, and ultimately to ourselves in
terms of professionalism and pride. Not only
must we ensure our design will work as specified; we are equally responsible for making it
reliable, ensuring that it’s safe to operate, making it environmentally friendly, and ensuring it
can be manufactured consistently and economically. In future columns, I will address engineering activities and procedures to achieve those
goals. The idea is simple: you want to become a
well-rounded engineer. But that’s easier said
than done. So, let’s begin with the basics.
WHAT’S IN A SPEC?
14
At the beginning of all our efforts has to be a
good, concise specification defining in no uncertain terms what we, the engineers, are expected
to deliver. Unfortunately, this is not always the
case. Sometimes such a specification is impossible to develop, namely for systems and products
with long life expectancy, where continuous
technological improvement takes the product
well beyond its original concept. The Internet is
a good example. Let’s imagine what the Internet
could be, had its original creators had the benefit of knowledge of technological advances of
the several following decades and developed the
architecture and the protocols accordingly.
Abuse of the Internet would be much harder
today, although I don’t believe the Internet
could be made bullet proof.
Only a few of us will ever have the opportunity to be at the birth of a system with the
impact and longevity of Internet, where inadequacies of the original specification could not
be prevented. In the majority of development
programs there is no excuse not to start with a
specification cast in stone, with no detail left to
a chance. It has been shown that most of product shortcomings are not the result of poor
design, but poor, inadequate specification. Why
is that?
One excuse is economics. There were times
when top-level system integrators, such as airframe manufacturers, had a team of experts
specifying subsystems and their components as
black boxes in minute detail. A specification for
a simple embedded controller would have been
a hundred or more pages long.
For those not familiar with the concept, engineers like to talk about black and white box
designs. These exist at all levels of a system.
The black box design defines all the functions,
interfaces and performance characteristics as
viewed from the outside with no regard for how
are those design goals achieved. White box
design is the actual implementation, typically
schematic diagrams and source code for an
embedded controller.
CIRCUIT CELLAR®
•
SPEC AMBIGUITY
Spec ambiguity is another pitfall we need to watch for.
Engineers preparing a higher level specification may forget that while they understand the system well, engineers
given the task to design to this spec may, without understanding the context, give it a different interpretation.
Specifying the working environment of an equipment
is one thing, but good engineer must also consider what
extreme use and environments the equipment may be
exposed to and take steps to ensure the equipment
remains safe even outside its specified operating envelope. You must expect the unexpected. Once I worked on
a controller where four sets of 380-V loads were connected to four separate MIL connectors, each about 4 inches
in diameter. Because inserting a wrong plug into a wrong
receptacle could result in catastrophic consequences,
each connector set was color coded and keyed. We convinced ourselves that mismatching the connectors was
impossible.
Yet, it happened on the very first unit being installed.
The fireworks, I was told, were spectacular. Fortunately,
no one was injured, only the controller was destroyed
and the customer extremely upset. Within hours an
•
engineer was flying across the continent to investigate.
He discovered that the installer, when the very first
cable connector he touched didn’t mate with the receptacle, decided to take a monkey wrench to it, forced the
plug into the wrong receptacle, breaking off the key in
the process, then proceeded hammering the remaining
three connectors in their wrong places. To prevent the
“impossible” from happening again, we added monitors
to the controller to stop the power from being turned on
should someone invented another way to mate wrong
connectors.
While much of circuit design can be performed by junior engineers under supervision, the top-level spec development requires in addition to solid engineering knowledge something no school can give you—experience.
LOOKING AHEAD
Moving forward, I’ll share my engineering experiences
with you. Hopefully, they will help you to avoid learning
the hard way from your own mistakes. I
George Novacek ([email protected]) is a professional
engineer with a degree in Cybernetics and Closed-Loop
Control. Now retired, he was most recently president of a
multinational manufacturer for embedded control systems
for aerospace applications in Canada. George wrote 26 feature articles for Circuit Cellar between 1999 and 2004.
NEED-TO-KNOW INFO
Knowledge is power. In the computer applications
industry, informed engineers and programmers don’t
just survive, they thrive and excel.
To learn more about George Novacek’s design tips and
projects, the Circuit Cellar editorial staff recommends
the following:
—
Building an Electric Airplane
by George Novacek
Circuit Cellar 150, 2003
Many engineers see electric motor technology as
the ticket to an energy-efficient, cost-effective
future. With this in mind, George shows you how to
design an electrically actuated system, with a focus
on control and braking. Topics: Aeronautics, Actuator, PWM, Control, Braking, EMF, Fly-by-Wire
Go to: www.circuitcellar.com/magazine/150toc.htm
—
Working with EMC
by George Novacek
George explains differences between magnetic
and electric emissions. You can design your next
project to handle both. Topics: EMC, EMI, Electromagnetic Compatibility, EMI, Protection, Shielding
CIRCUIT CELLAR®
To reduce non-recurring engineering costs, the burden to
generate the black box definition of a sub-system, in other
words the specification, was gradually shifted down the
supply chain until it eventually ended up on the controller
designer’s desk as perhaps a single paragraph of technical
requirements, with a laconic statement “you are the
experts, write your own spec”. Needless to say, expecting
an electronics engineer, even though he may have designed
similar controllers in the past, to be an expert on, say, aircraft dynamics is quite a stretch. What looked as a perfect
way to save money to an accountant usually results in
even higher cost of development, blown schedules and finger pointing.
No better results come from a similar approach called
“concurrent engineering,” where time and money should
be saved by having the black and white boxes developed
concurrently. This is an engineering version of the chicken
or the egg dilemma. Several teams are second guessing each
other, occasionally ending up in a dead end or having to do
things the complicated way because the simple route had
been blocked. Over-design, spaghetti code, less than optimum design, those are the results.
Another cause of a poor up-front specification is that its
importance is underestimated, its preparation considered
boring. Many a hardware designer likes to start drawing
schematics, just as a software designer likes to start by
writing code without a previously developed and understood architecture and a set of requirements. This is a sure
way to get in trouble. No engineer should start white box
design before he has shown the black box to be correct,
traceable to the top level spec and testable. In my experience the time spent up front developing a solid spec supported by black box description has always brought the
best results.
15
F EATURE
ARTICLE
by Denny Lin (USA)
Ready to take a look inside an Ubuntu webcam server? Once you finish
this project, your viewers will have access to your streaming video via
any modern browser (Internet Explorer, Firefox, Safari, Opera, or
Chrome) and desktop OS platform (WindowsXP, MacOS, or Linux).
T
he Department of Mathematics and Computer
Science at La Sierra University (Riverside, CA) where I
work maintains a display area in which we showcase many
of the cool gadgets we develop. One such design is a computer-controlled monkey we affectionately call Kiki, which users
can control from any Internet connection and view via a
LinkSys webcam (see Photo 1).
The webcam originally had a serious shortcoming. It
required the use of Internet Explorer, an ActiveX module, and
a series of obscure procedures to enable the module. As a
result, people using non-Microsoft browsers couldn’t see
Kiki. We simply needed a video stream that was compatible
with as many browsers as possible.
Fortunately, I had a used IBM PC camera (which I’d bought
a)
16
at a thrift store for less than $4) and a bunch of computer
parts from a good friend. As a result, I decided to build a web
camera server from the ground up, one that would offer
broader browser compatibility.
While assembling this server, I realized that displaying the
inner workings of a functional server would effectively turn a
video webcam server into useful, and educational, eye-candy.
So, for $67, I bought a transparent computer case that came
with three 80-mm fans illuminated by eye-catching blue
LEDs (see Photo 2).
SYSTEM OVERVIEW
The design features several key pieces of hardware: an Intel
motherboard, a 1.70-GHz Intel processor, 256-MB memory, a
b)
Photo 1a—The Kiki computer-controlled monkey is located in a display case at the entrance to the Mathematics and Computer Science
Department at La Sierra University where I work. My webcam server monitors Kiki. b—This is the Altera Nios web server controller board. It is
the creation of Dr. Enoch Hwang at La Sierra University.
CIRCUIT CELLAR®
•
a)
b)
c)
Western Digital 6-GB hard drive, a BenQ
DVD-ROM (16×), an NVIDIA GeForc
video card, an Antec 300-W power supply, a Logisys Model CS888CL case, and
an Xirlink IBM PC Camera KSX-X9903.
Because I was building this system on
the cheap, the next step was to find a
free operating system. I chose the popular Linux distribution called Ubuntu,
and downloaded the entire operating
system as an image file to be burned
onto a bootable CD-ROM (which can
even run on a computer without a hard
disk). Figure 1 shows the relationship
between the Ubuntu webcam system I
describe in this article and the Kiki
Altera webserver displays. Figure 2
shows the components that make up the
Ubuntu webcam system.
Although there is a server edition of
Ubuntu 9.04, I chose the desktop version
because it’s easy to configure and provides a simple-to-learn graphical user
interface (GUI). Many configuration
steps still require the use of the command prompt; but fortunately, these
simple commands can be documented in
step-by-step instructions. This is how
•
the idea for this article began!
Building this Ubuntu webcam server
involves three main steps. Let’s consider
each one.
The first step is to install Ubuntu
onto the computer. This step is straightforward since the installation program
on the Ubuntu CD-ROM includes a
step-by-step “Wizard.”
The second step is to install, configure, and update the servers. This configures the computer’s operating system to
be used as a server and installs the two
servers that run the webcam. Critical
security updates are installed at the end
of this step.
The third step is to configure start-up
scripts to automatically start the webcam server. Doing so loads the IBM PC
camera driver that links the camera
images to the computer’s video device,
and it creates a start-up script for the
webcam server.
I added two additional steps. Although
they’re optional, I recommend them.
One, set up a remote desktop. Doing so
makes it possible to manage the server
from a remote computer. Next, set up a
firewall. This step disables all ports that
are not essential for the functioning of
the webcam server.
While the hardware and software cost
very little, considerable time and effort
went into learning about and setting up
the software. I found websites that contained bits and pieces of information
needed to get my hardware and software
to work. If your hardware or software
differs significantly from mine, use common sense (and lots of patience) to troubleshoot any issues.
The configuration files were created
and modified by a text editor called Pico,
which is included in this Ubuntu distribution. While Pico lacks a GUI, I found
it easy to use. Context-sensitive commands, listed on the bottom of the
screen, can be accessed by pressing a
combination of CTRL and letter keys.
For example, help (labeled ^G Get
Help) is accessed by pressing CTRL-G.
INSTALLING UBUNTU
To begin, find and download the 32-bit
Desktop Ubuntu 9.04 disk image. Any
computer with a browser connected to
CIRCUIT CELLAR®
Photo 2a—An IBM PC camera is pointed at Kiki. b—This is my completed
webcam server housed in a transparent case. c—This is a close-up shot of
the server. The system is pretty eye-catching, especially with its blue LEDs.
17
Internet
Forwards off-campus requests for
192.156.214.81 to 10.1.31.24
Forwards off-campus requests for
192.156.214.80 to 10.1.31.23
IP Address forwarding firewall
Ethernet network traffic
Ethernet network traffic
Network switch
Local IP Address:
10.1.31.24
Local IP Address:
10.1.31.23
Video
IBM PC
USB
Camera
Drumming
monkey
3-V DC
On/Off
Relay
115-VAC
to 3-VDC
Adapter
Ubuntu webcam server:
Host server’s main page, the
webcam video, and embeds
the Altera web server page
Altera web server
and Kiki controller:
Host embedded page
containing the monkey
on control button
Welcome to Kiki’s Webcam!
Webcam video
(Java applet)
Existing Altera
controller monkey
display by
Dr. Enoch Hwang
Inline frame (iframe) tag embeds the
Altera web server page within the
Ubuntu webcam server’s main page
The server’s
index.html
main page
Monkey On
Ubuntu webcam
system display
Figure 1—This block diagram shows the relationship between the Ubuntu webcam system and the Kiki Altera web server displays.
the Internet, equipped with a CDROM drive and burner software, can
be used to create (burn) the CD-ROM.
Using a Windows XP computer, I
saved the image file, and
right-clicked on its icon to
burn an Ubuntu-bootable CDROM.
I configured my computer’s
ROM-BIOS to boot from the
CD-ROM drive. I completed
the installation using the
steps noted in Figure 3.
IBM
USB
Camera
Sends video to
Supplies video feed to
IBMCAM USB
Driver module
18
Runs on
Configuration
and
maintenance
tools
Executes
Configured and
maintained by
Runs on
Webcam
server
Starts
Loads driver
CONFIGURE SERVERS
First, the computer is set to
a static Internet protocol (IP)
address and domain name
server settings are configured.
Second, the Apache web server is installed and configured
to host the video stream produced by the webcam server.
Third, an HTML file that
serves the video stream from
the webcam server is configured. The Update Manager
must be allowed to download
users have a fixed target to connect
to the webcam server. In addition,
having a registered domain name
makes it easier to access your server. If
critical security updates that have
become available since Ubuntu was
released.
You need a static IP address so that
Startup
modules,
scripts, and
symbolic links
Provides
webcam
video to
Runs on
Ubuntu 9.04
operating
system
Runs on
Runs on
Sends web page to client via
Network
interface
Apache web
server
Forward client request to
Get requests from
and send images to
Internet
Figure 2—This is a context diagram of the components described in the article that make up the
Ubuntu webcam system.
CIRCUIT CELLAR®
•
The Newest Products
For Your Newest Designs
Get Connected Faster.
CFP Compliant Ethernet Connectors
& Components
mouser.com/tycocfpcompliantconnector
Molex iPass+™ HSC CXP Interconnect System
mouser.com/molexipassplus
Ultra Hard Metric (UHM)
Backplane Connector
www.mouser.com/3muhm
USB 3.0 SuperSpeed
mouser.com/AmphenolUSB3
WARNING: Designing with Hot, New Products
May Cause A Time-to-Market Advantage.
Micro USB, Industrial Ethernet, Hard Metric, Fiber Optic, MicroTCA, HDMI,
SFP, Mini SAS, SATA, QSFP, SFP+, Displayport get the newest Interconnect
products and technologies. Experience Mousers time-to-market advantage
with no minimums and same-day shipping of the newest products from more
than 400 leading suppliers.
mouser.com
Mouser and Mouser Electronics are registered trademarks of Mouser Electronics, Inc. Other products, logos, and company names mentioned herein, may be trademarks of their respective owners.
(800) 346-6873
the computer is
turned off and on.
If you are creating
a web server of any
sort, I recommend
purchasing a static
IP address plan
from your ISP and
getting a domain
name from a
domain name registrar. Another
alternative is to
use a free domain
name service that
can translate your
dynamic address to
a fixed domain name that you choose.
Since this webcam server is located
at my university, I picked an IP address
from a pool of static addresses reserved
1. Insert the Ubuntu 9.04 CD ROM and boot computer from the CD ROM
2. Select English
3. Select Install Ubuntu, press Enter
a. Wizard Step 1: Select English, click Forward
b. Wizard Step 2: Pick your region and city, click Forward
c. Wizard Step 3: Pick a keyboard layout (I used Suggested option: USA), click
Forward
d. Wizard Step 4: Select Use the entire disk, click Forward
e. Wizard Step 5: Enter your name, login and password, and computer
information.
i. Enter your full name: [administrator]
ii. Pick a login name: [administrator]
iii.Pick your password <enter this twice>
iv. Name your computer: [administrator-desktop]
v. Click Log In Automatically
vi. Click Forward
f. Wizard Step 6 is automatically skipped. In Wizard Step 7: Click Install
4. Restart computer: Click Restart Now
5. Remove CD ROM when prompted, and press Enter
Fi gure 3—The steps for installing Ubuntu
your ISP provides a dynamic IP address
(using the dynamic host configuration
protocol or DHCP), your server’s IP
address could be different every time
by our campus network administrator.
I also set up the domain search and
name server (DNS) look-up information in the webcam server’s resolv.conf
file. I received valuable help and advice
from the university’s network administrator while setting up domain names
for the webcam server.
Next, I had to download and install
servers. I used Donn Morrison’s webcam-server program to make the video
stream available to the Apache web
server. Morrison’s Sourceforge.net page
provides the webcam_server (notice the
underline in the name) program, but
that did not recognize my webcam.
The webcam-server (notice the hyphen
in the name) program (also authored by
Morrison) works well with my webcam, and it is the version that is available using the apt-get command in
1. Start Terminal: Go to Applications -> Accessories -> Terminal
2. Type sudo -i
a. press Enter.
b. enter the administrator’s password.
3. Define a static IP address:
a. Go to the network directory. Type: cd /etc/network
b. Type: pico interfaces
c. Modify the interfaces network settings file so the contents of that file looks like listing 1. In our case, we set the static IP address to 10.1.31.241.
You must use a different IP address, so consult your Internet Service Provider (ISP) or your network administrator for details about your particular
environment.
d. Exit pico: CTRL-X, Y(es) to save changes, press Enter to accept filename and exit.
4. Restart the computer. Type: reboot
5. Repeat steps 1 and 2 to use Terminal to get super user access
6. Configure the domain search and name server settings in the resolv.conf file.
a. Go to the /etc directory. Type: cd /etc
b. Type: pico resolv.conf
c. Edit the resolv.conf file so the contents of this file looks like listing 2. In our case, the domain search is lasierra.edu, and our DNS server address is
192.156.214.14. You must use a different address, so consult your Internet Service Provider (ISP) or your network administrator for details about
your particular environment.
7. Install webcam-server. Type: apt-get install webcam-server
8. Install apache2 server. Type: apt-get install apache2
a. Type Y to continue installation.
9. Install the Java browser plugin. Type: apt-get install icedtea6-plugin
a. Type Y to continue installation.
10. Copy webcam server files to apache server directory.
Type: cp /usr/share/doc/webcam-server/applet/* /var/www
11. Change to /var/www. Type: cd /var/www
12. Edit webcam.html and change the URL to your IP address; change the frames per second to 5, using listing 3 as reference:
a. Type: pico webcam.html
b. Look for the phrase http://localhost:88882 and replace “localhost” with the IP address you picked in step 3. For example, if your server’s IP address
is 10.1.31.24, change
this to http://10.1.31.24:8888
c. Look for the phrase FPS value= “1” and replace “1” with “5”.
13. Copy webcam.html to index.html. Type: cp webcam.html index.html
14. Go to the Update Manager
a. Click Check to look for new updates.
b. Enter the administrator password.
c. Click Install Updates
15. Restart computer, click Restart Now
20
1
The webcam server is actually located within a private sub-network; this is why I have a static IP address that starts with “10.” It is normally not possible for offcampus computers to directly connect to any devices in our private sub-network. So, a special public IP address “192.156.214.81” and a domain name of
“cswebcam.lasierra.edu” were created by my network administrator to redirect traffic from off-campus machines to the webcam server in our private sub-network.
On-campus users must either type the private IP address or use the private domain name “cswebcam.int.lasierra.edu” (notice the .int. in the domain name) to
access the webcam server.
2
The webcam-server uses port number 8888 to send the video stream. If the client’s ISP blocks this port, they will get an “Error connecting to the video”
message, though the rest of the web page will be visible.
Figure 4 —The steps for installing and configuring servers
CIRCUIT CELLAR®
•
Figure 4 (for installing and configuring servers). Morrison’s program provides a simple webcam_server.log
file, in the /var/log directory, with
connection statistics such as the
date, time, and requestor IP address.
Apache is used to serve the webcam
page containing the video stream. This
full-fledged web server is capable of
hosting Java applets and even keeping
track of website traffic. The access.log
file can be used to analyze web traffic.
It is found in the /var/log/apache2
directory and details every connection
request (including IP address, operating system and version, browser and
version, access date and time, which
pages were visited) made to the web
server during the current week. In
addition, it is automatically rotated to
access.log.1 at the end of each week.
Li sting 1—Contents of the interfaces file, found in /etc/network directory
# File path and name: /etc/network/interfaces
# Define the loopback interface lo
auto lo
iface lo inet loopback
# Define the interface eth0 with static IP address
auto eth0
iface eth0 inet static
address 10.1.31.24
netmask 255.255.255.0
network 10.1.31.0
broadcast 10.1.31.255
gateway 10.1.31.1
L isti ng 2 —Contents of the resolv.conf file, found in the /etc directory
# File path and name: /etc/resolv.conf
search lasierra.edu
# Search list for host-name lookup
nameserver 192.156.214.14
# IP address of the campus name server
L isti ng 3 —Contents of the webcam.html and index.html files, found in the /var/www directory
22

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Kiki's Webcam</title>
<style type="text/css">
html, body, div, iframe { margin:0; padding:0; height:1020; }
iframe { display:block; width:100%; border:none; }
</style>
</head>
<body>
<table cellspacing="15" width="80%" align="center">
<tr>
<tr>
<CENTER><H1>Welcome to Kiki's Webcam at La Sierra University!</H1></CENTER>
</tr>
<td>
<APPLET CODE = "WebCamApplet.class" archive="applet.jar" WIDTH = "352" HEIGHT = "240">
<param name=URL value="http://10.1.31.24:8888">
<param name=FPS value="5">
<param name=width value="352">
<param name=height value="240">
</APPLET>
</td>
<td>
Kiki's Webcam runs on Linux Ubuntu 9.0.4, which can be downloaded for free
<A HREF="ubuntu-9.04-desktop-i386.iso">here</A>.<P>
You can get step-by-step instructions to make an Ubuntu webcam, in
<A HREF="UbuntuWebcam.doc">MS-Word format</A> and in
<A HREF="UbuntuWebcam.pdf">PDF format</A>.<P>
The computer that runs Kiki's Webcam was donated by Craig Kinzer, alumnus of the
<A HREF="http://www.lasierra.edu/cs">Math & Computer Science</A> department.<P>
This webcam display, and the Ubuntu webcam setup instructions were created and written by
<A HREF="http://www.lasierra.edu/~dlin">Denny Lin</A>, Assistant Professor of Computer Science.
</td></tr>
</table>
<div>
<iframe src="http://10.1.31.23" height="1020" width="100%">
<p><a href="http://10.1.31.23">Kiki Cam</a></p>
</iframe>
</div>
</body>
</html>
CIRCUIT CELLAR®
•
What is the
missing component?
Electronics instructor Ollie Circuits planned to show his class of freshman electrical
engineering students how to use a super capacitor as a memory back-up capacitor,
but first he wanted to show how the students could make their own super capacitor
and demonstrate its charge/discharge cycles with the simple circuit above. Most of
the components were already on his workbench, the homemade super capacitor
would be made from several layers of lemon juice-soaked paper towels interleaved
between several layers of a mystery material to form a multi-layer stack.The stacked
layers would then be sandwiched between the two copper-clad PC boards and held
together with a rubber band. Ollie rushed to a nearby pet shop. What did he buy?
Go to www.Jameco.com/teaser7 to see if you are correct and while you are there,
sign-up for our free full-color catalog.
1-800-831-4242 | www.Jameco.com
2. Type sudo -i
a. press Enter
3. Change to the etcetera directory. Type: cd /etc
4. Type: pico modules
5. Listing 4 is the /etc/modules file, which loads the ibmcam module. Make sure the contents of the options following ibmcam are on a single line.
6. Change to the startup directory. Type: cd /etc/init.d
7. Type: pico webcam_server
8. Create the startup script by typing the contents of listing 5.
9. Exit pico: CTRL-X, Y(es) to save changes, press Enter to accept filename and exit.
10. Grant execute permission to the file you created. Type: chmod 755 webcam_server
11. Define a symbolic link so the script is automatically executed when the computer starts. Type: update-rc.d webcam_server defaults 24
Fi gure 5 —The steps to configure automatic start-up scripts
24
the word “localhost” with the server’s
asked for the administrator’s password
Older files are automatically comstatic IP address. Then, the frames per
before updates were installed, and I was
pressed and saved as access.log.n.gz,
second (FPS) setting was increased from
prompted to restart the computer for
where the number n is two or larger for
1 to 5. I picked a value which would
the updates to take effect.
older files.
not burden the server with a too-high
Regarding the static IP address to
Next, I downloaded and installed the
frame rate, but would still allow the
10.1.31.24 in Figure 4, note that the
Java browser plugin. (Windows and OS X
user to observe the motion of Kiki beatwebcam server is actually located withcan run Java applets without this Java
ing her drum. Finally, for testing, I
in a private sub-network. This is why I
plugin, so step 9 in Figure 4 for
entered http://10.1.31.24/webcam.html
have a static IP address that starts with
installing and configuring servers is
“10.” It is normally not possible for
arguably optional.) The video stream
into a browser running on a different
hosted by the webcam server is a Java
computer and modified the webcam.html off-campus computers to directly connect to any devices in our private subapplet that runs within the web browsfile on the server until I got the
network. So, a special public IP address
er. In order to view the video stream
desired results.
“192.156.214.81” and a domain name
while using the server’s web browser, I
Once I was satisfied with the webof “cswebcam.lasierra.edu” were creatinstalled the Iced Tea 6 Java plugin.
cam.html file, I copied it to
ed by my network administrator to
This project aimed not only to show
index.html, the default file that is
redirect traffic from off-campus
video images of Kiki, but also to allow
automatically displayed whenever a
machines to the webcam server in our
users to control Kiki. The webcam-servuser types the IP address or the domain
private sub-network. On-campus users
er package provides a default web page
name of the webcam server into a
(named webcam.html) that serves a sim- browser: http://10.1.31.24 (notice there
must either type the private IP
ple video stream. I made several changes is no /index.html after the IP address).
address, or use the private domain
to this file so that it is seamlessly
I then ran Update Manager. This tool
name “cswebcam.int.lasierra.edu”
linked to Kiki’s controller web page
lists patches and updates that keep the
(notice the .int. in the domain name)
(served by a stand-alone Altera Nios
server current and protected from vulto access the webcam server.
board) while serving the video stream. I
nerabilities. Ubuntu automatically runs
As for the note regarding the phrase
used inline frame tags to seamlessly
the Update Manager to look for and
http://localhost:8888 in Figure 4, the
link Kiki’s controller web page to the
install updates (step 14 in Figure 4 for
webcam-server uses port number 8888
video stream in the webcam.html page,
installing and configuring servers). I was
to send the video stream. If the client’s
so there are no wires that
directly connect the webLi sting 4—Contents of the modules file, found in the /etc directory
cam server to Kiki’s controller web page. Since I
# File path and name: /etc/modules
wanted to turn this webibmcam init_hue=96 init_color=250 init_brightness=48 init_contrast=110 lighting=0
site into a teaching tool,
init_model2_yb=170 init_model2_sat=28
links to the step-by-step
# Loads the ibmcam module (driver) upon startup
instructions for creating
# All of the numbers have a range from 0 to 255:
#
The initial hue set is 96
this webcam server were
#
The initial color is set to 250
posted in index.html.
#
The initial brightness is set to 48
Clearly, your circum#
The initial contrast is set to 110
stances will vary, so
#
The lighting (photosensitivity of camera) is set to type 0
#
The init_model2_yb parameter adjusts color balance, and is set to 170
adding inline frame tags,
#
The init_model2_sat parameter adusts the saturation, and is set to 28
for example, may not
apply to your project.
# More information about options for the ibmcam module can be found here:
I first edited
# http://www.mjmwired.net/kernel/Documentation/video4linux/ibmcam.txt
webcam.html by replacing
CIRCUIT CELLAR®
•
Listi ng 5 —Contents of the webcam_server file, found in the /etc/init.d directory
#!/bin/sh
# File path and name: /etc/init.d/webcam_server
# The first two characters in this file MUST be a number and an exclamation mark, because
# they tell the operating system that this is a script, to be executed by the Bourne shell
# interpreter found in /bin/sh
webcam-server -s -c " Kiki The Monkey %D %H:%M:%S Viewers: %%v"
# The -s option starts the webcam-server in daemon mode
# The -c option defines the caption to be displayed:
#
%D inserts the system date
#
%H inserts the current hour from system time
#
%M inserts the current minute from system time
#
%S inserts the current seconds from system time
#
%%v inserts the number of connected viewers to the webcam-server
ISP blocks this port, they will get an
“Error connecting to the video” message, although the rest of the webpage
will be visible.
Listing 1 shows the content of the
interfaces file found in /etc/network
directory. Listing 2 shows the content
of the resolv.conf file found in the /etc
directory. Listing 3 shows the content
of the webcam.html and index.html
files found in the /var/www directory.
This step lists the IBM PC camera
driver in the modules file, so the driver is loaded when the server starts up.
In addition, this step creates a startup
script that launches the webcam server in daemon mode. Instructions to
configure automatic startup scripts are
listed in Figure 5.
During start-up, Ubuntu loads drivers listed in the modules file. The
ibmcam module allows the IBM PC
camera to send its images to the
/dev/video0 device. I found that when
the module was loaded without any
options, the images appeared overexposed; under minimal light, the image
had a greenish tinge. I came across a
website describing the options available for adjusting the video signal. I
spent several hours of trial-and-error
testing to figure out the best combination of settings for initial hue, color,
brightness, contrast, and lighting. Subtle changes in lighting require different adjustments, so be patient while
trying different settings.
I named the webcam server startup
script “webcam_server” in order to
differentiate it from the executable file
(which contains a hyphen instead of an
•
CIRCUIT CELLAR®
START-UP SCRIPTS
25
26
underline). All of the
Li sting 6—Listings of the /etc/rc2.d and /etc/rc0.d directories
options for the webcamserver program can be
>ls /etc/rc2.d
README S20hotkey-setup
S50rsync
S90binfmt-support
viewed by typing “webS01policykit
S24hal S50saned
S91apache2
cam-server –h” in a termiS10acpid
S24webcam_server
S50system-tools-backends
S98usplash
nal window. I only needed
S10apmd S25bluetooth
S70bootlogs.sh S99acpi-support
two options for my setup.
S10sysklogd
S30gdm S70dns-clean
S99laptop-mode
S11klogd
S50avahi-daemon
S70pppd-dns
S99ondemand
The -s option was used to
S12dbus S50cups S89anacron
S99rc.local
launch the server in daeS16ssh S50NetworkManager
S89atd S99rmnologin
mon mode, so its execuS20apport
S50pulseaudio S89cron S99stop-readahead
tion is not tied to a termi>ls /etc/rc0.d
nal session. The -c option
K01gdm K63mountoverflowtmp
S30urandom
defined the caption that
K02usplash
K74bluetooth
S31umountnfs.sh
identifies the video and
K09apache2
K99laptop-mode S35networking
shows the date, time, and
K20apport
README S40umountfs
K24webcam_server
S01linux-restricted-modules-common
S60umountroot
number of users connected.
K25hwclock.sh S15wpa-ifupdown S90halt
The webcam_server
K50alsa-utils S20sendsigs
startup script is kept in
the /etc/init.d directory;
it’s made executable by using the chmod command in step
sequence, remove the symbolic links. First, type:
10 in Figure 5 (for configuring automatic start-up scripts). It
“update-rc.d - f webcam_server remove.” Then reissue the
turns out that the order for executing this script is very
update-rc.d command in step 11 from Figure 5 (for configurimportant. The Ubuntu Hardware Abstraction Layer must
ing automatic start-up scripts) to establish a symbolic link.
be started before the webcam_server script.
Listing 4 shows the content of the modules file found in the
Most users will run their Linux desktops at run level 2, 3,
/etc directory. Listing 5 shows content of the webcam_server
4, or 5, which are multi-user modes. By default, Ubuntu uses file found in the /etc/init.d directory. Listing 6 shows the conrun level 2. Run levels 0 and 6 are used to kill processes durtent of the /etc/rc2.d and the /etc/rc0.d directories, produced
ing system shutdown or reboot, while run level 1 is used
by the ls command displayed at the top of each list.
when Ubuntu is in single-user mode. Each run level has a
Let’s move on to the topics of setting up the remote desktop
matching control directory that contains executable symboland firewall. Both are optional.
ic links to services that are started or ended during startup,
shutdown, or reboot. Thus, for run level 2, the symbolic
REMOTE DESKTOP SETUP
links are in /etc/rc2.d, while the symbolic links for run level
Remote Desktop is a virtual network computing (VNC)
0 are stored in /etc/rc0.d. These symbolic links are identified server program that runs on the webcam server, allowing
by the initial letter S to start processes, or with the initial
remote control of the server through a VNC client viewer.
letter K to kill processes.
The computer in our display case is not connected to a
The number after the letter S of every symbolic link
monitor, but VNC allows me to see what the webcam server
denotes the order in which those services are started. Since
is displaying. While I can and do use a secure shell terminal
the webcam_server must be started after the hardware
emulator program (PuTTY) to quickly access the server,
abstraction layer (HAL), its number should be equal to or
VNC provides access to Ubuntu’s GUI desktop environment.
greater than the number for the HAL service (labeled as
VNC is platform-independent, so any computer running a
S24hal). There should be S24webcam_server links in the
VNC client can be used to control your webcam server. You
/etc/rc2.d, /etc/rc3.d, /etc/rc4.d, and /etc/rc5.d directories. For
can download the “VNC Free Edition for Windows” if you
wish to control the webcam server from a Windows PC. Try
run levels 0, 1, and 6, there are symbolic links that start with
“Chicken of the VNC” if you wish to control the webcam
K, which kill processes during shutdown, single-user mode,
server from a Macintosh. Refer to the resources listed at the
or reboot. There should be K24webcam_server links in the
end of this article for more information.
/etc/rc0.d, /etc/rc1.d, and /etc/rc6.d directories. These startTo connect to your webcam server through the VNC
up and shutdown symbolic links were created using the
client, enter the IP address of the webcam server and the
update-rc.d command in step 11 (for configuring automatic
password chosen in step 2 of Figure 6 (to set up Remote
startup scripts) shown in Figure 5.
Desktop). Pick a secure password that is at least eight charIf you need to change the start-up or shutdown
acters long.
Set up the VNC server pro1. Start Remote Desktop. Go to System -> Preferences -> Remote Desktop
2. On the Remote Desktop Preferences dialog box:
gram so that authorized users
a. Check the “Allow other users to view your desktop” and “Allow other users to control your computer” checkboxes.
can view and control your
b. Un-check the “You must confirm each access to this machine” checkbox.
c. Check the “Require the user to enter this password” checkbox, and enter a secure password.
computer, and remove the
d. Click Close
requirement to confirm each
Figure 6—The steps to set up Remote Desktop
access to the webcam server.
CIRCUIT CELLAR®
•
2. Type: sudo -i
a. press Enter
3. Check the ufw status. Type: ufw status (you should get a Status:
inactive)
4. Set ufw to deny all connections by default. Type: ufw default deny
5. Allow browser traffic. Type: ufw allow 80
6. Allow secure shell traffic. Type: ufw allow ssh
7. Allow webcam traffic. Type: ufw allow 8888
8. Allow remote desktop traffic. Type: ufw allow 5900
9. Enable ufw. Type: ufw enable
10. View the status of the firewall, and list ports you have opened. Type
ufw status
Figure 7—The steps to set up the firewall
D. Morrison, Webcam Server Project, http://webcamserver.sourceforge.net/.
IBM C-it USB Camera Driver FAQ, www.linux-usb.org/
ibmcam/ibmcamFAQ.html.
IBM C-it USB Camera Driver for Linux, www.linuxusb.org/ibmcam/.
Official Ubuntu Documentation, https://help.ubuntu.
com/.
PuTTY: A Free Telnet/SSH Client, www.chiark.green
end.org.uk/~sgtatham/putty/.
During set-up, I kept a monitor connected to the webcam
server while connecting from a remote VNC client to
make sure that my remote computer could display the
same images as the webcam server’s monitor.
README for the Linux device driver for the IBM CIt USB Video Camera, www.mjmwired.net/kernel/
Documentation/video4linux/ibmcam.txt.
RealVnc, www.realvnc.com.
Ubuntu’s uncomplicated firewall (ufw) front-end program easily removes avenues used by hackers to attack
your server. I shut off all network ports to the webcam
server and then reopened just the ports the server requires
to provide its services.
Port 80 must be open to serve web pages. Port 8888
must be open to allow the webcam-server software to
stream video. I opened port 22 (ssh) to allow secure shell
terminal emulator access. Port 5900 was opened to allow
Remote Desktop (the VNC server) to connect. Figure 7 is
a list of the steps associated with setting up the firewall.
DESIGN SUCCESS
This project taught me a lot about setting up an Ubuntu
server and interfacing it with a USB camera. And now it
serves to inspire curiosity and excitement about computer
technology in our department’s display case. Each day,
curious students are lured to the display because of the
attractive lights on the server and Kiki’s drumming. One
student called his father, who lives several hundred miles
away, to watch and control Kiki with the webcam server.
You too can connect to the webcam server and control
Kiki at http://cswebcam.lasierra.edu. I
Denny Lin ([email protected]) has a BA in Music and a BS and an
MSc in Computer Science. He is an Assistant Professor of Computer Science and a Senior Programmer/Systems Analyst at La
Sierra University. Besides teaching C++ Programming, Software
Engineering, and Systems Analysis courses, Denny also teaches
Computer Music Synthesis courses in the Music Department and
Electronics in the Physics Department.
Ubuntu 9.04 (Jaunty Jackalope), http://releases.ubuntu.
com/jaunty/ubuntu-9.04-desktop-i386.iso.
SOURCES
Nios embedded processor
Altera | www.altera.com
NEED-TO-KNOW INFO
industry, informed engineers and programmers
don’t just survive, they thrive and excel.
For more need-to-know information about topics
covered in Denny Lin’s Issue 240 article, the Circuit
Cellar editorial staff highly recommends the following
content:
—
Build Your Own 8051 Web Server
by Jim Brady
Building a web server doesn’t have to be difficult.
Jim built an 8051 server and describes the project in
this article. Topics: Web Server, TCP/IP, Ethernet
Chicken of the VNC, http://sourceforge.net/projects/cotvnc/.
—
Web Camera Design
by Minas Kalarakis
This versatile web camera system can take pictures, pan a camera, and change its IP and gateway
address to match a network. An Ethernet module
transmits the packets over the Internet. Topics:
Webcam, Servo, Ethernet, Packet Transmission
Dynamic Network Services, www.dyndns.com/.
Go to: www.circuitcellar.com/magazine/228.html
RESOURCES
•
CIRCUIT CELLAR®
FIREWALL SETUP
27
F EATURE
ARTICLE
by Devlin Gualtieri (USA)
Build an X10 Controller (Part 1)
Circuit Design
This article series details the process of building an X10 controller from
start to finish. Here you learn about the circuit design before moving on
to the software.
B
ack in the 1980s, I combined my interest in
carrier-current communication technology and
embedded systems to design my first timed automatic control system. The design featured a Zilog Z80 microprocessor and a carrier-current transmitter that I built myself.
Three decades later, I still find myself rethinking my first
design. But now I have a lot more experience under my belt
and new technologies to put into play.
In this article series, I’ll describe my original
X10 controller and then present my newest
design. This article covers the circuitry (see
Photo 1). Next month, I’ll describe the software
side of the project.
When I went off to college, I saw that the same trick was
used by the campus radio station. It had transmitters on
the AM broadcast band coupled to the power lines in each
of the dormitories. In that way, a low-power signal, legal
under the FCC Part 15 rules, could reach a large audience.
It was still done using vacuum tubes.
Carrier-current communication didn’t enter my mind
again until I was a homeowner in the early 1980s wrestling
CARRIER CURRENT
28
When I was a grammar-school student, my parents had what I considered a magical device in
their kitchen. It was an intercom that allowed
them to talk with our next-door neighbor who
had an identical unit. The amazing part about
these intercoms was that there were no wires
connecting the devices. They just plugged into
an electrical outlet. Later, when I was in high
school and more familiar with electronics, I realized that they were carrier-current devices. They
had a low-frequency amplitude-modulated transmitter and a complementary AM receiver. The
transmitter signals were injected into the power
lines, and that’s how they traveled between
houses. Everything was done using vacuum tubes
because it was the 1960s at the time.
Photo 1—The finished controller. The ICSP connector is visible at the bottom of
the board. The back-up lithium cell, in a home-built battery holder made from a
small plastic bottle and some switch parts, is at the lower right.
CIRCUIT CELLAR®
•
•
CIRCUIT CELLAR®
with the unreliable mechanwith the zero crossings of the
8.333
ical timers I was using to
AC line, they’re able to exam5.556
control the exterior lighting
ine the proper time interval
at my home. At about the
to determine the presence or
2.778
same time, the X10 system
absence of this signal. The
1
hit the stores. X10 is a carripresence of 120 kHz at the
er-current system that
appointed time is interpretallows control of lights and
ed as a logical one, and its
appliances. It’s a nicely inteabsence is a logical zero.
1
grated system of modules
Only a single bit is trans2.778
that replaces the light
mitted at each zero crossswitches and utility outlets
ing. The data protocol calls
5.556
in your house. These modfor a 4-bit preamble, fol8.333
ules have the same function
lowed by an 8-bit house
as the switches and outlets
code and a 10-bit device
Figure 1 —Timing in milliseconds of the carrier-current signal for
they replace, but they
number
or function code.
X10 communication. The data is transmitted as 1-ms bursts of a
include remote switching
The 8-bit house code is actu120-kHz carrier synchronized with the zero crossings of the 60-Hz
voltage sine wave. Each bit is transmitted redundantly three times
circuitry. I was able to purally just four bits of informain a half cycle.
chase these X10 devices
tion, but each bit is followed
with a companion control
by its complement. If a logibox that injected carrier-current sigcal one is transmitted on one zero
specifications, and I was able to build
nals into the power lines to turn
crossing, a logical zero is transmitted
a reliable X10 controller using 1980slights on and off, brighten and dim
on the next. The same is true for the
era technology.
lights, and turn appliances on and off.
device number/function code. It’s just
All this was nice when I was present
five bits alternated with their compleTHE X10 SYSTEM
to push the buttons on the control
ments. Figure 2 summarizes the transThe X10 signal, as shown in Figure 1,
box, but I wanted a timed automaticmission protocol. Table 1 shows the
is a pulsed 120-kHz oscillator that’s
control system. I was doing embedded
binary values of the house codes and
synchronized with the zero crossing
systems work in the same period, so I
device codes. Table 2 shows the binary
of the AC power line. The X10 innohad the tools to roll my own concommand codes. The preamble, which
vation, which made it a patentable
troller. In those days, this was no sim- idea, was this synchronization of the
is not interleaved with its complemenple task. An embedded system
tary bits, is always 1110. Since it has
datastream with the AC power line
involved a Z80 microprocessor, a mul- zero crossings. The receiver modules
three logical one bits in series, it can’t
titude of peripheral chips, an ultraviobe confused with the home or
know when to look for data, and this
let-light-erasable EPROM burned with leads to noise immunity and
device/function data.
an assembly language program, and a
enhanced reliability. The transmitter
few kilobytes of static RAM with a
accommodates the existence of
INTO MODERNITY
hefty NiCd battery backup.
receivers on any of three possible
My original X10 controller had an
There was another problem. I needphases of the AC line by transmitting
event table that was updated by a
ed to design and build my own carriredundant signals at times where the
serial data link, so it survived my
er-current transmitter to mimic the
zero crossings at these other phases
transition from an early CP/M comX10 signals. These were the days
would occur.
puter system to an MS-DOS computbefore the Internet. But fortunately,
The data signals are transmitted in
er. After a time, computers became
some hobby magazines published
short bursts of the 120-kHz oscillator.
ubiquitous enough for manufacturers
information about the X10 signal
Since the receivers are synchronized
to notice, and an X10 controller with
a similar serial access became
available. This controller was
the CP290, which came with a
1
2
3
4
5
6
7
8
9
10
11
simple DOS program for programming events and enough
H1 * H1 H2 * H2 H4 * H4 H8 * H8 D1 * D1 D2 * D2 D4 * D4 D8 * D8 D16 * D16
Preamble
information for programmers to
1
1
1
0
0
1
1
0
1
0
0
1
0
1
1
0
0
1
1
0
0
1
roll their own advanced serial
interface. I purchased a CP290
“Start” code
and retired my Z80 controller,
Device code “7”
House code “A”
1110
01010
0110
which was still operating after
many years of service. More
Figure 2—The data format of an X10 command. A command requires 11 full cycles of the power
than ten years ago, I migrated
line. After the transmission of the start code, the data bits and their complements are transmitted
on successive half cycles of the AC voltage sine wave.
from Windows computers to
29
Code
M
N
O
P
C
D
A
B
E
F
G
H
K
L
I
J
H8
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
H4
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1
H2
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
H1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
Device
13
14
15
16
3
4
1
2
5
6
7
8
11
12
9
10
D16
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
D8
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
D4
0
0
0
1
0
1
1
1
0
0
0
0
1
1
1
1
D2
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
D1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
Table 1—The house codes and device codes are listed in order of ascending binary values.
!
New
$99
OSD-232+
RS-232/TTL controlled on-screen
composite video character and graphic
overlay in a small 28 pin dip package.
Intuitive Circuits
www.icircuits.com
(248) 588-4400
30
Linux—first Red Hat, then Gentoo,
and now Ubuntu. I happily migrated
nearly all my tools to Linux, but I
still needed to keep a DOS diskette
with the CP290 control program on
it. I dutifully booted from this
diskette once each month to update
the CP290 event table with new
times.
Several years ago, my CP290 started
to do strange things. It started to turn
on lights at odd times. A dump of the
on-board memory showed that it was
corrupted, a fact that I attributed to
an exhausted back-up battery, which I
replaced. The problems were solved
for a few months, but they reappeared. The back-up battery was again
exhausted, which led me to suspect
that the CP290 power supply wasn’t
putting out its rated voltage. As is
usual for equipment that’s been operating continuously for many years,
the electrolytic capacitor in the power
supply had gone bad. It was a simple
fix, but I started thinking of how I
would use today’s technology to build
another X10 controller.
PL513 & EMBEDDED MCUs
Controlling X10 devices is simpler
now, since you don’t need to build
your own carrier-current transmitter.
The PL513, shown in Photo 2, is an
X10 carrier-current transmitter module that’s available from many sources
for less than $20. The PL513 has circuitry that converts power line zero
crossings to a logic signal, and it
accepts a logic signal to activate a
120-kHz carrier-current transmitter.
Using the PL513 lets you safely
interface with the AC power line. It’s
a well-designed circuit that has optical isolation between itself and your
circuitry. One of its claimed benefits
is that it enables you to use the X10
system commercially without
infringing on any of the X10 patents.
This may have been true in the past,
but the fundamental X10 patents
(4,200,862, 4,628,440, and 4,638,299)
have now expired.
However, it would be
Command
D16
D8
D4
D2
D1
hard to duplicate the
All devices off
1
0
0
0
0
PL513 for $20, so
Dim
1
0
0
1
0
that’s not recommendOn
1
0
1
0
0
ed. The only problem
All lights off
1
0
1
1
0
with the PL513 is that
All lights on
1
1
0
0
0
it’s supplied without a
Bright
1
1
0
1
0
mating connector or
Off
1
1
1
0
0
cable, but more on
that later. There is a
Table 2 —The command codes are listed in order of ascending
similar X10 module,
binary value.
CIRCUIT CELLAR®
•
appealing to hobbyists because of the
availability of free software tools and
the fact that these tools will run
under Linux. I’ve been using the PIC
microcontrollers with the PICBasic
Pro compiler for many years, so I used
what was familiar. PICBasic Pro has
built-in statements for X10 control,
which makes for easy programming.
CONTROLLER HARDWARE
Photo 2 —The PL513 X10 carrier-current transmitter module
the TW523, which both sends and
receives X10 signals. This would be
useful for an alarm-type system, but
it wasn’t needed in my home system.
It might be worthwhile buying the
TW523 instead of the PL513, since it
contains the PL513 circuitry and it
would allow a future alarm system
feature. The TW523 is just a little
more expensive.
Today’s designer has a choice of
many microcontrollers, most of
which are more than adequate for a
simple task such as an X10 controller. The Arduino system, which
uses an Atmel chip, is especially
Figure 3 and Figure 4 are schematics
of the X10 controller. A PCB layout
and parts placement guide are available online. Since I often etch my own
circuit boards, this board is laid out as
single-sided with a few necessary
jumper wires. Through-hole parts are
used, since surface-mounted components don’t make sense for a one-off
board built by a hobbyist. The
Microchip Technology PIC16F877
microcontroller I used has a built-in
hardware USART for the serial interface, another feature that makes the
programming easier. The PIC16F877 is
a 40-pin DIP package, which might
appear overkill for this application.
EMBEDDED
SERVER
Standard SIB
(Server-In-a-Box)
Starting at $305
Quantity 1.
Since 1985
2.6 KERNEL
! Power Supply Included
! Locked Compact Flash Access
! Analog SVGA 3D Video
! Optional Wireless LAN
! EMAC Linux 2.6 Kernel
! XP Embedded & WinCE 6.0
www.emacinc.com/servers/standard_sib.htm
OVER
25
YEARS OF
SINGLE BOARD
SOLUTIONS
Phone: (618) 529-4525 · Fax: (618) 457-0110 · www.emacinc.com
•
CIRCUIT CELLAR®
! Fanless x86 500MHz/1GHz CPU
! 512MB/1GB DDR2 RAM On Board
! 4GB Compact Flash Disk
! 10/100 Base-T Ethernet
! Reliable (No CPU Fan or Disk Drive)
! Two RS-232 Ports
! Four USB 2.0 Ports
! Audio In / Out
! Dimensions: 4.9 x 4.7 x 1.7” (125 x 120 x 44mm)
31
Figure 3 —This is a schematic of the X10 controller showing the PIC16F877 microcontroller and most of the peripheral circuitry.
I had a few of them in my surplus
parts drawer—a design decision
known to all designers.
The PIC16F877A is slightly different from the PIC16F877. It has a
comparator option for some input
pins, but this feature is disabled at
power up to make it code-compatible
with the PIC16F877. I used a 20-MHz
chip, since high clock speeds are
Aside from its hardware USART feature, it has other redeeming features.
There are its 256 bytes of EEPROM
for storing events, a large data and
code memory, and, most importantly,
32
CIRCUIT CELLAR®
•
always good when doing serial communication. The clock signal in this
design is derived from an oscillator,
not a crystal or ceramic resonator,
but those can be used instead since
the precision of the timebase is not
that important. The processor configuration data needs to specify the
timebase type when programming
the device. There’s an in-circuit serial programming (ICSP) connector on
board, which is nearly essential for
program development when high-pincount chips are used. Repeatedly
removing such chips to program
them in an external programmer
would be possible only if a zeroinsertion-force (ZIF) socket is used.
Such sockets are expensive and they
take up a lot of board space.
Every controller for real-time
events needs a real-time clock. The
PIC16F877 has enough horsepower to
act as its own real-time clock, and the
AC power line zero-crossing signal
from the PL513 would make an excellent timebase. However, battery backup for timekeeping during power outages would be a problem, so I chose a
standalone real-time clock chip, the
Maxim DS1307. The DS1307 is inexpensive, it communicates using an I2C
serial connection, and PICBasic Pro
has built-in I2C commands. The clock
chip handles its own battery back-up
with a 3-V lithium cell. The serial
data link is buffered by the usual
MAX232 chip that provides the bipolar voltage signals that an RS-232
serial port expects. Of course, many
computers now eschew such serial
ports, but there are many USB-to-RS232 converters on the market that
2
3
solve this problem.
I use a few transistors for driving
external LEDs and the PL513. The
PL513 driver circuitry is similar to that
4
Figure 5—The connections to the PL513 X10
carrier-current transmitter module
•
CIRCUIT CELLAR®
1
Figure 4—This schematic of the X10 controller shows the MAX232 serial interface
and the voltage regulator. Although the voltage regulator accepts a 12-V input, a 9-V
input is acceptable, and it allows for greater
energy efficiency. The controller draws about
35 mA in its quiescent state, about 70 mA
when an X10 command is being sent, and a
maximum current of less than 100 mA.
recommended by the manufacturer. I
made a small change to preserve signal polarity on the output driver to
maintain compatibility with the
PICBasic Pro X10 commands. The
output driver as published by the
manufacturer inverts the data signal.
The drivers have diode protection circuitry to prevent damage from voltage spikes. It is possible to drive
LEDs directly from the PIC pins, but
I never like that approach. I like
bright LEDs, which translates to a lot
of drive current, and a microcontroller shouldn’t be too abused. One
LED is used to indicate whenever an
X10 signal is being transmitted, and
the other indicates that the back-up
battery is still good. Either or both of
these LEDs and their associated components can be eliminated at no real
loss of functionality, although there’s
another function of the transmit
LED. If the connection from the controller to the PL513 is broken, the
LED will remain lit when a command is attempted but no power line
zero crossings are detected. One
33
advantage of the PIC microcontroller is the availability of
analog-to-digital conversion on some pins. For the
PIC16F877, this is a 10-bit conversion that has plenty of
resolution for most applications.
The PICBasic Pro compiler has commands that simplify
handling analog input. Since I had an overabundance of
input pins, I decided to use one of them to monitor the
back-up battery voltage. The Maxim datasheet specifies a
nominal 3-V lithium cell and recommends a backup battery voltage of between 2.0 and 3.5 V. It would be nice to
continually monitor the battery voltage to give an indication of when the backup battery is reaching its end of life,
but there’s a problem here. Maxim says that a back-up battery will last for 10 years, so attempts to monitor the battery voltage will actually reduce the battery lifetime, even
if the monitor impedance is several megohms! I’ve taken
the approach of using a removable jumper to one of the
PIC analog inputs so the battery can be checked occasionally without a voltmeter. In normal operation, the jumper
is removed so that the battery is not unnecessarily
drained.
A front panel push button can be used instead of the
jumper, and a push button with an integrated LED would
be ideal. Because of the availability of many input pins, I
used jumpers to set the house code. The house code also
can be set with software at the loss of one event location
in the event table in EEPROM. However, the house code
is rarely changed once it’s set, so the jumper approach is
not a problem. One house code allows control of 16
devices. If you have a large house that needs more devices,
you could modify the software to include different house
codes in the event table, but the number of possible
events in the table would be reduced because of the extra
storage requirement.
Photo 1 is the finished circuit. The ICSP connector is
visible at the bottom of the board. The power source is a
regulated 9-V wall transformer. It’s one that stands vertically, so it can be plugged into the same outlet as the
PL513. You could probably build the PL513 and power
supply into your controller box so it can use a single plug
connection to a wall outlet.
CONNECTIONS
34
The connector on the PL513 mates with those old telephone and modem cables you have at the bottom of a
drawer somewhere, but they’re not going to work.
Although the connectors have four pins, inexpensive telephone cables typically contain just two or three wires. The
PL513 requires four wires. I found that handset cords, the
coiled cords that connect old-style telephone handsets to
the telephone body, use all four wires.
Telephone cables purchased at home improvement stores
usually have four wires, and I used one of these for connection to the PL513. The mating jack for the controller side is
available at nearly every home improvement store.
Figure 5 shows the PL513 connections. The TW523 connections are the same, except that pin 3 is an open-collector
output for received signals. If you aren’t using this signal,
this pin can be grounded to maintain compatibility with the
PL513 connection at the controller.
THE PROGRAM AHEAD
In Part 2 of this article series, I’ll review the controller
program and a utility I wrote to convert a human-readable
event file to the serial data the X10 controller understands.
The utility is written in ANSI C, so it compiles and executes under both Linux and Windows. Executables will be
provided for those of you without compilers. I
Devlin Gualtieri received a PhD in Solid-State Science
from Syracuse University in 1974 and worked for many
years doing research for a major aerospace company. Now
retired, he spends his time doing various embedded systems projects. He can be reached at [email protected].
PROJECT FILES
To download the code, go to ftp://ftp.circuitcellar.com/
pub/Circuit_Cellar/2010/240.
SOURCES
PIC16F877 Microcontroller
Microchip Technology, Inc. | www.microchip.com
PL513 X10 Module
X-10 (USA), Inc. | www.x10.com
NEED-TO-KNOW INFO
covered in Devlin Gualtieri’s Issue 240 article, the
Circuit Cellar editorial staff highly recommends
the following content:
—
XTstat
A Portable Wireless X10 Thermostat
by Henry Jacobs
Henry’s XTstat wireless thermostat allows you to
adjust a room’s temperature. The PIC-based device is a
great starting point for wireless temperature control
projects. Topics: X10, Wireless Communication, RF
—
Radio Riot
by Tom Cantrell
Going wireless doesn’t need to be difficult. Tom
has the scoop on several essential wireless solutions.
Topics: Wireless, RFID, ZigBee, Bluetooth, IEEE
802.15.4, Spread Spectrum
CIRCUIT CELLAR®
•
UNIQUE PRODUCTS & SUPPORT
www.saelig.com
event-time or pulse-counting data
Standalone USB temp / hum / volt /
current loop data logger.
$57+
FTDI USB ICs
CAN-USB
FT232RL
USBI2CIO
L
Pr owe
ic st
es
USB to I2C
Mini-logger with built-in temp/hum/
pressure/3-axis accel sensors.
without a PC! No drivers needed.
Versitile Function/Arbitrary/Pulse
Generator offering precise waveform
control demanded by engineers.
Serial-Ethernet Cable
Lorlin Switches
Connect any SD slot (phones/PDAs)
to a USB memory stick. Send files
Network serial product easily without
Keyboard Simulator
Instant Ethernet
Ethernet-IO
FPGA Systems
FEZ Domino / FEZ Mini
Tiny open source boards using
USBizi chipset:
Kits $98
Compact, economical smart OLED with
graphics drive from USB or RS232.
CGR-101
www.saelig.com
Mention offer# SBW
Sound Module
Temp/RH Sensors
RTG005
Ether-IO 24
transducers with integral electronics.
Wireless Solutions
Analog input, bluetooth wireless
modules 433/868/915MHz.
RF Modules
RS232 to 422/485
9p-9p or 25p-25p self-pwrd,
isolated RS232-RS422/485
non-contact SAW digital rotary torque
14-pin module plays back prestored audio files from microSD card
designs for beginners and experts!
Simultaneously transmit composite
video and stereo audio signals.
Configurable, patented USB-output
Fantastic array of stock and
custom switching devices.
Novel ambient sensors & modules
accurately measure temp/RH.
USB-Serial
1/2/4/8/16 x RS232
KK Systems
Easy OLED Display
$89
Ready-to-go out-of-the-box FPGA/DSP
AWM6XX TX/RX
.NET Board
a PC using this 28” cable.
UDP/IP-controlled 24 digital
I/O board 3 x 8-bit TTL ports.
uOLED-96-G1
No OS needed. TCP/IP offload,
ICs improve system performance.
WIZ110SR / W5100
Intelligent CAN connection
from PC’s USB port. $299
PoKeys55T
Popular UART and FIFO chips.
Upgrade Legacy designs to USB.
10-bit A/D inputs, 1 x 10-bit analog O/P.
PS9200
TorqSense
RF Testing / EMI Tents
50MHz AWG
RWT320
SD to USB Adapter
“Drop-in” solution connects PC to
I2C/SMBUS + 32 I/O lines. $89
USB board adds 55 I/O and 5 x
APSIN6000
Portable RF test enclosures &
shielding tents with external frame.
SOMO-14D
MSR145S
RTR-50
Log and display temp, hum, volt,
High-res, extremely low-noise,
portable 6GHz RF generator.
Add 1-16 COMports via your
PC’s USB Port easily.
UPSICAP / DLP-TH1
USB Loggers
Multiparameter Loggers
CANbus, Ehternet, USB2.0, CF.
Embedded controller series: 2 x
CE-USB
Kits turn your PC into vehicleelectrics diagnostic tool.
!
exerciser and analyzer.
RF Testing/EMI Tents
TG5011
Versatile USB 2.0 I2C protocol
or arbitrary waveform generator.
RF Generator
EW
I2C Xpress
Wave Xpress
USB2.0 speed 16-bit digital pattern
CANminiBOX
2-ch 12GHz sampling scope for
high-speed electrical signals.
N
Automotive Testing
Handheld Palm PC-based
2.7GHz Spectrum Analyzer.
Lorlin
I2C Xpress
RF & EMF Spectrum Analyzer
1Hz to 7GHz for WiFi, mikes, etc.
While supplies last
- not available with
any other offers
Worlds’s Fastest
EmbedRF / Adeunis
Protocol exerciser/analyzer for standard
SPI and non-standard 4-wire and 3-wire
serial protocol interfaces up to 50 Mbps.
to get a free Starbucks
Card with your >$50 order!
EMC Spectrum Analyzer
emPC-x133
Intuitive full-featured 16-ch 4MB
200MHz sampling memory. $240
12/16 input 1kS/s 10/12-bit
PC-connected voltage logger.
PL1012 / PL1216
EMC Spectrum Analyzer
FREE COFFEE
Call 1-888-772-3544
PSA2701T
SPI Bus Analyzer
SPI Xpress
16-Ch Logic Analyzer
Waveform Generator
Wireless Data Loggers
CS328
Mephisto
Analyzer and Signal Generator. $1259+
Compact compressed-serial output
camera module for any host system.
LAD-16128U
100MHz Scope, + Spectrum/Logic
Alan Lowne
Saelig CEO
Multichannel DAQ
eCOV-110-P
$699 +
2-ch DSO 16-bit DSO, FFT, VM,
logic analyzer, standalone + 24 I/O.
PS3423 / KLARI-MOD
analyzers and generators.
microCAM
EL-USB-1/2/3/4
Packet-Master™ - USB 1.1/2.0
Ethernet/3xUSB/SD.
Mixed-Signal Scopes
PS2203/4/5
B
Va es
lu t
e
USB Bus Analyzers
USB12 / 480+ / 500AG
2-ch 40/100/200MS/s 8-bit scope
range with 5/10/25MHz. $246 +
Windows PC. LX800/512MB/4GB/
“I really like this scope adapter
- it’s meant for teaching electronic
experiments but it’s ideal for
engineers too.”
USB-COM
probe! Up to 100MS/s. $193 / $280
Touch-input 10.2” LCD 12V-powered
uCAM-232 / uCAM-TTL
10/25MHz USB powered scope-in-a-
EMC RF & EMC Spectrum
XP Emb Touchpanel
$180
CircuitGear CGR-101™ is a unique
new, low-cost PC-based instrument
which provides the features of seven
devices in one USB-powered compact box:
2-ch 10-bit 20MS/sec 2MHz oscilloscope,
2-ch spectrum-analyzer, 3MHz 8-bit
arbitrary-waveform/standard-function
generator with 8 digital I/O lines. It also
functions as a Network Analyzer, a
Noise Generator and a PWM Output
source. What’s more – its’ open-source
software runs with Windows, Linux and
Mac OS’s!
Only $199!
Mobidapter
Pen Scopes
CUPC-80
25MHz 2-ch /16 logic scope
and logic analyzer.
$1095
Scope/Logger
Low-Cost Scopes
DS1052D
DS1000E
N
RI EW
GO !
L
50/100 MHz scope. $595 / $795
2-ch 1GSa/s (25GSa/s equiv.)
Amazing 7 in 1 Scope!
CAN-USB
20MHz / 60MHz rugged handheld
USB 2-ch scope.
$593 / $699
Scope + Analyzer
PS2104/PS2105
Handheld Scopes
PDS5022S / PDS6062T
N
RI EW
GO !
L
2-ch + trigger standalone USB
bench scope.
$287 / $419
2-ch 1GSa/s Scopes
HDS1022MN / HDS2062M
Se Bes
lle t
r
Color LCD Scopes
A complete CP2102 USB-serial
converter in a DB9 shell. $26
Above are some of our best-selling, unique, time-saving products - see our website for 100s more:
WiFi/910MHz antennas, wireless boards, LCD display kits, Ethernet/IO, USB/RS232/485, USB-OTG,
instant Ethernet-serial, CAN/LINbus, USB cables/extenders, line testers, logic analyzers, color sensors,
motion controllers, eng. software, wireless boards, SMD adapters, I2C adapters, GPS loggers,
automotive testing, security dongles, video motion detectors, crystals/oscillators, custom switches,
barcode scanners, DSP filters, PLCs, Remote MP3 players, etc. FREE Starbucks card with your $50 order!
Check www.saelig.com often for special offers, bargains, business hints, blog, etc.
F EATURE
ARTICLE
by Tom Struzik (USA)
Wireless Data Exchange
Build a 2,700-lb. Bluetooth Headset
Building a Bluetooth-enabled device can be difficult if you don’t have any
experience with the technology. Finding the right module can greatly simplify
the process. The Bluegiga WT32 is a highly integrated Bluetooth module that
fits the bill. Here you learn how to put Bluetooth to good use.
E
36
nvy is a strong motivator. My wife bought a new
Toyota Prius and it had one exceptionally “geekworthy” feature: a built-in Bluetooth interface. As if by
magic, when you start the car, it automatically connects
to her Palm Treo. She answers incoming calls by simply
pressing a button on the steering wheel. The audio,
which comes in over the car’s stereo speakers, automatically mutes whatever is playing when a call is received.
All in all, it’s a slick, simple interface.
Unfortunately, instead of a Prius, I have an old (2,700-lb.
tare weight) Acura, and its only upgrade is a radio with
an auxiliary input. My thought was that I should be able
to build a Bluetooth-to-aux-in adapter that functions
much like the Prius’s Bluetooth interface. However, I figured I should be able to support playing MP3s from my
smart phone with a custom adapter. Plus, if I could
access the phone’s voice command functionality, I
thought I would really “out geek” the Prius!
Thus, I started my project to construct a “2,700-lb.
Bluetooth headset.” I had three goals for the project. The
first was to learn something about Bluetooth. The second
was to find a Bluetooth module I could use for this Bluetooth project and others. My third goal was to one-up the
Prius’s Bluetooth functionality!
to be a very DIY-friendly shop. They provide breakout
boards for their SMD parts, manuals, user guides, and
even a support forum. They’re also one of the few parts
sources that offer low-cost U.S. Postal Service shipping,
so you don’t have to pay a $10 shipping fee for a $2 part!
So, off to SparkFun I went, and they had several different
Bluetooth modules in stock (all with breakout boards).
FINDING A MODULE
Fi gure 1—The Bluegiga WT32 is a highly integrated device. In addition to the radio, it contains multiple digital and analog I/O sources,
a programmable MCU, and a DSP.
When I look for small-quantity parts, I tend to start
searching at SparkFun (www.sparkfun.com). I find them
USART/USB
Flash
PIO
RAM
Antenna
2.4GHz
Radio
I/O
DSP
Audio in/out
MCU
PCM, I2S, SPDIF
XTAL
RESET
SPI
CIRCUIT CELLAR®
•
of the WT32 datasheet, the iWRAP
user guide, and the circuit diagram for
the WT32 breakout board. With this
in hand, I thought I should be able to
power up and begin interacting with
the WT32 device.
One thing I noticed while working
on powering up the WT32 was that
the SparkFun breakout board has differences in the pin nomenclature and
c)
d)
wiring between it and the actual
WT32 device. For example, the breakout board has VDDIO and VDD_BAT
wired together and AGND and DGND
have a solder jumper.
With power to the board, the next
step was to gain access to the UART.
The UART has CMOS input/outputs,
so a MAX3232 line driver chip was
required for interfacing to my PC serial port. With the UART connected, I
confirmed the basics were working by
Photo 1a—A WT32 module without its cover. The heart of the WT32 is a Cambridge Silicon
running HyperTerminal on my PC
Radio BlueCore 5 Multimedia chip. b—The underside of the WT32 module. Don’t place bare
copper on your board underneath the WT32! See the WT32 datasheet for board layout recand connecting at 115,200 bps, 8N1 to
ommendations. c—The SparkFun WT32 breakout board with the cover intact on its WT32. I
the WT32 UART interface. When I
added the legs on the breakout board in order to facilitate easy breadboard experimentation.
applied power to the WT32, HyperTerd—This is the breadboard with the breakout board on the left.
minal displayed the iWRAP boot
screen (see Figure 2). Success!
Now, with a live connection to the
Only one, however, the Bluegiga WT32,
build a complete Bluetooth applicaWT32’s iWRAP firmware, the fun
supported anything more than simple tion using nothing more than the
really began. The next logical step was
WT32 and its UART or USB inter“serial cable replacement.”
to create an hands-free profile (HFP)
face. It appeared that the Bluegiga
Further investigation revealed that
link between the WT32 and my smart
WT32 was an almost-perfect candithe WT32 was quite a find. Per the
phone. With this link, I could explore
date for my project.
datasheet, it has one of just about
the basics of Bluetooth link setup and
Figure 1 and Photo 1 show the
everything. On the analog audio side,
how the iWRAP firmware operated.
WT32 module’s details. Photo 1c
it supports stereo input and output
The iWRAP user guide was a bit light
with differential signal levels. On the shows the SparkFun breakout board.
on how to actually set up an HFP link,
digital audio side, it has an integratso I went over to the Bluegiga website
ed DSP with stereo codec and supGETTING STARTED
and signed up for their tech support
port for multiple digital audio forMy initial goal for working with the
forum. This move turned out to be
mats (PCM, I2S, and SPDIF). The digWT32 was to breadboard the simplest
excellent, as the tech support forum
possible configuration and validate
ital I/O interface has support for
had not only firmware updates and PC
that I could indeed make a Bluetooth
USB, RS-232, and SPI—and on top of
tools, but also WT32 application notes
connection to my phone. I also felt
that, 10 individually programmable
like the “Hands-Free and Headset Prothat I needed to gain some experience
I/O pins. Most importantly, in the
files” iWRAP application note. From
with how the WT32, iWRAP, and
area of Bluetooth, it supports multithis, there were two main things
ple Bluetooth profiles (A2DP, AVRCP, phone interact.
I started by reviewing the documenHFP, HFP-AG, SPP, OPP, and HID)
required to configure an HFP link. The
tation that SparkFun had available on
and has an integrated antenna, giving
first was to enable the required Blueits website. This information consisted tooth profile. The second was to define
the device a 30-m range. And if all
that wasn’t enough, throw in a
the Class-of-Device (CoD).
built-in LiPo battery charger!
Enabling and disabling profiles
WRAP THOR AI (3.0.0 build 165)
On top of this massive hardwas straightforward. The
Copyright (c) 2003-2008 Bluegiga Technologies Inc.
ware feature list, Bluegiga
iWRAP command was SET
READY.
loaded the WT32 with its
PROFILE {profile_name}
iWRAP firmware. The program[SDP_name]. Note that leaving
Figure 2—When I applied power to the WT32, HyperTermiming language provided by this
out the SDP_name disables the
nal displayed the iWRAP boot screen. Success! (Source:
Bluegiga Technologies)
firmware makes it possible to
specified profile. For my simple
•
CIRCUIT CELLAR®
b)
a)
37
READY.
HFP 1
CONNECT 0 HFP 2 00:17:e8:3f:d9:e2
<= Initial Autocall connect
<phone moves out of range>
NO CARRIER 0 ERROR 305 SDC_OPEN_SEARCH_FAILED_PAGE_TIMEOUT
NO CARRIER 0 ERROR 305 SDC_OPEN_SEARCH_FAILED_PAGE_TIMEOUT
<phone moves into range>
<= Autocall reconnect
Figure 3—After rebooting the WT32 or phone, I saw the connection automatically reestablished. (Source: Bluegiga Technologies)
38
would need to use a different CoD. For
HFP test, it appeared a “SET PROFILE
that configuration, I used the generaHFP ON” would work nicely. I knew
tor to produce the CoD 200418 {Major
that later I’d need to enable multiple
Service: Audio, Major Device:
profiles in order to complete my projAudio/Video, Minor Device: Handsect; but for the time being, I only
free and Headphones}.
needed to enable the HFP profile.
With the basic HFP profile configThe more cryptic part of the appliured and put into the iWRAP
cation note was setting the WT32’s
firmware, I cycled the power on the
CoD attribute. The application note
WT32 and initiated a pairing from my
said to use 200408 with no explanaphone. During the pairing process, I
tion what the number actually meant.
noticed that changing the CoD altered
I took a trip to www.bluetooth.com
how the WT32 appeared to the phone.
and the official specification docuDepending on the CoD, the phone
ments hosted there to turn up inforwould use different icons to represent
mation. CoD is used during the Bluethe WT32 device. Once the phone and
tooth device discovery process and is
WT32 were paired, I called my phone
how the queried device responds with
from another number and was able to
its device type and available services.
In the Bluetooth “Assigned Numbers – see the incoming call information
through the iWRAP interface. At this
Baseband” document, CoD is defined
point I believed I had confirmed that
as three octets consisting of Major
everything was working properly. That
Service class (11 bits), Major Device
was, until I power cycled the WT32. I
class (5 bits), and Minor Device class
could still see that the devices were
(6 bits). Even the various bit definipaired, but they were not connected.
tions for the fields were included in
this document.
Another clue came from the
iWRAP user manual. It referenced a
ATtiny26L
Bluetooth CoD generator webpage.
The CoD generator let me simply
PC
PBTNS
check boxes for services and classes
Term
and then it displayed the resulting
Mic
CoD number. From this informaWT32
Car
tion, I determined the CoD in the
stereo
application note—200408—actually
Differential
maps to Major Service: Audio;
Major Device: Audio/Video; and
Op-amp
AUX-IN
Minor Device: Hands-Free device.
Single-ended
Armed with this knowledge, I
could now issue the required “SET
Figure 4—A more complex solution required using
BT CLASS 200408” command and
an AVR processor to control the WT32 through the
at least understand what it was
iWRAP interface. The PC can still be connected to
doing.
the system for troubleshooting. The op-amp is
For my final configuration using
used to convert the audio-out signal from differential to single-ended mode.
both the HFP and A2DP profiles, I
The phone didn’t show an active headset icon, and incoming calls no longer
displayed information in iWRAP.
After more digging in the iWRAP
user guide, it became apparent that
there had to be a CALL command to
open the Bluetooth link between the
phone and the WT32. Evidently, the
act of pairing just happened to also
make the HFP link active, but it was
not a permanently active link. In fact,
even after a CALL was issued, if either
the phone or the WT32 was rebooted
or moved out of range, the Bluetooth
link dropped and was not automatically reestablished.
I finally found a solution in the
iWRAP user guide: the iWRAP
AUTOCALL command. Per the documentation, AUTOCALL will automatically open the specified link to the
first paired device it finds and will
even attempt to reopen the link if the
link is dropped. This sounded like just
what I needed. The format of the
AUTOCALL command was SET
CONTROL AUTOCALL {target}
{timeout} {profile}, which was all relatively obvious, except for the {target}
field that was specified as “the UUID
of the target channel.” Both the
iWRAP user guide and the Bluetooth
“Assigned Numbers – Service Discovery” document stated that the UUID
for the HFP-AG channel was 111F
hex. The HFP-AG channel UUID
must be used rather than the HFP
channel UUID since it’s the handset
making the call to the gateway.
Thus, a “SET CONTROL AUTOCALL 111F 5000 HFP” appeared to
be the answer. Sure enough, after
rebooting either the WT32 or the
phone, I could see the connection
being automatically re-established
(see Figure 3).
As a side note, it turned out that
the AUTOCALL function was a bit
of a double-edged sword. It provided
the connect/reconnect functionality;
however, it always tried to connect.
When working manually with
iWRAP, if the phone was unavailable, the reconnect attempts disrupted any typing into the iWRAP
interface. I found that it was sometimes easier to quickly key in a
“SET CONTROL AUTOCALL”
CIRCUIT CELLAR®
•
24th International Trade Fair
New Munich Trade Fair Centre
09–12 November 2010
erolpxe
the possibilities of tomorrow.
Automotive
e-Mobility
Displays / e-Signage
Embedded systems / software
Medical / MEMS
Photovoltaics
Time for electronics. Time for the future.
Key topics, trends and technologies. The latest components, systems and applications.
Visit electronica 2010, the international trade fair that will show you today what is
important tomorrow and generate momentum for real growth.
Parallel event: hybridica. Trade fair for hybrid-component production. www.hybridica.de
get the whole picture
electronica 2010
components | systems | applications
www.electronica.de/en
Fi gure 5—What do you think of this solution? An ATtiny26 AVR controls the WT32.
immediately after the WT32 booted
in order to disable the AUTOCALL
function.
At this point, I had verified that I
could successfully connect the WT32
to my phone. However, the ultimate
goal of one-upping the Prius required
establishing A2DP and AVRCP links
in addition to the HFP link. The need
for multiple links quickly became
another problem. The AUTOCALL
function only supported making one
CALL, and I needed to make three.
The only way to establish the additional links was to manually issue the
multiple CALL commands.
THE SIMPLEST OPTION
40
After digging through the documentation and on-line forums, it appeared
there were potentially several solutions
to the multicall problem. The simplest
solution relied on a “trick” using the
programmable GPIO pins and the
iWRAP firmware command SET CONTROL BIND. Together, these functions
would allow me to automatically trigger additional iWRAP commands based
on GPIO pin state transitions. Better
still, by specifying different priorities,
multiple BIND commands could be
made for the same pin transition.
The other critical iWRAP command
to make this option work was the SET
CONTROL CD command, which ties
a specified GPIO pin to the WT32’s
Carrier Detect state. So, in theory, this
solution would work thusly: AUTOCALL would make (and maintain) the
State
Powerup
Call Setup
In Call
Idle
Streaming
Button #1
Reset
Answer
Mute
Play
Pause
initial HFP link. When this link was
made, the WT32 Carrier Detect state
changes and the SET CONTROL CD
raises a GPIO pin. This action then
triggers multiple SET CONTROL
BIND commands, one for each
required CALL. Even if the link was
broken, the CD signal would be deasserted. When AUTOCALL reestablished the HFP link, the CD pin
would be reasserted and the BIND
commands would then (re)CALL the
other links as well. It seemed like a
simple, elegant solution until I actually tried to implement it. Several
Button #1 held
Button #2
Button #3
Reject
Hangup
Voice Cmd
Voice Cmd
Prev Track
Next Track
Table 1—This is how I mapped the push button functions to the various states.
CIRCUIT CELLAR®
•
<= iWRAP begins booting
READY.
<= iWRAP completes boot
HFP 1
<= iWRAP invokes the autocall
<= iWRAP connect event
call 00:17:e8:3f:d9:e2 17 a2dp
<= AVR placing the AVRCP call
CALL 1
CONNECT 1 A2DP 23 00:17:e8:3f:d9:e2
<= iWRAP connect event
HFP 0 STATUS "service" 1
<= iWRAP continuation
HFP 0 STATUS "call" 0
<=
of the Hands Free
HFP 0 STATUS "callsetup" 0
<=
profile startup
HFP 0 STATUS "callheld" 0
HFP 0 STATUS "signal" 5
HFP 0 STATUS "roam" 0
HFP 0 STATUS "battchg" 5
call 00:17:e8:3f:d9:e2 19 a2dp
<= AVR placing the A2DP call
CALL 2
HFP 0 READY
<= iWRAP HFP status continues
HFP 0 UNKNOWN (7): \r\n+COPS: 0,0,"AT&T"\r\n
<= iWRAP connect for left
<=
and right channel
Figure 7—The exchange looked like this from the iWRAP interface. (Source: Bluegiga Technologies)
•
so in the end I could not guarantee
which link ID would end up going
with which profile.
As the final nail in the coffin, on
the simple option, with a marginal
signal, some links would drop while
the others remained active. If the
HFP link wasn’t the one that
dropped, the other links never would
be reestablished.
BETTER, BUT COMPLEX, OPTION
While the SET CONTROL BIND
command could be extremely useful
for a single-profile device, in my case
it was not enough. I needed to investigate other solutions. Thus, the second
solution (see Figure 4) was born,
which consisted of using an Atmel
ATtiny26 AVR processor to control
the WT32 directly through the iWRAP
interface (see Figure 5). The concept
was that the AVR would monitor the
WT32 state through event messages
displayed in the iWRAP interface and
it would monitor the state of several
push buttons. The AVR could then
issue the necessary iWRAP commands
to manage the links and issue the
commands to activate the various profile functions such as ANSWER,
HANGUP, and PLAY.
There were several advantages to
this solution. For one, it would be
easy to simulate and troubleshoot the
system by using HyperTerminal on
the PC connected to the iWRAP
UART interface. The AVR could
remap push-button functions based on
the current WT32 state, which would
have been impossible to do using the
simple option.
There were three main things the
AVR needed to track in this option.
The first was tracking and managing
the state of the links. The second was
tracking the phone state (the HFP profile). The third was tracking the
streaming audio state (the A2DP profile). With knowledge of the states, the
AVR could map the push buttons to
the iWRAP commands necessary to
control the HFP and the A2DP functions. Refer to Table 1 to see how I
mapped the push-button functions to
the various states.
I started with building a simple
state diagram to outline what the AVR
CIRCUIT CELLAR®
course change any
time a different
Link state management
Power-up
phone was paired
I: “READY”
with the WT32,
and changing the
Button#1 held
Ready
address would
Ready
not connected
connected
require using the
I: “CONNECT 0 HFP 2”
iWRAP interface to
Pending
reset
update the address
I: “PAIR”
string stored in the
I: “READY”
I:
“CONNECT
3”
CALL command
Call
A VRCP
Issue
that was in each
reset
SET CONTROL
BIND statement.
I: “CONNECT 1”
I: “ERROR 409”
While this might
I: “SYNTAX ERROR”
have been fine for a
one-off device, I
*Any*
Call
A2DP
wanted to produce
a more general-purpose solution.
Figure 6—The AVR manages the state of the HFP, A2DP, and
Second, once mulAVRCP links. After the HFP Autocall, the AVR makes the other two
tiple
links are active,
calls. Error-handling is brute-force at this point and just resets the
device. I:“<string>” represents the iWRAP text that the AVR watches profile commands
for to drive the state transitions.
like ANSWER,
HANGUP, or
additional problems arose during
AVRCP PLAY need to specify which
implementation.
link ID should receive the command.
First, unlike the AUTOCALL comThe problem resulted in the link ID
mand, the CALL command required
appearing to be nondeterministic. It
that I specify the Bluetooth address of
depended on the order in which the
the destination. The address would of
CALL commands actually completed,
41
42
would need to do. Using HyperTerminal and the iWRAP interface, I
watched the WT32 as it responded to
the various iWRAP events, link states,
and phone actions. From this information, I was able to produce a state diagram (see Figure 6).
For managing the link state, I determined that I would still use the
iWRAP AUTOCALL function to control the HFP link, and then the AVR
could daisy chain off the event messages coming from the AUTOCALL.
From the iWRAP event resulting from
the successful AUTOCALL, the AVR
could also find the necessary Bluetooth
address. The AVR could then issue the
two CALL commands, inserting the
Bluetooth address where necessary, to
make the A2DP and AVRCP links. The
exchange looked like Figure 7 from the
iWRAP interface.
Managing the phone call state was
relatively simple. An incoming phone
call that was answered and then disconnected looked like Figure 8 in iWRAP.
Managing the streaming audio was
even easier. The same AVR push button would control both the PLAY
and STOP commands. The AVR
would monitor the streaming state to
determine which iWRAP command
to issue. From iWRAP, hitting the
Play and then Stop button looked
like Figure 9.
The voice command was simple as
well. Since it’s done completely by the
phone, all the WT32 has to do is send
audio of the voice command to the
phone. The phone does the heavy lifting of recognizing and acting on the
voice command. Refer Figure 10 for the
iWRAP exchange for voice command.
At this stage, I had completed the
design of my “2700 lb Bluetooth Headset” that would even beat the Prius by
providing streaming audio in addition
to regular phone headset capability. I
had also learned some Bluetooth basics
and I found a Bluetooth module, the
Bluegiga WT32, which I could use in
my future Bluetooth projects.
Even though I only brushed the surface of the WT32 module’s capabilities in this article, I hope this whetted your appetite to investigate further. With a little work, you too can
easily add a Bluetooth interface to
HFP 0 STATUS "callsetup" 1 <= incoming call, iWRAP start call setup
HFP 0 RING
<= iWRAP, the phone rings
HFP 0 CALLERID "5551111234" "" 80
<= iWRAP, caller ID
HFP 0 RING
HFP 0 CALLERID "5551111234" "" 80
HFP 0 RING
HFP 0 CALLERID "5551111234" "" 80
< AVR Answer button pressed>
set 0 select
<= the AVR selecting the HFP link
answer
<= the AVR issues answer command
HFP 0 OK
<= iWRAP, successful answer, the call is made
HFP 0 STATUS "callsetup" 0
<= iWRAP, end call setup
RING 4 00:17:e8:3f:d9:e2 SCO
<= iWRAP, incoming Audio link to support the call
< AVR Hang-up button pressed>
set 0 select
<= AVR selects the HFP link
hangup
<= AVR issues hang-up command
HFP 0 OK
<= iWRAP ends calls
NO CARRIER 4 ERROR 113 HCI_ERROR_OETC_USER <= iWRAP says phone dropped sco link
Fi gure 8—An incoming phone call which was answered and then disconnected looked like
this in iWRAP. (Source: Bluegiga Technologies)
your next project.
ADDITIONAL NOTES
Don’t worry if your WT32 from SparkFun initially comes up with a warning
that the firmware is “evaluation only.”
You just need to upgrade the firmware.
You can find the latest firmware on the
Bluegiga technical forum. Just sign up to
get access to the firmware, documentation, PC software, and application notes.
set 1 select
avrcp play
A2DP STREAMING START
set 1 select
avrcp stop
A2DP STREAMING STOP
cable. The circuit diagram for the SPI
interface cable and the necessary software drivers can be found on the
Bluegiga forum.
One word of caution: it is entirely possible to configure non-standard data rates
with the iWRAP SET CONTROL BAUD
command. Verify your SET CONTROL
BAUD command before you hit the
Return key. Make doubly sure you have
not mistyped the data rate parameter!
<AVR Play/Pause button pressed>
<= AVR selects the AVRCP link
<= and issues the play command
<= iWRAP sees streaming start
< AVR Play/Pause button pressed>
<= AVR selects the AVRCP link
<= and issues the stop command
<= iWRAP sees streaming stop
F igure 9 —This is what it looks like when you hit the Play button and then the Stop button.
(Source: Bluegiga Technologies)
The support staff at Bluegiga was
helpful when I had questions or problems that weren’t covered in the
FAQs or forums.
The WT32 firmware can be upgraded
over UART or SPI. However, to use
SPI you need the necessary interface
The normal PC serial port won’t do
non-standard data rates. If you misconfigure the WT32, you won’t able to
communicate with iWRAP via the
serial port. However, it is still possible
to use the SPI and the PSTOOLS program (also available on the Bluegiga
<AVR Voice Cmd button pressed>
set 0 select
<= AVR selects the HFP link
AT+BVRA=1
<= AVR issues voice dial command
RING 4 00:17:e8:3f:d9:e2 SCO <= iWRAP incoming audio link (for voice cmd)
HFP 0 OK
< voice command processing happens on phone>
NO CARRIER 4 ERROR 113 HCI_ERROR_OETC_USER <= iWRAP sees audio link close
Fi gure 1 0—The iWRAP exchange for voice command (Source: Bluegiga Technologies)
CIRCUIT CELLAR®
•
forum) to reset the iWRAP data rate. Your only other alternative is to find a
serial terminal that will perform non-standard data rates. Or you can do as I
did and use a USBee SX signal generator to manually produce the non-standard
RS-232 signals! I
Author’s note: I thank Bluegiga Technologies for granting me permission to show the
header with its copyright statement in some of this article’s figures.
Tom Struzik ([email protected]) has been building and taking things apart from
an early age. He built his first Heathkit project at age 12 and sold his first computer
program at age 16. Tom has a BSEE from Purdue University. He works for a Fortune100 chemical company in its engineering organization as an IT systems architect.
Tom continues to write software and build hardware projects at home to “keep his
hands dirty.” He is currently in a state of wonder over his young daughter. The bets
are on to see if she gets Tom’s engineering gene, her mother’s music gene, or
some combination thereof. Visit Tom at www.JenRathbun.com/Electronics/.
PROJECT FILES
To download a circuit diagram and code, go to ftp://ftp.circuitcellar.com/
pub/Circuit_Cellar/2010/240.
RESOURCES
WT32 User Guides, Application Notes, and More, Bluegiga Tech Forum,
www.bluegiga.com/support.
SOURCES
ATtiny26 Microcontroller and AVR assembler
Atmel Corp. | www.atmel.com
WT32 Bluetooth module
Bluegiga Technologies | www.bluegiga.com
NEED-TO-KNOW INFO
Knowledge is power. In the computer applications industry, informed
engineers and programmers don’t just survive, they thrive and excel.
For more need-to-know information about topics covered in Tom
Struzik’s Issue 240 article, the Circuit Cellar editorial staff highly recommends the following content:
—
Wireless Sensor Network Design
by Diego Méndez Chaves, et al.
A group of designers presents a wireless sensor network platform. It’s a testbed for the evaluation of different algorithms related to the communications
stack of a network (e.g., routing policies and medium access control). Topics:
Wireless, Network, Mote, Kernel, Sensors
—
Wireless Thermostat System
by Darryl Uchitil
Darryl designed a wireless zone thermostat system that can remotely
read temperature, calculate the dew point, and force a setback. Topics:
Wireless, Thermostat, ZigBee
•
CIRCUIT CELLAR®
F EATURE
ARTICLE
by Matt Bunting & John Rogers (USA)
Hexapod Kinematics
Motor Communication, Motion Planning, and Image Processing
This article covers advanced topics in robotics applied to a completely
autonomous hexapod design. The concepts of gait dynamics, feedback
control, and optical flow for motion depiction are covered.
I
n this article, we’ll demonstrate advanced topics in
robotics applied to a uniquely designed hexapod robot
that is a 2-kg completely autonomous unit using an Intel
Atom z530, which is a low-power processor capable of
powerful computation (see Photo 1). The hexapod has the
ability to be completely self-contained while simultaneously combine walking and balancing algorithms, together
with sophisticated optical image-recognition processing.
We’ll describe advanced robotic concepts—such as inverse
kinematics and optic flow—in terms of foot placement
with path optimization and gait dynamics, feedback control
for behavior modification, and optical flow for motion
depiction, all of which are all essential to robotic operation.
PRECISION CONTROL
Hexapods are six-legged walking creatures commonly
found in nature. Having six legs eliminates many of the
complexities involved with maintaining physical balance;
however, the relatively large number of extremities
requires intricate control during movement. But what about
a hexapod robot? With the capabilities of contemporary
a)
b)
44
microprocessors, complex floating-point calculations
involving inverse trigonometric functions can be performed
with an unnoticeable effect on the system’s speed. As such,
inverse kinematic functions that model a hexapod’s legs
provide precision control that enables fluidic movement.
The hexapod robot design involves mechanical engineering, electrical engineering, and computer science. We created the mechanical design with 3-D CAD software and a 3-D
printer at the Robotics and Neural Systems Laboratory
(RNSL) at the University of Arizona. The electronics were
designed purely to pipe control and sensory information to a
computer, utilizing USB for connectivity. The computer utilizes hyper-threading, a method of simultaneous computational capability to increase application performance.
Machine vision is included as the fundamental sensor for
the future integration of various behaviors.
A system model using forward kinematics was fabricated
to predict the end effector locations based on the set
angles of the motors. Inverse kinematics were then
derived to set the angles of the motors based on desired
end effector locations. Using inverse kinematics, trajectories for each foot were created to generate walking gaits.
A dynamic system was then employed to place the feet of
the hexapod based on the body location. Various walking
gaits were possible using the dynamic walking techniques.
Photo 1a—The assembled hexapod
features an Intel Atom processor
and Robotis AX-12, RX-10, and RX28 servomotors. b—The Atom
processor performs kinematics calculations together with complex
vision-processing at nearly 900 Hz.
A Point Grey Chameleon machine
vision camera is used for its compactness. An inexpensive lens is
used to get a 150° field of view.
CIRCUIT CELLAR®
•
•
CIRCUIT CELLAR®
Instead of placing all
apply the greatest torque, the RX-28
six legs circularly symmotors are used. The last motor is
metric about the center
the tibia motor, using the AX-12,
of the body, we used a
with the angle denoted as θt and
simple ellipse. The 3-D
length denoted as dt.
rendering and printing
The Robotis motors offer a mountprocess is shown in
ing point on the same axis of the
Photo 2.
output shaft, which can be used for
Each leg was designed rigid joints. Bearings are fixed on this
with three degrees of
mounting point. Each link, for the
freedom (3DOF). In a 3coxa, femur and tibia, was designed
D space, 3DOF are
to wrap around the motor.
needed to position the
foot at an arbitrary
COMPUTATION
Photo 2 —This is a 3-D printer creating pieces. The white
point.
Legs
with
only
The main computational element
material is support material used to create overhanging
two
motors,
or
2DOF,
that
performs inverse kinematics and
objects. The printer is currently creating a femur, body plate,
do not have the flexibil- other control processes needs to
and coxa brackets.
ity of operating the end
employ little electrical power and
effector in three dimensions. The
must be compact enough to fit on
Finally, balance gestures were includonly possible space for the end effecthe hexapod, all while having powered to avoid any mechanical limitation
tor of a 2DOF leg is topologically
ful computational capabilities. While
within the individual legs.
equivalent to a torus. The first motor many microcontrollers and FPGAs
operates the leg’s horizontal motion
offer what is needed, they also can be
MECHANICS
and is denoted as the coxa. The coxa
incredibly difficult for performing
The mechanical portion of the
applies the least amount of torque
more complex computations such as
design process involved the physical
while the hexapod walks, and thereimage processing. The easy-to-use
design and the use of servomotors.
fore the RX-10 servomotor is used
Atom z530 processor can implement
Numerous companies distribute and
for each coxa joint.
a variety of available operating sysmanufacture servomotors. Hobby
The coxa angle is denoted as θc.
tems. A Fit-PC2 available from Commotors are commonly used and inexpulab sports the z530 processor in a
pensive, but they offer very little in
The second motor operates the first
compact form.
terms of feedback to the main conmain link of the leg, called the
trol system. Robotis offers the
femur. Since there is a physical disThe computer consumes around 5 W
Dynamixel series of servomotors that tance from the coxa motor axis of
of power at 1.6-GHz clock. Ubuntu
enables you to read various paramerotation to the femur motor axis of
9.04 is the preferred operating systers of the motor, including current
rotation, the distance needs to be
tem since C++ is the chosen proposition, applied torque, input volttaken into account for kinematics.
gramming language. OpenCV
age, and motor temperature. It also
This distance will be denoted as dc.
libraries are installed to implement
features an emergency shutdown
easy machine vision processing.
The femur rotates upward from the
operation in case one of the paramecoxa. The angle of the femur is
ters exceeds a settable threshold. We
denoted as θf, and the length of the
ELECTRICAL & VISION
used Robotis RX-28, RX-10, and AXfemur is denoted as df. Since this is
In order to communicate with the
12 motors, with 28, 10, and 12 kg-cm the joint that generally needs to
motors, we built a simple circuit to pipe
of torque, which are more than
strong enough for each joint in the
hexapod.
We used a Dimension BST-1200es
3-D printer for the mechanical construction. The 3-D printer is a rapid
prototyping machine that can create
parts in ABSplus plastic. Solidworks
was used for the design. We used
dimensional drawings of the motors
to model mechanical components to
fit with extreme precision. The layout of each leg was inspired by biological features; however, the exact
physical positioning was completely
Figure 1—Sending and receiving data from both a half-duplex TTL serial bus and RS-485 bus
arbitrary.
45
Photo 3—This is a simulation built with forward kinematics. The simulation was created using OpenCV by drawing individual lines
from point to point using the described
equations.
motor control protocol to the motors
from the computer. A Microchip Technology PIC18F4550 microcontroller
includes a USB hardware peripheral,
which can be used to connect directly
to the Fit-PC2. Microchip offers a
CDC library to allow ease of integration of USB for embedded projects.
The PIC transmits motor protocol
through both a half-duplex TTL-level
serial bus and a bidirectional RS-485
bus for the AX and RX series motors,
respectively. All Robotis motors use
the same motor protocol, so a simple
circuit was designed to integrate their
serial buses together (see Figure 1).
Also, contact sensors on the foot can
be easily integrated through direct
connection to the PIC.
With the computational capability
and hyper-threading ability, the
Atom processor performs kinematics
calculations together with complex
vision processing at nearly 900 Hz. A
Point Grey Chameleon machine
vision camera is used for the compactness. An inexpensive lens is used
to get a 150° field of view.
MOTOR CONTROL
46
The simple servomotor operates by
providing either a signal or protocol
specific to the motor which translates to the output shaft position.
Typical hobby-style servomotors
have a single-input pin that senses
the high part of a PWM signal. If the
PWM’s high time changes, the
mechanical output angle of the
motor changes accordingly. The servomotors used for the hexapod
require a slightly more complex protocol, involving an RS-485 bidirectional communication bus. The
Robotis motors have a few advantages over the standard PWM controlled servomotors. When a position
is sent to a Robotis motor, the motor
stores the position in memory and
attempts to move the output shaft to
the set angle. A standard hobby
motor operates differently in that it
needs a constantly fed PWM signal
for it to maintain a position. Also,
each hobby motor needs a separate
connection with a unique position
signal, whereas the Robotis motors
are daisy-chained on the same bus.
This makes wiring much easier, and
the position protocol can be sent to
all motors simultaneously.
Many low-quality motors on the
market work well for human-controlled applications, but their linearity for precise positioning in terms of
robotics can be poor. Poor linearity
results in inaccurate expectations of
forward kinematics. For example, if a
motor is calibrated to positions at 0°
and 90°, and if we send a signal based
on the calibration to set the motor to
180°, then a motor with good linearity should result in an output position very near to 180°. Motors with
sufficient linearity have a tolerance
of ±2°, while a motor with poor linearity will vary from ±5° to ±45°.
With any servomotor, calibration
is needed due to small manufacturing differences. Calibration is essential to the hexapod’s smooth operation. Given two measurement points
of desired output position—θd1 and
θd2—and the motor commands
required to move the motors to the
desired positions—L1 and L2, respectively—then any output command L
to move the output shaft as a function of desired angle θ is:
L (θ) = (θ − θ d1 )
L1 − L2
+ L1
θ d1 − θd2
[1]
KINEMATICS
Kinematics involves connections
of links and their respective angles.
It is possible to compute the position
of each link from the connections
and the angles, and most importantly, the end of the final link is called
the end effector. In the case of the
hexapod, if the three motor angles of
a given leg are known, then the position of the foot can be determined
through simple trigonometry.
The method of calculating the end
effector position as a function of the
angles is called forward kinematics.
Usually, it is more desirable to set a
desired foot position and determine
the angles required to set the foot at
the given position. This is particularly useful when planning a path or
trajectory for the foot. The process of
computing the angles as a function of
desired position is called inverse
kinematics.
FORWARD KINEMATICS
Although not particularly useful
for direct control of the hexapod, performing the forward kinematics can
be useful for understanding inverse
kinematics and for creating simulations. Refer to Photo 3 for a simulation created with OpenCV using forward kinematics. Figure 2 depicts the
mechanical configuration of one of
the hexapod’s legs. When performing
the derivation of forward kinematics,
it is easiest to break the system into
smaller sections.
First, consider dc and θc, the angle
and distance of the coxa. The position of the femur joint xF and yF can
be computed as in Equation 2 and
Equation 3, respectively.
x F = d C cos (θC ) + x L = d C cos (θC )
y F = d C sin (θ C ) + y L = d C sin (θ C )
z F = zL = 0
[2]
[3]
[4]
By knowing the projection of the first
link and the corresponding angle, it’s
possible to project the next point
based on the femur angle θf femur
length df to find the location of the
tibia joint, T.
x T = x F + d f cos (θf ) cos (θC )
y T = y F + d f cos (θf ) sin (θ C )
z T = z F + d f sin (θ f )
[5]
[6]
[7]
Finally, the coordinates of the end effector E (or foot) are the last projection,
CIRCUIT CELLAR®
•
which is determined from the tibia
angle θt and the tibia length dt.
x E = x T + d t cos (θ t + θ f − π ) cos (θ C )
y E = y T + d t cos (θ t + θf − π ) sin (θ C )
z E = z T + d t sin (θ t + θf − π )
θt
[8]
df
z
y
θf
INVERSE KINEMATICS
When you use your arm to reach
for an object, you easily operate your
muscles to angle your joints so as to
place your hand at the object’s location. Such a case involves inverse
kinematics, which is an inverse operation to forward kinematics. Similar
to forward kinematics, the easiest
method to inverse kinematics derivation is by simplification. With
desired end effector coordinates—xE,
yE, and zE—trigonometry can be used
to solve for θc, θf, and θt.
Again, consider the simple case
with the leg at the origin. First, from
an overhead view and the constraints
of the system, it becomes clear that
the only angle that affects the horizontal movement is θc. Thus, θc can
be solved first.
[11]
Note that atan2 is used instead of
tan–1 because it can return the full
possible range of an angle (–π ≤ θ ≤
π). Knowing the angle and the distance of the coxa link dc, Equations 2
and 3 may be applied to get the
known location of the femur joint. It
is now possible to construct a triangle with the femur, tibia, and an
(xE, yE, zE)
dc
The abovementioned equations only
apply when the leg is at the origin;
however, this can be generalized when
leg locations L (not equal to zero) are
applied.
θc
x
imaginary distance from the end
effector to the femur joint. This distance will be denoted at di.
di =
(xE
− x F ) + ( y E − y F ) + (z E − z F )
2
2
2
[12]
Knowing all lengths of the triangle,
the law of cosines may be applied to
solve for the tibia angle θt.
⎛ − d 2 + d 2f + d 2t ⎞
θ t = cos−1 ⎜ i
⎟⎠
2d f d t
⎝
[13]
A similar process can be performed to
solve for θf.
⎛ d 2 + d f2 − d i2 ⎞
−1 ⎛ z E ⎞
θ f = cos −1 ⎜ i
⎟⎠ + sin ⎜⎝ d ⎟⎠
2d f d i
⎝
i
[14]
Even though the simple case has been
shown for the leg starting at the origin, it’s easy to expand to legs not
located at the origin—and perhaps
rotated about the z-axis.
MOTION PLANNING
With the ability to decide the end
effector location of each leg, it is easily possible to create any arbitrary trajectory for the end effector to follow.
z
θ = b + 2π n
For example, if the hexapod were to
walk straight forward, then a possible trajectory would involve a backward motion of the foot, followed by
a lift-and-forward motion. The body
will move in space based on the
motion of each end effector relative
to the body.
The desired path when the hexapod walks forward is a straight line,
followed by a lift and a return to the
original position. Since the trajectory
needs to be continuous and oscillatory, the function will mimic some
properties of a sinusoidal function. If
a semicircle is desired for the lift,
then a characteristic equation for the
trajectory where Z is the lift and Y is
translational movement as a function of θ and a set break or lifting
point b may be synthesized, as in
Equation 15, under the definition and
constraint of Equation 16. The trajectory is graphed in Figure 3.
⎡ y⎤
⎢ z ⎥ = f (θ, b) =
⎣ ⎦
2θ′ ⎤
⎧⎡
⎪ ⎢1 − b ⎥ 0 ≤ θ′ < b
⎪⎢
⎥
⎦
⎪ ⎣0
⎪
⎡
− 2π ) π ⎞ ⎤ b ≤ θ ′ < 2π
θ
⎛
′
(
⎨ cos
⎪ ⎢ ⎜⎝ b − 2π ⎟⎠ ⎥
⎥
⎪⎢
⎪ ⎢ ⎛ (θ′ − 2π ) π ⎞ ⎥
⎪ ⎢sin ⎜ b − 2π ⎟ ⎥
⎠ ⎥⎦
⎩ ⎢⎣ ⎝
[15]
0 < b < 2π, θ ′ = θ ( mod 2π)
y
-1
•
1
Figure 3—An example of a trajectory
for simple forward
motion where n is
any integer
[16]
The function may be implemented
into walking forward with a maximum lift height gain h and travel
distance gain dT, and a frequency ω
CIRCUIT CELLAR®
1
θ = 2π n
dt
[9]
[10]
θC = atan2 ( y E , x E )
Figure 2—Kinematics
of a single hexapod
leg. The black, blue
and red lines represent the coxa,
femur, and tibia,
respectively. Servomotors change the
angles θc, θf, and θt.
47
and arbitrary phase φ to set the end
effector distances. The resting or pivot
positions xP, yP, and zP are constant
and translate the trajectory in space.
⎡ yp ⎤
⎡ y E ⎤ ⎡d T 0 ⎤
⎢ z ⎥ = ⎢0 h ⎥ f (ωt + φ, b) + ⎢ z ⎥
⎢⎣ p ⎦⎥
⎣ E⎦ ⎣
⎦
xE = x P
48
[17]
[18]
zero. The initial resting position of
the foot with respect to the body will
be denoted as xr and yr. When the
hexapod begins to move, the resting
position moves as well. The current
resting position in space, called the
pivot positions, is denoted as xP and
y P.
⎡ x p ⎤ ⎡cos (θb ) − sin (θb ) ⎤ ⎡ x r ⎤
⎡ xb ⎤
⎥⎢ ⎥ + ⎢ ⎥
⎢ ⎥= ⎢
cos (θb )⎦⎥ ⎣ y r ⎦
⎢⎣ y p ⎥⎦ ⎢⎣sin (θ b )
⎣ yb ⎦
filter. If α is close to zero, then the
foot takes a step closer to the current
pivot point. If α nears one, then the
feet take a much longer time before
they settle into the current walking
motion. A visualization of the
described dynamic walking mechanism is shown in Figure 4.
(
)
(
)
x DEST = x p + α x p − x OLD
y DEST = y p + α y p − y OLD
[20]
[21]
[19]
Different phases may be applied to
each leg so that the legs will lift at
0 ≤ α<1
[22]
different times. Various walking gaits
The foot is stationary as the body
may be generated by simply choosing
moves. It begins repositioning at a
different phases φ and breakaway or
The trajectory function in Equation
specific time. Before the foot lifts,
lifting points b.
the previous position is recorded into 15 without any applied gain results in
As for dynamic walking, it’s easy
–1 ≤ y ≤ 1 and 0 ≤ z ≤ 1. Using the
xOLD and yOLD. A simple trajectory
to mathematically fabricate specialvalue of y, a smooth motion can be
can be created to move the foot to
created from the old position to the
the current resting position. Indecase end effector trajectories to perdestination position. If the leg is at the
pendent of the initial foot location,
form a specific body movement. In
time to lift, then y = –1 and y increasthe foot returns to a stable position.
the aforementioned example, walking
es until the foot reaches the destinaThus, the foot control is dynamic.
forward is accomplished by moving
tion position at y = 1.
all feet in a similar oscillatory motion Returning to the pivot is, however,
not the most natural way for a creafrom front to back. All of these cases
1+ y
1 − y⎤
⎡
ture to walk. A more natural way
have a relatively constrained way of
⎢ x DEST 2 + x OLD 2 ⎥
⎡xE ⎤ ⎢
⎥
would be to project the foot beyond
motion; however, as they are not
1+y
1 − y⎥
⎢ ⎥ ⎢
y
y
=
+
y
the pivot point. The possible destinaadaptable to changes in motion.
DEST
E
O
LD
⎢ ⎥ ⎢
2
2 ⎥
⎢⎣ z E ⎥⎦ ⎢
⎥
tion of the foot, xDEST and yDEST, is
Instead, a dynamic way of generating
hz
⎢
⎥
trajectories would be more adaptable
shown in Equations 20 and 21,
⎢⎣
⎥⎦
in terms of different motion controls.
which looks similar to an α tracker
[23]
A simple way to have the
hexapod move would be to
WALKING GAITS
move the body in space and
A walking gait is the
have the legs take care of
sequence in which various
themselves. This way any
legs oscillate to move the
complex movement could be
body. Gaits vary with a creadest
applied to the body with
ture’s number of legs. A
ease.
biped is limited in its possi( d)
The dynamic walking
ble gaits. As exemplified, a
method will be dynamic in
gait can be picking up one
p
space but not in time. Thus,
foot while the other is down,
d
each leg will have a known
and vice versa. Also, both
lifting time. When the hexalegs can hop at the same
pod is first initialized, the
time, but that defeats the
old
legs slowly lift the hexapod
definition of walking. A
body upward to a standing
quadruped is again limited
position. The position of
by which leg can be lifted
each foot relative to the body
while remaining stable.
before the body translates in
Hexapods have the advantage
space is called the resting, or
of being stable over a wide
initial, position. The six
variety of gaits. We’ll cover
Figure 4—Visualization of dynamic walking. The blue outline is
locations are stored in memthree such gaits. With the
the previous hexapod position and foot locations, designated as
ory for the duration of the
ability to set the phase and
old. The gray body is the current body position. The dotted lines
program. The body location
lifting point of each leg, we
represent the current resting or pivot position, denoted as p.
and orientation in space are
can easily implement differThe gray circles represent the destination position for the feet
denoted as xb, yb, and θb,
denoted as dest. The feet are projected a total of αd beyond the ent gaits. Figure 5 shows the
current pivot position, where d is the distance from old to p.
various walking gaits with
which are all initialized to
CIRCUIT CELLAR®
•
arm and hope to reach
it. Instead, you move
Tripod gait
1
4
and angle your body
2
4
1
1
accordingly to help
4
1
2
5
3
extend your reach. In
4
2
5
other words, you bal3
6
5
5
6
3
6
ance your body with
2
respect to your limbs.
1
4
1
3
2
1
4
6
These balance gestures
2
5
are also applied the
4
2
5
3
hexapod, which helps
4
6
6
3
1
5
the hexapod avoid
3
6
2
reaching mechanical
5
1
1
2
4
limits. Balance gestures
4
1
6
are especially useful
3
2
5
3
2
5
4
with the application of
6
3
terrain adaptation
6
6
3
5
behaviors, when the
exact terrain is
unknown. When the
Figure 5—Various hexapod walking gaits. A colored circle signifies a leg in the lifting state. Notice the differhexapod is traversing
ence in leg indexing for the ripple gait.
over rugged terrain, the
body always remains
centered with respect to each foot
corresponding leg indexing.
of the gait allows for good stability;
location.
The tripod gait gets its name from
but with two wave gaits, the refresh
Imaginary springs with identical
having three legs supporting the body rate can be double that of the wave
spring constants k have been
at one time while the other three
gait, thus making it much faster than
attached from each foot to the correreposition themselves. Once the
the wave gait.
sponding coxa joint (see Figure 6).
three legs have been repositioned,
When the springs have the least posthe three previously supporting legs
π
4π
φi = 2π − i, b =
sible amount of tension, the position
lift and reposition. This is similar to
3
3
and orientation of the body is considhow a biped walks. Under the appli[26]
ered to be in balance with respect to
cation of the trajectory function
BALANCE GESTURES
the foot positions. It is assumed that
above, the phases and lifting point
When you see a coin on the
the hexapod to which the balance
can be set as in Equation 24. The triground, you don’t stretch out your
pod walking gait has the advantage
of being the fastest gait, but it does
y
not look natural.
Wave gait
Ripple gait
E4
[24]
E1
The wave gait gets the name from
the wavelike motion of the leg lift
order. Here, only one leg is lifted at a
time. This gait is the slowest possible
gait, but it is very stable and looks
fairly natural.
π
5π
φi = 2π − i, b =
3
3
•
E5
c5
c1
Өb
yb
b
yc6
c2
c6
[25]
The ripple gait has a balance
between speed and stability, and it has
the most natural look. This gait is a
variation on the wave gait, where there
is an individual wave gait on both
sides of the hexapod. The wave nature
c4
yE5
c3
E2
Xc6
E6
E3
x
Xb
XE5
Figure 6—Spring model for balance gestures. Black lines represent ideal springs, with arbitrary spring constant k. Body center and orientation are represented by b and θb. Coxa locations with respect to body center represented by c. End effector locations represented by E.
CIRCUIT CELLAR®
φ1, φ3, φ5 = 0, φ2, φ 4, φ5 = π, b = π
49
gestures are being applied is fairly symmetric. Due to
symmetry, the summation of all six coxa locations with
respect to the body is shown in Equation 27.
6
∑x
i=1
6
= 0,
ci
∑y
i=1
=0
ci
[27]
The tension due to each spring Fi, as shown in Figure 6,
is determined through observation of the orthogonal components, as in Equation 30.
(
Fxi = x Ei − x ci cos (θb ) + y cisin (θb ) − x b
(
)
Fyi = y Ei − x cisin (θb ) − y ci cos (θ b ) − y b
[28]
)
[29]
∧
∧
Fi = Fxi i + Fyi j
[30]
Under the assumption of symmetric coxa locations, the sum
of the orthogonal components of spring tension should be
zero, as there otherwise would be translational accelerations.
6
∑ (x
i =1
Ei
)
− x ci cos (θb ) + y ci sin (θ b ) − x b k = 0
[31]
6
∑ (x
i =1
6
Ei
6
∑ (x
i =1
Ei
(
)
− x b ) + ∑ − x ci cos (θ b ) + y cisin (θ b ) = 0
i =1
[32]
− xb ) = 0 ⇒ xb =
6
1
∑ x Ei
6 i =1
[33]
The resulting location results in the simple average location of the feet. A similar process is done for yb, the body
location in y.
yb =
1 6
∑ yEi
6 i =1
[34]
The remaining parameter to be solved is the angle of the
body, θb. Since the position has been solved, it is possible
to find the total tension of the springs as a function of
the body angle. By taking the derivative of the magnitude
of the tensions with respect to the body angle, and setting the value equal to zero, the minimum or maximum
tension can be solved. The resulting angle can then be
solved using Equation 36.
∂
∂θ b
6
∑ F (θ ) = 0
i =1
i
b
[35]
6
50
θb = π − tan −1
∑ ⎡⎣ x ( y
ci
b
i=1
6
∑ ⎡⎣ y ( y
ci
i=1
b
− y Ei ) − yci ( x b − x Ei )⎤⎦
− y Ei ) + x ci ( x b − x Ei )⎤⎦
[36]
The described balance gesture equations only apply to a
two dimensional case, but the same principles are used
in the expansion to three dimensional balance gestures.
RESULTS
The ABS plastic body and leg components from the 3-D
printer are rigid, thus enabling accurate kinematic predictions.
The Atom z530 processor calculates the control algorithms
with ease at nearly 200 Hz without code optimization. The
PIC18F4550 pipes motor control protocol from the computer
through USB 1.1 to the Robotis motors through two different
serial interfaces at 1-Mbps data rate.
Through the implementations of the described algorithms,
the hexapod is able to traverse smoothly in a simple flat environment. The implementations of inverse kinematics enable
the effortless control of the end effector’s Cartesian placement
for each leg. The simple trajectories enables the hexapod to
walk forward. The inclusion of dynamic foot placement
enables facile operation of the body placement in space, where
the feet are automatically projected to a stable location.
Mechanical limits are rarely encountered due to the inclusion
of balance gestures. The various walking gaits are easily implemented by simply assigning different lifting points and leg
phase shifts.
The only unsatisfactory element of the hexapod is contained
within the servomotors. Unfortunately, the motors implement
only programmable nonlinear proportional (P) controllers, with
a limited resolution of 10 bits per 300°. The hexapod is capable
of performing much better with PID controller-based motors,
preferably with a higher resolution at a minimum of 12 bits
per 300°.
LOOKING FORWARD
You’re now familiar with how we designed and constructed
a hexapod robot. The biologically inspired body and leg design
was modeled in Solidworks and created with a 3-D printer.
In this article, we described equations for both forward and
inverse kinematics. We also presented a simulation created
with forward kinematic methods, and we explained how controlling the hexapod involves inverse kinematic methods. We
built simple trajectories for the end effectors of each leg to
synthesize a simple walking motion.
In addition, we explained how to implement dynamic foot
placement techniques for complex body motion. Using simple
phase and lifting point selection, it’s possible to generate various walking gaits. Balance gestures were included using a
spring model to help avoid legs reaching mechanical limits.
The hexapod provides a good foundation to perform higherlevel control algorithms. We included a machine camera in the
mechanical design, although we haven’t implemented it at
this stage. In the future, we plan to implement various behaviors based on vision. One such behavior is machine learning,
specifically Q-learning using optic flow to sense movement.
Force-sensing resistors (FSRs) are also included in the foot
design, but they are not connected at this time. We’ll explore
terrain adaptation based on vision and foot contact, in addition
to machine learning, in the future. I
Authors’ note: We want to acknowledge Dr. M. Anthony Lewis of
the Robotics and Neural Systems Laboratory (RNSL) at the
CIRCUIT CELLAR®
•
University of Arizona for his invaluable assistance with the project.
We also thank Stewart Christie and the Embedded Design Center
(EDC) team at Intel Corp. for their gracious support and assistance
to bring the project to its current level.
Matt Bunting ([email protected]) earned a degree in Electrical Engineering at the University of Arizona. His interests include
robotics, specifically in the areas of cognitive and neural systems.
John Rogers ([email protected]) earned a degree in Computer
Engineering at the University of Arizona. His interests include
robotics and computer engineering.
REFERENCE
[1] H. Kopka and P. W. Daly, A Guide to LATEX, 3rd
Edition, Addison-Wesley, 1999.
SOURCES
Atom z530 processor
Intel Corp. | www.intel.com
Microchip Technology | www.microchip.com
Chameleon camera
Point Grey | www.ptgrey.com
AX-12, RX-10, and RX-28 servomotors
Robotis | www.robotis.com/zbxe/main
NEED-TO-KNOW INFO
covered in Matt Bunting and John Rogers’s Issue 240
article, the Circuit Cellar editorial staff highly
recommends the following content:
—
Robot Navigation and Control
by Guido Ottaviani
Build a navigation control subsystem for an
autonomous explorer robot. Here you learn about a
robotic platform that drives motors and controls an Hbridge. Topics: Robot, Navigation, Motor, Telemetry
—
Vision-Guided Robotics
by Hanno Sander
Want to build a vision-guided, balancing robot that
interacts with its environment? You can tackle this
project with a Parallax Propeller, a handy design kit,
and an inexpensive camera. Topics: Robot, PID,
Fuzzy Logic, Vision, Frame Grabber
microOBD 200
www.microOBD.com
OBD-II interpreter module in a
DIP-24 package. UART interface,
bootloader, 100% compatible with
the ELM327 command set.
$49
OBDLink
www.OBDLink.com
$99
#1 PC based OBD-II scan tool. USB,
RS232, Bluetooth, and WiFi options.
Free OBDwizTM software included.
ECUsim 5100
www.ECUsim.com
Professional multiprotocol OBD-II
ECU simulator. Functional and
physical addressing, three virtual
ECUs, software configurable.
Full support of all OBD-II protocols
Free regular firmware updates
Discount code for free shipping: CC1006FS
Questions? Email [email protected]
•
CIRCUIT CELLAR®
$549
51
BOARDS, BOOKS, DVDs AND MORE AT WWW.ELEKTOR.COM/SHOP
Elektor Shop
The world of electronics
at your fingertips!
!
EW
N
Get started quickly and proceed rapidly
Python Programming
and GUIs
This book is aimed at people who want to interface PCs with hardware projects using graphic
user interfaces. The programming language used
is Python, an object-oriented scripting language;
a higher level language than, say, C. The book
guides you through starting with Linux by way of
a free downloadable, live bootable distribution
that can be ported around different computers
without requiring hard drive installation. Practical demonstration circuits and downloadable, full
software examples are presented that can be the
basis for further projects.
224 pages • ISBN 978-0-905705-87-3 • $47.60
Books
ore
This book and m
are available at
m/books
www.elektor.co
Microsoft
approved!
Learn more about C# programming and .NET
A laser alarm, USB teasing mouse, soundswitch and much more
C# 2008 and .NET
programming
50 PIC Microcontroller projects
for Electronic Engineers
This book contains 50 fun and exciting projects for PIC microcontrollers such as a laser alarm,
This book is aimed at Engineers and Scientists
USB teasing mouse, eggtimer, youth repellent, soundswitch, capacitive liquid level gauge, “finger
who want to learn about the .NET environment
in the water” sensor, guarding a room using a camera, mains light dimmer (110-240 volts), talking
and C# programming or who have an interest in
microcontroller and much more. Several different techniques are discussed such as relay, alternating
interfacing hardware to a PC. The book covers the
current control including mains, I2C, SPI, RS232, USB, pulse width modulation, rotary
Visual Studio 2008 development environment,
encoder, interrupts, infrared, analog-digital conversion (and the other way around), 7-segment
the .NET framework and C# programming lan-
display and even CAN bus. Three PIC microcontrollers are used in this book, the 16f877A, 18f4455
guage from data types and program flow to more
and 18f4685. It is also discussed how you can migrate your project from one microcontroller
advanced concepts including object oriented
to another - 15 types are supported - including two example projects.
programming.
440 pages • ISBN 978-0-905705-88-0 • $58.10
240 pages • ISBN 978-0-905705-81-1• $47.60
Prices and item descriptions subject to change. E. & O.E
CD/DVD-ROMs
Bestseller!
See the light on Solid State Lighting
DVD LED Toolbox
New!
Reign with the Sceptre
dsPIC Control Board
This open-source & open-hardware project
This control board has been designed for in-
This DVD-ROM contains carefully-sorted com-
aims to be more than just a little board with a
corporation into typical industrial electronics
prehensive technical documentation about
big microcontroller and a few useful peripherals
applications like controlling motors or ad-
and around LEDs. For standard models, and for
— it seeks to be a 32-bit ARM7 fast prototyping
justment of static up- or down-converters.The
a selection of LED modules, this Toolbox gath-
system. To justify this title, in addition to a very
objectives were to obtain a board with a large
ers together data sheets from all the manufac-
useful little board, we also need user-friendly
number of pulsewidth modulation (PWM)
turers, application notes, design guides, white
development tools and libraries that allow fast
generators, which enables us to control several
papers and so on. It offers several hundred dri-
implementation of the board’s peripherals. Am-
motors and static converters at the same time.
vers for powering and controlling LEDs in dif-
bitious? Maybe, but nothing should deter you
The cost of the control board needed to be
ferent configurations, along with ready-to-use
from becoming Master of Embedded Systems
as low as possible too. In addition, it must be
modules (power supply units, DMX controllers,
Universe with the help of the Elektor Sceptre.
possible to construct the board using a
dimmers, etc.). In addition to optical systems,
light detectors, hardware, etc., this DVD also
addresses the main shortcoming of power LEDs:
heating. This DVD contains more than 100 Ele-
0#"POPULATEDANDTESTED
TESTSOFTWARELOADED
Art.# 090559-91• $143.60
ktor articles on the subject of LEDs.
soldering iron, without requiring use of a reflow
oven.
0#"POPULATEDANDTESTED
Art.# 090073-91 • $225.90
ISBN 978-90-5381-245-7 • $46.00
Elektor is more
than just your favorite
electronics magazine.
It’s your one-stop shop
for Elektor Books,
CDs, DVDs,
Kits & Modules
and much more!
DVD Elektor 1990
www.elektor.com/shop
ment of every electronics lab. However, the
increased complexity of a switch-mode design
This DVD-ROM contains the full range of
scares away many potential builders, even
1990-1999 volumes (all 110 issues) of Elektor
though it actually isn’t all that complicated if
Electronics magazine (PDF). The more than
you use a suitable combination of well-known
2,100 separate articles have been classified
(month/year), but are also listed alphabetically
by topic. A comprehensive index enables you to
search the entire DVD.
ISBN 978-0-905705-76-7 • $111.30
A power supply with adjustable output voltage
and current limiting is part of the basic equip-
through 1999
chronologically by their dates of publication
UniLab
technologies. This circuit is suitable for building
Elektor US
PO Box 180
Vernon, CT 06066
USA
Phone: 860-875-2199
Fax:
860-871-0411
E-mail: [email protected]
a single or dual power supply.
0#"ANDALLCOMPONENTSLESSPOWER
TRANSFORMER
Art.# 090786-71 • $103.30
Kits & Modules
110 issues, more than 2,100 articles
L
ESSONS FROM THE TRENCHES
by George Martin (USA)
Rules and Assignments
In this third part of the Sudoku problem-solving exercise, you learn about a
rules package and then receive some assignments. You’ve put C language
to the test. Now a solution is within reach.
L
54
et’s begin with a lesson. I’ve introduced
some new (to us) C language operators—for example, all the file I/O. But we’re
not doing enough work with files to make
much of a point by focusing on all you can do
with the file operations. I’ll save file I/O for a
later date. I would like to talk about the % and
/ operators. The modulus operator (%) computes the remainder that results from performing integer division. So, 11 % 3 = 2. If
you were to just divide x = 11/3, you would
get the 3 as an answer. That’s the integer portion and the remainder is lost. If you want the
remainder, you need to use the % operator.
So, what’s the big deal? I’ll show you what.
Remember that our database was organized as
Puzzle.Block[b].Cell[c].Numb[n], where b, c,
and n represented a specific block, cell, and
number, respectively. But as humans, we want
to talk about the rows and columns of a puzzle. We need a routine that takes a row number and column numbers and returns a block,
cell, or number. And to do that, we’ll use the
% operator.
Consider a 16 × 16 grid. Start in the top left
and fill in the grid with the following: 0123,
0123, 0123, 0123 for the first row; 4567, 4567,
4567, 4567 for the second row; 89AB, 89AB,
89AB, 89AB for the third row; and then CDEF,
CDEF, CDEF, CDEF for the fourth row. You
need to repeat this three more times to fill in
the grid. Doing so numbers each of the cells in
the puzzle.
In the routine INT16 RowCol2iCell(INT16
Row, INT16 Col);, you find the statement
return (4*(Row%4) + (Col%4));. This routine is passed the row and column that interests you. Using the % operator, it returns a
number representing the cell. Look at the test
values in Table 1. If you take another 4 × 4
grid and fill in this routine’s results, you’ll see
that the table describes the cells of the puzzle.
Another routine takes row and column numbers and returns a block number. That routine
is INT16 RowCol2iBlock(INT16 Row, INT16
Col);. In that routine, you’ll find the statement return((Row/4) + 4*(Col/4));. This
routine is passed the row and column that you
are interested in. Using the / operator, it
returns a number representing the block. I
won’t build another test table, but you should.
Remember: the / operator returns the dividend but not the remainder. Isn’t it better to
see these operators in action that just remembering the rules by rote?
RULES
Now it’s time to implement the rules package. I started the rule design by assigning numbers to the rules starting with 1. (Brilliant!)
CIRCUIT CELLAR®
•
Rules 1, 2, and 3 require that there
only be one hex digit in each row,
column, and block, respectively. Let’s
take a look at the logic for Rule 3 in
Listing 1.
This rule is implemented by first
looking at each of the blocks and
then in a specific block looking at all
the cells. If in one of the cells a number is found to be CCI_YES (that
number is the valid number for the
entire cell), mark all the numbers in
the cell and CCI_NO (meaning it’s
not this number) and then restore the
original cell to CCI_YES. I implemented it in this manner because the
code was most straightforward. And
since the puzzle is held in RAM,
doing all the writes to memory is not
a time or performance penalty. This
design would be much different if the
RAM were slow and expensive (such
as flash). We would apply more code
to minimize the RAM usage.
Rules 1 and 2 are the same, but
they are applied to the rows and
columns. In those routines, you will
find that we start the searching by all
the rows and then all the columns.
But then you need to change from
the Row Column notation to the
Row
0
0
0
0
1
1
Column
0
3
4
7
0
3
Cell
0
3
0
3
4
7
Table 1—These are the test results for the
RowCol2ICell() routine.
Block Cell notation. Calls to simple
routines convert from Row Column
to Block Cell. You should look this
over carefully. In your designs, I bet
there will come a day you need to
access a database you’ve created
using several points of view. This is
a very common situation you’ll
encounter as your problems get more
and more complicated.
Great. We’re all done.
SOLUTION GENERATION
With the three basic rules running,
puzzle solutions will just fall out.
Well, I started running the code with
inputs from a file. I ran the puzzle
against Rules 1, 2, and 3 and then
sent the output in SHOW format to
the screen. The output of the puzzle
in the SHOW format filled the screen
and it was very difficult to see if anything was changing. So, my first
addition was to keep a running total
of all the numbers printed on the
SHOW screen. This seemed to make
sense. As numbers were eliminated,
the sum was reduced. I found that
numbers stopped being eliminated as
I kept applying the rules, but the
puzzle was not solved. “Oh, no! This
won’t work,” I thought. “I can’t
solve the problem!”
I needed to look at the database in
another format to see what was
going on. I came up with the LIST
command to print out the database
in text format. Again, there was a lot
of data. (I have since reduced the
command’s output by eliminating all
the CCI_NO responses.) I found that
after applying Rules 1 through 3
some of the cells had all the numbers
but one tagged as CCI_NO. The
other number was tagged as
CCI_MAYBE. Common sense tells
you that whenever you come across
this situation you should mark the
lone CCI_MAYBE as CCI_YES. And
Rule 4 does just that. The logjam
was broken.
Li sting 1—Rule 3 with only one number per block
for (Block = 0; Block < 16; Block++) {
for (Cell = 0; Cell < 16; Cell++) {
for (Numb = 0; Numb < MAX_CANDIDATES; Numb++) { // for all numbers
if (puzzle.block[Block].cell[Cell].numb[Numb] == CCI_YES) {
Changed = 1;
for (iCell = 0; iCell < 16; iCell++) {
puzzle.block[Block].cell[iCell].numb[Numb] = CCI_NO; // Mark all as NO
}
puzzle.block[Block].cell[Cell].numb[Numb] = CCI_YES; // Restore original
}
} // end of for (Numb = 0; Numb < MAX_CANDIDATES; Numb++) {
} // end of for (Cell = 0;iCell < 16;iCell++) {
} // end of for (Block = 0; Block < 16; Block++) {
return(Changed);
}
// end of INT16 RunRule3(void) {
//********************************************[ Public ]*******
//
// Rule 3 Only one number per Block.
//
//*************************************************************
INT16 RunRule3(void) {
INT16 Block, Cell, Numb; // variables for loop counters
INT16 iBlock, iCell, iNumb; // variables for loop counters
INT16 Changed = 0;
55
rules until none of the rules report any changes. What a
great situation. If you are going to use this program, I
suggest you get the rules reporting to work. I see a “challenge” here. Get the rules properly reporting the changed
status and I’ll mention your name in a future article. If
you plan to participate, please tell me if you’re a C beginner or a black belt. I’ll include that information.
Listing 2—The “ideal” rules package
Changed = 1;
while (Changed) {
Changed = 0;
printf("Rule 1");
Changed += RunRule1();
printf(" 2");
printf(" 3");
printf("\n\r");
}
MISSING RULES
I believe there will be more rules for complicated (difficult) problems. Consider the following in a block:
Now, running Rule 4 reduced the SHOW command’s
output further. Rerunning Rules 1, 2, and 3 again produced lower results when running the SHOW command. In
fact, running all the rules over and over generates a solution to the puzzle. Ta-da!
A little bit of a back step. My intention was to have
each rule return a value indicating if running that rule
caused anything to be changed (1 = a change, 0 = no
change). Well, I was trying to get the rules working, deal
with my Rule 4 issue, and meet production deadlines. I
never got the changed Y/N return for the rules to work
reliably. Just think: if that had worked, I could have run
the code fragment in Listing 2.
The code fragment in Listing 2 will keep applying the
READER FEEDBACK
I recently received some interesting feedback
from readers about this article series. Two examples
follow.
Ivey Cole wrote:
“In the March issue of Circuit Cellar, you mention looking at a variety of C compilers. Was wondering if you’ve looked at Open Watcom, which is
the ‘old’ Watcom compiler that went open many
years ago. In my opinion it and the Borland packages were fairly comparable in the ‘good old days,’
with each having different features.”
For more information about Open Watcom, go to
www.openwatcom.org/index.php/Main_Page.
56
Dave Brown wrote:
“I have been following your Circuit Cellar articles on the C language with interest. The Sudoku
puzzle-solving Program looks like fun. I think I will
join in. FYI, here is another free C compiler that
may interest you. My son who is a grad student in
chemistry needed a C compiler for a project. I
pointed him to the Visual C++ Express Edition,
which is free: www.microsoft.com/express/vc/.”
Cell A has 1, 2, and 3 as candidates (CCI_MAYBE)
Cell B has 2 and 3 as candidates
Cell C has 2 and 3 as candidates
I could make up a rule that would set Cell A to 1 since
it’s the only place the number can be used in the block.
And similar rules would apply for the rows and columns.
I’ve seen this for solving very difficult 3 × 3 puzzles. I’ll
leave it up to you to design this rule set and any other
that comes along. If any come my way, I’ll see they get
published.
REVIEW
Let’s review the steps we took to solve this problem.
We had our initial design approach described using UML
diagrams (Part 1, Circuit Cellar 236). These diagrams
contained the routines gathered into modules and database design described in the notes. As work progressed,
we discovered new routines and better methods for solving the problem (Part 2, Circuit Cellar 238). We also reorganized the code as work progressed. And now, with this
article, we have a final solution with rules for solving the
problem. Some of the rules were anticipated and one of
the rules was a surprise. Think about how we worked
our way through the problem. The most important thing
to take away from this project is that conditions change
and you should be flexible enough to address those
changes. We didn’t even include any changes from a marketing department or a customer. Expect changes in the
real world.
As the design now stands, would you change it in anyway? I did not make the code universal for 2 × 2, 3 × 3,
and 4 × 4 puzzle sizes. Would you change the database
design, I/O features, or reorganize the routines in any
way? Think about this and perhaps your next design will
start off better.
SUCCESS WITH C
Well, did we just destroy Elektor’s Hexaduko puzzle
challenge? Perhaps I should be looking for an alternative
career. Just kidding. Hopefully, we just increased participation.
On a more serious note, let me know what you think
about this article series. I’ve already heard from some
readers. (Refer to the sidebar.) Did you learn anything
about writing C programs and UML? Are you now more
CIRCUIT CELLAR®
•
confident about attempting such a design? Do you think this series has
helped you read and understand C programs? I
Author’s note: There are many C-like languages and many procedural-type languages. For instance, Joachim Schueth of Bonn has published a Hexadoku solver
written in Ada (www.schlaupelz.de/hexadokusolve.adb). No comments are shown as
part of the code. But I’m mentioning this to show you that other solutions exist.
Also, think about commenting your code if you intend to describe what’s going on
and give others insight into what you are doing. I’m not criticizing Joachim Schueth;
it looks well-written. Just think about commenting yours. Think about my comments. Are they sufficient to document the design?
George Martin ([email protected]) began his career in the aerospace industry in 1969.
After five years at a real job, he set out on his own and co-founded a design and manufacturing firm (www.embedded-designer.com). His designs typically include servomotion control, graphical input and output, data acquisition, and remote control systems. George is a charter member of the Ciarcia Design Works Team. He is currently
working on a mobile communications system that announces highway info. He is also a
nationally ranked revolver shooter.
SOURCES
CC386 C Compiler
LadSoft | http://ladsoft.tripod.com/cc386.htm
MagicDraw Modeling tool (v.9.5)
No Magic, Inc. | www.magicdraw.com
NEED-TO-KNOW INFO
Knowledge is power. In the computer applications industry,
informed engineers and programmers don’t just survive, they
thrive and excel.
For more need-to-know information about topics covered in
George Martin’s Issue 240 article, the Circuit Cellar editorial staff
highly recommends the following content:
—
A Sudoku Puzzle-Solving Program
by George Martin
Ready to write a program in C language? Here you learn about
starting a program for solving a Sudoku puzzle. Topics: C Language,
Program, Sudoku, IDE, UML, database
Text Adventure Gaming
by Chris Cantrell
Do you like learning new languages? Try using a chip and the
Cog Coordination language (CCL) for an interesting gaming application. Topics: Language, Programming, CCL, RISC, SD Card, FAT
•
CIRCUIT CELLAR®
—
57
F
ROM THE BENCH
by Jeff Bachiochi (USA)
Application Communication
with USB (Part 2)
The Importance of Descriptors
You know how a USB device goes through a number of states as it progresses
during the enumeration process. The next topic to understand is the standard
device descriptor, which is the first item in a device’s datasheet.
I
58
must be the reincarnation of a cat. I’m
curious about many things. In fact, I’ve
used a number of my “nine lives” by letting my
curiosity take precedence over the self-preservation instinct. Do you know anyone who can say
they’ve never performed an act of stupidity at
some point in their life? Somewhere along the
line, we learn how fragile our bodies are and
that tempting fate should not be an everyday
experience. At this point, curiosity takes a
backseat to personal safely. We are more apt,
say, to learn about aerodynamics through references and modeling, rather than jumping off a
cliff with a hang glider.
I’ve been interested in USB since it was forced
upon us years ago. Even though many manufacturers have given us the tools to use it, without
having to know how it really works, I just can’t
leave it at that. I want to know more about its
operation. For those of you who might have
similar feelings, I hope I can shed a little light
on this by going through a simple USB project
that will enumerate and transfer application
data using assembly language. This project
began last month. I encourage you to start there
and get up to speed on the USB hardware
peripheral—the serial interface engine (SIE)
built into many of Microchip Technology PIC
microcontrollers. While each manufacturer may
arrange things differently, they all must obey
the same rules. Thus, you should be able to take
what you learn here and apply it to any micro.
We left off looking at the start of the enumeration process as seen from the Beagle USB 480
protocol analyzer I use to display the bus activity between the USB host (my PC) and the project’s device circuitry. After forcing a USB reset,
the host has requested a descriptor. Through the
SETUP packet received by the SIE, I discovered
that the host was requesting a standard device
descriptor (SDD) via a GetDescriptor transfer.
We know that the purpose of designing the
USB interface was to make it user-friendly yet
universal. That means that the host must be
able to figure out how to access a device automatically when a user plugs it into the bus. To
do this, a device must be able to answer all the
questions the host will ask it. These questions
have been predetermined by the USB specifications. So, as a device, we need to be ready to
respond with the answers. The answers come
from USB descriptors, a collection of information describing the device. Each descriptor uses
an identical format to simplify the exchange of
information. A descriptor begins with a length
byte followed by a type byte. Let’s begin with
the descriptors, as defined in this project, to
describe its configuration to the host, the
CIRCUIT CELLAR®
•
the index of any string
descriptor describing this
Offset Field
Size Format
Description
Project Value
interface. The last 2 bytes
0
bLength
1
Number
Descriptor length in bytes
0x12
define the attributes of the
1
bDescriptorType
1
Constant
DEVICE
0x01
configuration (self-powered
2
bcdUSB
2
BCD
USB Specification revision
0x0200
or supports remote wakeup)
4
bDeviceClass
2
Class
Class Code
0x02
and the maximum power
5
bDeviceSubClass
2
SubClass
Sub Class Code
0x00
consumption (if it were to
6
bDeviceProtocol
1
Protocol
Protocol Code
0x00
require power through the
7
bMaxPacketSize0
1
Number
Maximum packet size EP0
0x08
USB connector).
8
idVendor
1
ID
Vendor ID
0x04D8
In this project, the host
10
idProduct
1
ID
Product ID
0xFE96
discovers that this configu12
bcdDevice
1
BCD
Device Revision
0x0000
ration (1) has two interfaces
14
iManufacturer
1
Index
Manufacturer String Index
0x01
associated with it. It also
finds that the device is self15
iProduct
2
Index
Product String Index
0x02
powered and supports
16
iSerialNumber
1
Index
Serial Number String Index
0x00
remote wakeup. If its local
17
bNumConfigurations 1
Number
Maximum Device Configurations
0x01
power is removed, it would
Table 1—The standard device descriptor describes general information about the device. It indicates to
require 100 mA. While the
the host how many configurations it should look for that are associated with the device.
host discovered this configuration descriptor’s length
your device supports. In this project,
is only 9 bytes, it knew to look for a
device descriptor, the configuration
we have just one.
total length value within this descripdescriptor, and the string descriptor.
My SDD is 0x12 bytes long, so how
tor. It can continue to read in addican the host get this if I’m using a
tional transactions (using the “in”
SDD
maximum packet size of eight? If you
token) until it has all of this informaA USB device has only one SDD as
refer back to the first part of this
tion. The host can then break down
it is a global description of the device
series, you’ll see that the host follows
this data into additional descriptors
and all its configurations. Table 1
its GetDescriptor set-up transaction
looking for more information on each
shows that this descriptor contains
of the two interfaces.
the version number of the USB specifi- with additional “in” token requests
until it has received all it requires. But
Appended to, and as part of, the
cations that it conforms to, so the
hold on to that thought; we haven’t
total SCD are any additional descriphost will understand the rules it can
discussed the “in” and “out” token
tors that are required to fully define
play by.
yet.
the device’s configuration. Here we
Next are the device’s class and subhave two interfaces to define, and
class. These define the kind of
each will have its own interface
device—in our case, it is the commuSCD
descriptor. The first interface (internications device class, covering
Unlike the SDD, there may be more
face descriptor 0) has no alternative
telecommunications and networking
than one standard configuration
setting—so its value is zero (the first
devices. The communication device
descriptor (SCD). Remember, the
and only)—and the number of endclass has no subclass or device protonumber of configurations was defined
points used for the interface is one.
col, so these bytes are zero. The packback in the last byte of the SDD. In
Also within the interface descriptor
et size byte tells the host a packet’s
this case, we’re just dealing with a
are communication interface class (2),
maximum size. For EndPoint0, that’s
single configuration; therefore, we
subclass (2), and protocol (1) values. In
8, 16, 32, or 64 bytes.
need only one configuration descripNext, the device descriptor lists the
tor. The configuration descriptor is very this descriptor, the host learns that
interface 0 is of the communication
vendor ID (that’s Microchip Technolocomplex. It is a container, if you will,
class using the abstract control model
gy), the product ID (that’s mine), and a
for all the interface descriptors associathaving one endpoint.
release or version number for my
ed with this configuration. It uses the
application. Because all this is kind of
standard descriptor format so the host
We know that USB devices require
cryptic, the descriptor includes a num- knows the descriptor’s type and size by
specific device drivers to be loaded on
ber of bytes that can point to string
the first 2 bytes (see the table posted on
the PC. To support all possible USB
descriptors that will describe the man- the Circuit Cellar FTP site).
devices, this driver would be mamufacturer, the product, and the serial
The SCD’s next 2 bytes define the
moth. So, the driver that gets loaded for
number using ASCII text. These may
total size of all the descriptors associeach device contains only the subset
show up in various places on your PC
ated with this configuration. Follownecessary for its associated device. The
when installing drivers or viewing sys- ing this is the actual number of interdriver gives the host the ability to
tem information. Finally, one last byte faces supported by this configuration,
know what descriptors to find for that
indicates the number of configurations the number of this configuration, and
particular device. The subclass 2
•
CIRCUIT CELLAR®
Standard Device Descriptor
59
Standard String Descriptor0
Offset
Field
Size
Format
Description
Project Value
0
bLength
1
Number
0x04
1
bDesriptorType
1
Constant
STRING
0x03
2
wLANGID
2
Number
Country Code
0x0409
Offset
Field
Size
Format
Description
Project Value
0
bLength
1
Number
0x36
1
bDesriptorType
1
Constant
STRING
0x03
2
bString
N
Number
Text
“Microchip Technology, Inc.”
Offset
Field
Size
Format
Description
Project Value
0
bLength
1
Number
0x4C
1
bDesriptorType
1
Constant
STRING
0x03
2
bString
N
Number
Text
“Imagine That
(www.imaginethatnow.com)”
Table 2—The standard string descriptors have two formats. String0 defines the language used by the host to talk to the user. The remaining
strings contain ASCII text presented in a UNICODE presented string.
60
(abstract control model) has its own set
of descriptors that are used to help
describe its functions through an interface. There are 25 subtype descriptors for
the communications class. We will be
using four for this project: header, call
management (CM), abstract control
management (ACM), and the union
functional descriptor. You can read
about all of them in the communication
class specifications. The header descriptor indicates the version of the specifications that the interface is adhering to.
CM defines the call-handling capabilities—and there are none, in this case.
The ACM identifies the capabilities of
the interface. These include: network
connection support; sending break, line,
and serial states; and COMM features.
Finally, the union descriptor defines the
master interface (interface 0 in this case)
and the subordinate interfaces. In this
instance, there is only one other interface—interface 1.
Following the subclass descriptors for
this interface, you must include the endpoint descriptors for this interface. One
endpoint is for this interface and its
descriptor identifies the actual endpoint
(EP1In) using an interrupt transfer type.
This short descriptor also includes the
packet size (64 bytes) and how often the
host will poll for data (2 ms).
The second interface (interface
descriptor 1) has no alternative setting, so its value is zero (the first and
only) and the number of endpoints
used for the interface (two endpoints
here). The data interface class (10) has
no subclass and uses no protocol, so
these values are zero. In this descriptor, the host learns that interface 1 is
of the data class using no specific protocol and having two endpoints. This
interface has no additional descriptors
other than the two specified endpoint
descriptors. One endpoint is defined
as using endpoint (EP2Out) using a
bulk transfer type. The second endpoint is defined as using endpoint
(EP2In) using a bulk transfer type.
Both endpoints will use a packet size
of 64 bytes and request polls whenever there is bandwidth.
SSD
If any optional standard string
descriptors (SSD) are used, the first
descriptor must identify the language
or languages (see Table 2). The 16-bit
LANGIDs2 defines how the host
should present text to user (i.e., “Hit a
key to continue”). The remaining
strings can give the user an ASCII text
description of some cryptic value within the associated descriptor.
In this project, the second SSD (1) is
associated with the Manufacturer ID in
the SDD and the text reads “Microchip
Technology, Inc.” The third SSD (2) is
associated with the product ID in the
SDD and the text reads “Imagine
That (www.imaginethatnow.com).”
RETRIEVE DESCRIPTORS
I reprinted the first page of transactions as captured by the USB 480 Protocol Analyzer in Photo 1 because it
demonstrates how the host interacts
with this device when plugged into the
USB and enumeration begins. We left
off last month with the SIE setting the
transaction complete interrupt after
receiving the set-up token via EP0. The
buffer descriptor (BD0) associated with
default endpoint (EP0Out) pointed us to
the 8-byte EP0Out buffer that held the
packet data received by the SIE. Upon
moving and interrogating the data, it
was found to be requesting 64 characters via a GetDescriptor request (0x80).
The descriptor data we discussed above
has been placed at the top of the program memory space by the assembler,
so we know where it can be found.
You must prepare an 8-byte packet
response for the SIE before clearing the
interrupt. Once you clear the interrupt,
the SIE will acknowledge the SETUP
packet and the host will send an IN
packet to get our data. You best be ready.
To get the requested information,
USB_desc_ptr is preloaded with the offset address of the character we want
from the descriptor table. A call to the
Descriptor routine loads the address of
the beginning of the descriptor table into
a table pointer that will fetch characters
CIRCUIT CELLAR®
•
from the program memory space. The
offset is added to the table pointer, a
character is fetched from that address,
and returned. The character can be then
transferred to the EP0In buffer and the
next character sought. When EpoIn
buffer is filled we can leave the interrupt
routine and clear the interrupt, the SIE
acknowledges the host.
Refer to Photo 1a. The host has continued with an IN token. The SIE will
again set the transaction complete interrupt after sending the data we preloaded
into EP0In buffer. Note that the 8-byte
DATA1 packet of the IN token has the
first eight characters of the SDD. This
time the interrupt routine will discover
an IN token request from the USTAT
and branch to the ProcessInToken
routine. Since I stored the SETUP
token’s GetDescriptor request in
USB_dev_req, I know how I got here and
must continue to preload 8-byte packets
of data into the EP0In buffer until all the
requested data has been sent, I run out
of data to send, or the request is cancelled by a new SETUP token.
As you can see in Photo 1a, notice
that after the host acknowledged the
receipt of the first IN data, it sent an
OUT token. This is an acknowledgment
that the transfer request is complete. It
a)
62
decided it has what it wants. Not only
did it find out what the maximum packet size was (0x08), but also that the SDD
is 0x12 bytes in length (no sense asking
for more than that).
Note that this GetDeviceDescriptor
transfer contained three stages: a setup
transaction (in this case one), a data
transaction, and a status transaction.
Each transaction stage has a token
phase, a data phase, and a handshake
phase (see Figure 1).
ADDRESS STATE
Up to this point, the communication
has taken place on the default address
UADDR=0. The host is now ready to
assign the device a real address. The
host sends a SETUP token. The receipt
of the token by the device’s SIE causes
another transaction complete interrupt.
USTAT reveals it was via EP0Out. By
processing the EP0Out Buffer, we find
the 8-byte packet contains a
SET_ADDRESS request (0x00) and the
new address is 0x06. While you might
be tempted to change UADDR immediately, hold on. Since this is only the
SETUP stage of this transfer, we must
continue using the old address until the
transfer is complete. This is the only
time when you need to finish things
before making use of the data received.
Since this request has no DATA stage,
we just preload a zero-length packet into
the EP0In Buffer and it is used as the
HANDSHAKE stage. We exit the interrupt routine, clearing the interrupt and
the SIE ACK is the SETUP token.
The host sends an IN token (the
beginning of the HANDSHAKE stage),
the SIE receives the request and sends
the EP0IN Buffer, and sets the transaction complete interrupt. USTAT reveals
it was via EP0In. Since we stored the
SETUP token’s SET_ADDRESS request
in USB_dev_req, we know how we got
here. There is nothing expected as this is
the end of the SetAddress request and all
transactions are complete. We are now
clear to change UADDR to 0x06. And,
don’t forget to reset our USB_dev_req to
0xFF (NO_REQUEST). Once we leave
the interrupt routine, clearing the interrupt flag, we have transitioned into the
ADDRESS_STATE.
CONFIGURE STATE
Note that the next SETUP token
(and all subsequent tokens) are sent to
the new address 0x06. Now the device
pays attention only to those tokens
containing the new address. This new
SETUP token is again requesting the
b)
Photo 1a—The Beagle protocol analyzer’s capture of the USB traffic clearly shows the transfer requests of a device beginning the enumeration process. Every device begins with a default address of 0 and is reassigned a new address by the SetAddress request seen here. b—The
enumeration process ends when the host requests that a device be configured to one of its declared configurations. This project has a single
configuration. Once configured, the host will begin polling for information according to the Endpoint Descriptors, as seen at the bottom here
as IN_NAKs to EP2 and EP1.
CIRCUIT CELLAR®
•
Figure 1—This is a
potential sequence
of three transactions
that will complete a
control transfer.
Each transaction or
stage consists of a
Token, Data, and
Acknowledgement
Phase or Packet.
USB Reset
URSTIF
Start-of-frame
SOFIF
From host
From host
To host
SETUP Token
Data
ACK
From host
To host
From host
IN Token
Data
ACK
From host
From host
To host
OUT Token
Empty data
ACK
Set TRNIF
Set TRNIF
Set TRNIF
Transaction
Transaction
complete
RESET
SOF
SETUP
DATA
SOF
STATUS
Differential data
Control transfer*
1-ms frame
* The control transfer shown here is only an example showing events that can occur for every transaction.
Typical control transfers will spread across multiple frames
64
GetDeviceDescriptor. This time it is only asking for
0x12 bytes. The host will continue to gain knowledge about
the device by sending GetConfigurationDescriptor and
GetStringDescriptor requests. Once it knows everything, the host will request a SetConfiguration and finally GetLineCoding and SetControlLineState via the
class-specific requests.
The SetConfiguration request gives the device permission
to use a specific configuration. In this case, we have defined
only one. The configuration uses two interfaces and now we
set them up using EP1In as an ACM interface using interrupt transfers (the control interface) and EP2In and EP2Out
as the data interface using bulk transfers. The class-specific
GetLineCoding request informs the host that our (virtual)
UART is set to receive 8 data bits, with 1 stop bit, and no
parity at a data rate of 115,200 bps. These are values I chose
and stored in initialized variables so that they can be
recalled and changed if necessary. I could set up the microcontroller’s (physical) UART and this project would be a
USB-to-UART interface, but that is an application left up to
you. The SetControlLineState request presents the
device with RTS and DTR states, which are controlled by
the host.
The device is now in the CONFIGURE_STATE. It will be
listed as a (virtual) COMM port in any application running
on the PC that can use a serial port for data communication.
The host starts to poll both interfaces of the device. It is
polling EP1In looking for any control information that the
device might want to convey to the host. It is also looking
for any data on EP2In that the device may want to transfer
to the host. The device is happy with the status quo and
has no information to pass on, so the SIE has not been given
control over any of the IN endpoints. It will NAK all polls,
which tells the host that it’s busy and to try again later. By
“later” I mean (as defined in the standard endpoint descriptor) every two frames (2 ms) for the interrupt endpoint and
whenever there is room for the bulk endpoint.
When an application on the PC selects this COMM port
for use, things begin to happen once again. In my next article, I’ll look at the two conversations that take place
through the configured device’s newly established endpoints,
EP1 and EP2. I
Jeff Bachiochi (pronounced BAH-key-AH-key) has been writing for
Circuit Cellar since 1988. His background includes product design
and manufacturing. You can reach him at jeff.bachiochi@imaginethat
now.com or at www.imaginethatnow.com.
PROJECT FILES
To download the descriptor table, go to ftp://ftp.circuit
cellar.com/pub/Circuit_Cellar/2010/240.
SOURCES
Microchip Technology, Inc. | www.microchip.com
Beagle USB 480 Protocol analyzer
Total Phase, Inc. | www.totalphase.com
NEED-TO-KNOW INFO
For more need-to-know information about topics covered in Jeff Bachiochi’s Issue 240 article, the Circuit
Cellar staff recommends the following content:
—
Create a Hybrid Hub
by Jeff Bachiochi
Jeff shows you how to build a four-port USB hub
with a TPS2071 power controller, a data controller, and
a few other simple parts. Topics: USB, Hub, Power
—
Get Started with PIC USB Connectivity
by Jeff Bachiochi
Get started with USB. Jeff presents MCUs with USB
peripheral support, embedded hosting, and more.
Topics: USB, OTG, HNP Protocol, TPL
CIRCUIT CELLAR®
•
www.DevtoolsXpress.com
Pick a Tool.
Any Tool.
Find the Right Development Tool, Compare it to Other Tools, Evaluate It,
and Buy It from Digi-Key Tools Xpress -- Without Leaving Our Site.
The Digi-Key Tools Xpress
intuitive research engines are
used by engineers worldwide to
locate, compare and evaluate
hardware or software
development tools.
Digi-Key Tools Xpress, engineered
by Embedded Developer, is the
only site in the industry where
engineers can quickly find,
compare and buy the leading
development tools.
Compare before you buy: tools are
listed side-by-side, with features and
performance specs, availability, and prices,
so you can make an educated decision!
FIND. COMPARE. BUY.
S
ILICON UPDATE
by Tom Cantrell (USA)
Wall Whisperer
A Fresh Take on Powerline Communication
Powerline communication isn’t a new concept, but as the Rodney Dangerfield of
networks, it just can’t get respect. So, maybe your experience with early versions
was less than satisfying. Now Cypress is ready to give it another go. How about you?
“H
66
ome Automation” is seemingly
an oxymoron, one of those
“Tomorrow’s Technology of Tomorrow” that
just can’t get off the starting block. I’ve got a
relatively new home, and it’s about as “smart”
as a rock with just a few brain cells (e.g., a
thermostat and a garage door opener).
Half the battle is the skeletons in the closet.
Like a lot of you, I had my
fling with X10 many years
ago. And no doubt, like a lot
of you, it didn’t last long.
The performance and reliability was just too iffy for me
to convince myself to put a
lot of work into trying to do
something meaningful.
The concept of powerline
communication (PLC for
short) took another hit with
the misguided premise of
turning your power company
into an ISP. The concept of
“Broadband over Power
Line” (aka BPL) may have
sounded good at the time,
but as a practical matter, the
idea it can compete with
DSL and cable is fading
away.[1] With its dubious
track record, I haven’t really
Photo 1—Powerline communications is
bothered to keep track of
making a comeback with a new generation
PLC. But taking off the
of high-speed units such as the 200-Mbps
Powerline 200 AV+ from Netgear.
blinders, I can see that there
are some interesting developments afoot. You’ll
see what I mean if you Google acronyms like
“P1901,” “G.hn,” and “UPA.”[2] Or toddle over
to your nearest BUY MORE big box and check
out the latest PLC offerings from networking
heavyweights like Linksys and Netgear (see
Photo 1). Their newest powerline gadgets claim
to deliver up to 200 Mbps, with 500 Mbps and
even 1 G on the drawing board. While it’s generally understood these claims greatly overstate actual throughput, it’s still fast by any
measure. Beyond some metering and “smart
grid” applications, BPL may be dead as far as
your power company is concerned, but the concept would seem to have much better
prospects in smaller scale applications.
This is where Cypress Semiconductor steps
in with three new PLC-capable chips. Let’s
take a closer look and see how they let embedded applications hitch a ride on the power line.
Heck, the wires are already there, so why not
use them?
The Cypress lineup comprises three different
part numbers—the CY8CPLC10, CY8CPLC20,
and CY8CLED16P01—that target somewhat
different applications. As an interesting aside,
the silicon under the hood is the same for the
trio—namely, a standard PSoC, with differentiation provided using its programmable hardware and software capabilities. Let’s start with
the CY8CPLC20 (see Figure 1), which is the
most general-purpose of the trio.
The PLC20 is simple enough to describe: it’s
a PSoC flash memory MCU—specifically, a
CIRCUIT CELLAR®
•
static applications,
returned and you can configure the
but if the network
stack for up to four automatic retries.
FSK_IN
RSVD
2
27
configuration is large
After the address is a “Command”
PO[4], A, IO
FSK_OUT
3
26
and dynamic (i.e.,
byte. Sixteen commands are defined
RSVD
A, I, P0[1]
4
25
number of nodes
(Table 1) with another 32 reserved and
RSVD
TX_ SHUTDOWN
5
24
the rest available for application use.
unknown
and
changP2[5]
6
23
P2[6], EXTERNAL VREF
ing), consider using
After the command are “Payload
AGND
A, I, P2[3]
7
22
SSOP
RXCOMP_IN
A, I, P2[1]
8
21
Length” (0 to 31 bytes), a sequence
the “PLT + network
RXCOMP_OUT
RSVD
9
20
stack” option.
number (to detect lost/duplicate packXRES
12C SCL, P1[7]
10
19
ets) and a header CRC. These are folBehind the Cypress
P1[6]
12C SDA, P1[5]
11
18
lowed by the payload itself and an
network
stack
is
the
P1[4], EXTCLK
P1[3]
12
17
overall packet CRC.
fairly sophisticated
P1[2]
12C SCL, XTALIN, P1[1]
13
16
P1[0], XTALOUT I2C SDA
VSS
14
15
The media access strategy implepacket structure
mented by the stack features “carrier
shown in Figure 3.
sense” and “collision avoidance. Each
Let’s go through it to
Figure 1—With the same silicon under the hood, the new
Cypress powerline communication chips have a similar pinout
node waits until the line is idle, but
see some of the
and are available in 28-pin (shown here) and 48-pin versions.
then further waits a random amount
embellishments the
of time (between 85 and 115 ms)
stack provides, startbefore initiating transmission. WithPSoC-1 (i.e., M8C core)—whose program- ing with addressing. You can see the
out this “random backoff,” every node
network has plenty of elbow room
mable hardware is configured with a
would try to jump in as soon as the
power-line modem (see Figure 2), option- with 8-, 16- and 64-bit address options,
line goes idle and the network would
the latter taking advantage of the
ally fronted with a bit of software impleunique 64-bit ID each chip comes with get paralyzed by collisions.
menting a simple network protocol.
Of course, there’s no free lunch. The
from the factory. There are also
Now before you get too excited
price for the network stack’s reliabili“group” and “broadcast” options that
thinking about hooking up your
ty and ease of use may be worth it,
support one-to-many and many-to-one
HDTV, realize this modem is a simple
but there is a price to pay—namely,
associations.
half-duplex 2,400-bps unit with perthroughput. You can see that in the
The “Service Type” bit up front is
formance that harkens back to the
worst case (64-bit addresses) transferan innocuous name for a key feature
dial-up days. It’s speedy enough for a
ring a single byte payload would
of the protocol, namely acknowledgelot of simple and useful applications.
require sending a 22-byte packet (i.e., a
ment. When enabled, receivers will
Just be careful not to bite off more
22:1 packet/payload ratio). Fortunateautomatically acknowledge receipt,
bandwidth than the chip can chew.
ly, many applications will be able to
transmitters will timeout in half a
When you checkout the power line
use shorter addresses and/or longer
second if the acknowledgment isn’t
transceiver (PLT) object from the PSoC
payloads to reduce the
Designer library, there are
overhead. For instance,
three options to choose
Network protocol
with 8-bit addresses and
from: basic PLT, PLT + net31-byte payloads, the
work software, and PLT +
packet/payload ratio
network software + I2C.
Digital
would approach 1:1 (e.g.,
The basic PLT option
receiver
Digital
1.2 for a 38-byte packet
gives you raw byte-at-aHysteresis
transmitter
comparator
carrying a 31 byte paytime access to the data
Logic ‘1’ or
Low-pass
load). On the other hand,
link, much like a UART.
Logic
‘0’
Local
filter
External
oscillator
keep in mind that interThe automatically generatlow-pass filter
Modulator
Correlator
mittent interference will
ed API allows configuring
Square wave
be more likely to disrupt
the modem (e.g., bit rate,
at FSK
IF Band
a longer packet than a
gain, noise threshold),
frequencies
pass filter
Local
oscillator
shorter one.
checking if the line is busy
Mixer
Programmable
That brings us to the
(i.e., voltage higher than
gain amplifier
HF Band
PLC10 chip, which uticonfigured threshold), and
pass filter
lizes the final “PLT + nettransmitting and receiving
RX
work stack + I2C” option.
bytes by interrupt or
amplifier
In this case, access to the
polling. With this option
PLT and network stack is
you’ll have to handle all
Coupling circuit
rerouted via the external
the network administraI2C interface. You don’t
tion and error handling
yourself. It’s probably OK
have any direct access to
Fi gure 2 —Cypress takes advantage of the PSoC’s programmable hardware to implement a 2,400-bps, half-duplex FSK powerline modem.
for simple and relatively
the on-chip processor.
A, I, P0[7]
VDD
28
•
CIRCUIT CELLAR®
Receiver
Transmitter
Powerline modem PHY
1
67
Introducing Pololu’s new
Maestro USB
Servo Controllers
Conduct a
symphony
of servos.
Bit offset
7
0
6
0
5
0
#1354
Mini Maestro
18-Channel
$39.95
#1356
Mini Maestro
24-Channel
$49.95
Pololu's 6-channel Micro Maestro and
new 12-, 18-, and 24-channel Mini
Maestros take serial servo controllers to
the next level by incorporating native USB
control for easy connection to a PC and
programmability via a simple scripting
language for self-contained, host
controller-free applications. Whether you
want the best servo controller available or
a versatile, general-purpose I/O control
board, these compact devices will deliver.
Three control methods: USB, TTL serial,
and internal scripting
Free configuration and control
application with motion sequencer
Channels can be used for digital I/O or
up to 12 analog inputs
Individual servo speed and acceleration
control for each channel
Up to 8 KB of internal scripting memory
(~3000 servo positions)
0.25 us servo pulse resolution with
pulse rate up to 333 Hz
68
more information at
www.pololu.com/maestro
0
3
x
2
1
x
X
0
Byte
offset
0
X
0x02
1
0x01
2
0x09
xx
#1350
Micro Maestro
6-Channel
$19.95
#1352
Mini Maestro
12-Channel
$29.95
4
3
4
0x03
xx
Xx
5
0xA1
6
0xA2
7
0xA3
8
Xx
9
SA Type
Command ID
DA Type
Payload length
Destination address
Payload
Source address
Service type
Figure 3—The Cypress powerline communicaton stack raises the level of network abstraction
utilizing the packet structure shown here.
Rather the chip acts a self-contained
powerline “coprocessor” working
under the direction of an external
MCU of your own choosing.
Although most PLC headlines
revolve around 120/240-VAC, the
Cypress solution is just as well-suited
for lower voltage AC and DC powerline designs. Enter the
CY8CLED16P01, which targets LED
lighting applications, drawing from a
library of hardware and software functions such as dimming, temperature
compensation, color mixing and
closed loop optical feedback. Another
low voltage application that comes to
mind would be vehicles where data
would ride on the 12-VDC power bus.
Just the thing for moving some bits
bumper-to-bumper and everywhere inbetween (e.g., using lighting and accessory power connections) without
stringing a bunch of extra wire.
AC/DC
“Eavesdropping aliens and the
industrial archaeologists of the future
may well wonder why we built a
phased array across the north of England to beam credit card numbers
and digital images of naked ladies
into the atmosphere.”—Nick Long of
the Low Power Radio Association
(www.lpra.org).
These days, silicon is so
advanced
that most designs
Command ID Command Name
involve
little
more than slapping
0x01
SetRemote_TXEnable
down
a
few
Wunderchips
like Lego
0x03
SetRemote_ExtendedAddr
blocks.
And
so
it
is
with
the
0x04
SetRemote_LogicalAddr
Cypress chips which handle all
0x05
GetRemote_LogicalAddr
manner of applications with
0x06
GetRemote_PhysicalAddr
aplomb. But when it comes to
0x07
GetRemote_State
actually getting the data on and off
0x08
GetRemote_Version
a high-voltage powerline, there’s
0x09
SendRemote_Data
0x0A
RequestRemote_Data
still a need for some old-school
0x0B
ResponseRemote_Data
circuitry to deal with unique safe0x0C
SetRemote_BIU
ty and regulatory issues. Fortu0x0D
SetRemote_ThresholdValue
nately, Cypress has an excellent
0x0E
SetRemote_GroupMembership
app note that addresses the subject
0x0F
GetRemote_GroupMembership
in great detail.[3] Here are some of
the highlights.
Tabl e 1 —The network stack option eases applicaMaking “The last inch” contion software development by providing a number
of high-level commands.
nection involves three functions:
CIRCUIT CELLAR®
•
the Cypress high-voltage CY3274
development kit. More on the kit in
a moment; but for now, just note the
nontrivial collection of transformers,
op-amps, resistors, capacitors, diodes
and such (more under the board)
required to power-up your data.
Before moving on, I should say a
few words about one controversial
aspect of PLC alluded to in the quote
above—namely, potential pollution of
Accuracy
Sensitivity
transmit filtering and amplification,
receive filtering, and electrical coupling. And, of course, there’s also the
usual power supply circuits (e.g.,
120/240 VAC to 5-V DC switcher) to
power the Cypress chip itself along
with any other digital logic.
Put it altogether and the powerline
interface turns out to be the tail wagging the chip. A picture is worth a
thousand words and Photo 2 shows
Great Value in Test & Measurement
Photo 2—A pair of CY3274 kits makes it easy (if not cheap at $624 each) to evaluate the
new Cypress powerline communication chips. The kit includes a programmer for downloading
code and an I 2C “wiggler” (attached to the board on the right) for exercising the chip’s PLC10
I 2C co-processor mode.
from
$4,422
250MHz/350 MHz 2/4 Channel
Oscilloscope HMO Series
DSO/MSO with 8/16 logic channels
(Mixed Signal Opt. HO3508)
Serial Bus Trigger and decode
I2C, SPI, UART/RS-232 (Option)
Pass Fail Test based on Masks
Low noise fan
from
$1,419
Programmable 2/3/4 Channel
Power Supply HMP Series
HMP2020: 1x 0…32V/0…10A
1x 0…5.5V/0…5A
HMP2030: 2x 0…32V/0…5A
1x 0…5.5V/0…5A
HMP4030: 3x 0…32V/0…10A
HMP4040: 4x 0…32V/0…10A
from
$3,581
1GHz/3GHz Spectrum Analyzer
HMS Series with and without TG
Frequency Range 100kHz…3GHz
Measurement Range -114…+20dBm
DANL -125dBm/-135dBm (Preamp.)
Resolution Bandwitdh 100Hz…1MHz
Standard EMI RBWs (-6dB) included
6.5“ TFT VGA Display, DVI Output
Quality
$2,011
20Hz…200kHz
Basic Accuracy 0.05%
Functions:
L, C, R, |Z|, X, |Y|, G, B, D, Q, , Δ
Programmable U & I Bias
HAPRO Electronics
Photo 3—The Cypress chips aren’t really “new.” They’re standard PSoCs upgraded with powerline communication features using the chip’s programmable hardware and software capability, all under the purview of the familiar PSoC Designer tool.
•
Tel: +1-516-794-4080 · Fax: +1-516-794-1855
[email protected]
www.haproelectronics.com · www.hameg.com
CIRCUIT CELLAR®
Simplicity
LCR-Bridge HM8118
69
Photo 4—Configured as a PLC10, access to the powerline is via an I2C interface. The kit includes
“PLC Console” software that utilizes the included I2C “wiggler” to evaluate that capability.
the radio spectrum. Amateur radio
operators have been raising the red
flag for years and have kept regulatory
agencies’ (e.g., FCC) feet to the fire.
Presumably, the demise of the fullfledged BPL concept (i.e., power company as ISP) will provide some solace
for those concerned. Hopefully, limiting the reach (i.e., between walls vs.
overhead lines) will mitigate the spectrum pollution concerns as well.
Cypress is quick to point out their
design is compliant with a variety of
emission, immunity, and safety regulations and the fact they’ve incorporated specific features (e.g., transmit filtering) to try to be a good neighbor.
Now, as we’ve seen on Wall Street,
the fact something “meets regulations” doesn’t mean it’s not going to
cause problems if the regulations
and/or regulators are inept. But that’s
not a “technical issue,” it’s a political
one. Those who are concerned are
fully entitled to stand up for their
rights, but the place to hold the
protest rally is Washington, DC, not
Silicon Valley.
70
SHOCK IT TO ME
Hopefully, the clever phrase you just
read will get your attention. Needless
to say, if you’re going to be playing
around PLC, don’t! “Playing around”
isn’t something you do with a mains
voltage. It’s OK to poke willy-nilly at
digital logic, because if something goes
wrong, the chips lose. Make a mistake
with 120/240 VAC and you lose.
Turning our attention back to the
Cypress CY3274 development kit,
another “shocking” aspect becomes
apparent, namely sticker shock. These
puppies (mains- and low-voltage versions are available) are a whopping
$642 each not to mention that you
need at least two of them to do anything. Bracketing the CY3274 kit are
some simpler, and less costly ($299)
demo boards and at the top end ($951)
a CY8CLED16P01 kit with LED
daughter card.
The price is high, but at least the
CY3274 kit is replete with features.
The built-in breadboard and LCD
make for instant out-of-the-box gratification and easy prototyping. Recalling the difference between the PLC20
and PLC10 chips is a matter of programmability, the kit can support
development for both.
For developing PLC20 apps you utilize the popular and well-proven PSoC
Designer tool, with hardware and software libraries updated to support the
powerline transceiver module (see
Photo 3). The kit includes a USB plugin programmer to burn your own code
into the PSoC, noting that a fullfledged debugger is an extra cost item.
Using the programmer, you can
download the aforementioned “PLT +
network stack + I2C” configuration to
make the PLC20 on the board act like a
PLC10 co-processor under the direction
of an external host via an I2C connector
on the edge of the board. A very nice
touch is that the kit also includes a
USB-to-I2C wiggler and powerline control panel software for the PC. With
two kits and their wigglers you can
plug into two power outlets, two USB
ports and run two copies of the control
panel to exercise the network from the
comfort of your PC (see Photo 4).
The quick-start guide has you wire
up the LCD and then program each
board with a simple PLC20 application, one board as a transmitter and
one as a receiver. When you push a
button on the transmitter it reads the
on-board DIP switch and sends the
data to the receiver, which in turn displays the received data on its LCD.
I’m all for keeping it simple, but
Cypress could have made this demo a
little more useful. As an initial experiment, I wanted to do a quick site survey of my house wiring. Exercise is a
good thing, but the prospect of running
up and down stairs pushing the switch
on the transmitter and then checking
the receiver, for each outlet no less,
wasn’t appealing.
Fortunately the CD in the kit has all
the demo projects and software (see
Listing 1). It’s testimony to PSoC
Designer that even though I haven’t
used the tool in awhile, it was a nobrainer to tweak the demo code to
have the transmitter repeatedly send a
unique message without having to
push the switch. Now I could make
just make one pass through the house
with the receiver in hand to check
reception at each outlet.
Next, I reprogrammed the boards to
act like PLC10s and used the two I2C
wigglers with two copies of the control
panel software to get a little deeper
under the hood. Notably, the control
panel allows you to tweak key parameters, such as gain and noise threshold,
to see their effect.
So how’d it all work out? Quite well,
I’d say. But note that an exhaustive and
authoritative test would be a much
bigger job. So let me just tell you what
CIRCUIT CELLAR®
•
I saw, but as well what I didn’t see (i.e., issues a real-world
designer would have to question, test and design around).
Right off the bat, you’ll have to deal with the reality that
there’s good chance that two particular outlets won’t be able
to talk to each other. At least that’s true for locales that use
“split phase” distribution, which brings, for example, 240 V
to the panel and then splits it into separate 120-V legs. Sure
enough, my simple walk-around “Can you hear me now?”
test confirmed some half the outlets were inaccessible from
the other half. Interestingly, and a bit of good news, was the
fact that the legs seemed rather well intermixed—for example, between floors and side-to-side in my two story house.
There are a lot of AC outlets in a modern house—maybe
four to six per room (which is a big advantage over the alternative of piggybacking on phone wires and coax)—and it
generally seemed that if one outlet was dead a nearby one
worked. But your mileage may vary, and unless you’ve got a
wiring diagram, your safest option is to try before you buy.
A related gotcha is that all outlets aren’t created equal,
the irony being a “better” outlet in terms of power delivery
may be worse for communication. Sure enough, plugging
my test setup into a UPS with surge protector confirmed it
was real good at killing ones and zeros as well as surges.
That’s something to consider if a particular installation is
prone to lightning or otherwise destructive power spikes.
And what about power failures? Interestingly, I don’t see
any reason communication couldn’t continue which isn’t
the case for schemes, such as X10, that rely on zero crossing. Of course, there’s the catch-22 that since you can’t use
a UPS each gadget would need a battery, a dubious solution
counter to the fundamental premise. Better just figure if
the power goes off, so does you application.
Presuming you’ve got plain-vanilla outlets that can see
each other, the next question is how well the link works.
Show typical house wiring, with its ad hoc layout, un-terminated stubs, power strips, extension cords, etc. to a network guru and they would laugh out loud. The topology is
very inhospitable and things only get worse when you consider all the sources of noise such as motors, fluorescent
lights, dimmers, and so on.
While the Cypress network stack brings a lot of features
to the table, it doesn’t do everything. There is one API call
that automatically tests the line noise level and sets the
noise threshold for determining if the line is “busy.” But
beyond that your own software will have to handle features
such as adaptive gain settings. Another embellishment
would be security with the understanding that signals may
leak between properties that share a transformer.
Next I used PSoC Designer to further tweak the demo
software to track packet errors and display the count on
the LCD. I set up the units at opposite corners of the house
(bottom floor left, top floor right) and let it run overnight.
Checking next morning I discovered the unit had detected
a grand total of 10 errors, about one per hour. That’s quite
impressive considering I was sending an 8-byte packet (8bit addressing, 1-byte payload) about once per second,
making for an overall packet error rate of less than 0.01%.
Switching back to the PLC Console mode (i.e., using the
I2C wigglers) confirmed similarly good results, at least with
a simple one-way test, but digging a little deeper revealed
some issues. On the receiver side, somewhere along the
line between the I2C connection, USB and the PLC Console
void main()
{
BYTE bTemp;
LCD_Start(); // Start the LCD Module
PLT_Start(); // Start the Powerline Transceiver
TX_LED_Start();RX_LED_Start();BIU_LED_Start(); // Start the powerline comm. LEDs
// Set the baud rate to 2400 bps with a 3 kHz deviation
PLT_Memory_Array[Modem_Config] = (Modem_TXDelay_7ms|Modem_FSKBW_3M|Modem_BPS_2400);
PLT_Memory_Array[TX_Gain] = 0x0b; // Set the TX Gain to 0.375x
PLT_Memory_Array[RX_Gain] = 0x00;
// Set the logical address of the PLC20 Receiver board to 0x01
PLT_Memory_Array[Local_LA_LSB] = 0x01;
// Enable the TX and RX and allow the RX buffer to be overwritten
PLT_Memory_Array[PLC_Mode] = (TX_Enable|RX_Enable|RX_Override);
LCD_Position(0,0); LCD_PrCString("PLC20 RX");
while (1)
{
PLT_Poll(); // Run the network protocol to update memory array
bTemp = PLT_Memory_Array[INT_Status];
if ((bTemp & Status_RX_Data_Available) && (PLT_Memory_Array[RX_CommandID] ==
CMD_SENDMSG) && (PLT_Memory_Array[RX_SA] == 0x02))
// Process the next block of code only after a new message is received
{
LCD_Position(1,0); LCD_PrCString("RX Data = 0x");
LCD_PrHexByte(PLT_Memory_Array[RX_Data]); // Display the data received on the LCD
PLT_Memory_Array[RX_Message_INFO] = 0x00; // Reset the RX Message Info variable
}
}
}
•
CIRCUIT CELLAR®
Listi ng 1 —Taking advantage of the Cypress network stack API makes it easy to send and receive data via the powerline. Shown here is
the receive routine utilized by the demo software that comes with the CY3274 kit.
71
Edition
A
S
U
r
o
t
k
Ele
usa
/
m
o
c
.
r
o
t
k
www.ele
“Elektor? Prescribed reading for
our R&D staff because that’s where
we need professional guidance for
microcontroller technology.”
– Frank Hawkes, 39, development engineer –
[Analog Digital Microcontrollers & Embedded Audio Test & Measurement ]
January 2010
US $ 7.95 - Canada $ 7.95
[Analog Digital Micro
controllers &
Embedded December 2009
Audio Test &
Measurement
]
US $ 7.95 - Canada
$ 7.95
[Analog Digital Micro
controllers &
Embedded www.elektor.com
January 2010
Audio Test &
Measurement
]
US $ 7.95 - Canada
$ 7.95
www.elektorusa.com
There’s More
Preselector
forthan just USB!
Life
Elektoto
r SD
R
ded Audio ollers & Embed
l Microcontr
[Analog Digita
February 2010
ent ]
Test & Measurem
$ 7.95
US $ 7.95 - Canada
Her
e’s the
USB
Magic Eye
automat
ic
tuning upgrade
!
Put a Stop to Throwawayism!
www.elektor.c
om
Buses
USB and I2C
Energybeyond
Sav
om
www.elektor.c
Subscribe to Elektor now!
ings from
Home Autom
ation
Fourier Analysis using
Standards com
pared
LTspice & Excel
Tech
ATM18
There’s Mor
nolo
gy
to
Life
than just USB!e
Proj
N
ACKDOW
Dimmer
with a Micro
a Sto
PFCPutCR
ects
Logic
Analyzer
USB Magic
Eye
Throw
led!
eayism
Christmas Hol p toto
awa
rev
facs Cirrcuit
Buses mp
riday
we
po
13 pag
yo
s
esnd
La be
of beg
US
Br’san
inne
andd
I2C
junk
box projects
My First AVR-USB
ER
olsPOW ing
[
Fou
TAwTH
S
rieerToE
An
ign
aly
des
VEingGO
,NEW
DIY
U’18
lingsis
usi
The
YO
ngPROJECT
ATM
rc , coo
ling
ard
trol
7
u
con
0
follo
,
w
LTs
o
picthe
S
e story
ds on test
l
Analogue Digita
Embedded controllers &
Audio Micro
nt ]
Test & Measureme
Join the fascinating world
of electronics worldwide!
February 2010
2pen
Han
Lo
gic AnalOyz
TS
6 NEW PROJerEC
ty
paci
ery Ca
Batt ker
Chec
& Excel
oth
ueto
TTL Blgle
Don
ain
ke Ch
orbi
Mot
elektor.com
stem
PA Sy
LET THE
SCEPTER
RULE! 7
A 32-bit ARM g
fast prototypin
system
Creative & fresh articles since 1961
applications
design for AVR
Low power
e
h DMX Interfac
LED Mixer wit
Chip
Linux on a
ar
Parking Rad
...
Ultra-compact
ULO D
MOD
amplifier
fledged stereo
$ 7.95
US $ 7.95 - Canada
A full-
PUZZLE
H ELECTRONICS
WIN $100 WIT
Electronics projects to delight and educate in
diverse areas of electronics
ues
11 Iss the
ing
includ
e
doubl
r
e
m
t
sum
or jus
f
e
u
s
is
5*
$39.9
Each project is built, tested, and approved by
Elektor’s laboratory staff
Access to a wide range of products to support
magazine projects in our online shop
Cheaper than 11 issues from the newsstand:
Save 57% of the cover price of $92.45
* Offer available in US and Canada only. Canada add $11 per year.
Take out your subscription now:
XXXFMFLUPSDPNVTBt1IPOF
electronics worldwide
electronics worldwide
GUI things were getting bogged down to the point that the
receiver couldn’t keep up with even a few incoming messages per second. I noticed the little “CPU tachometer”
widget on my PC screen was redlining, something that
almost never happens. Without turning on the software
option that allows received packets to be overwritten—or,
alternatively, the option to insert a delay between packets
on the transmitter side—the receiver couldn’t keep up.
Finally, I performed a pseudo-full-duplex stress test by
having the nodes send and receive messages to each other
simultaneously, a scenario which caused the number of
errors to skyrocket. One explanation might be that the collision avoidance scheme isn’t sophisticated enough to deal
with such blatant abuse. But it just as well could be issues
with the PC/I2C/GUI lash-up as described above.
Clearly, a real evaluation of performance and reliability
will be better served with a dedicated test configuration.
For a PLC20, that would be native code test routines running on the internal PSoC, while for the PLC10 it would be
diagnostic software on an external MCU. Bottom line, my
experience is a reminder to keep your expectations in
check understanding performance is limited, the link isn’t
perfect, and errors will occur.
closed loop acknowledgement capability at least you’ll
know when a packet is lost. Detected errors are a chore,
but undetected ones are a showstopper.
Although less obvious, I’m thinking the sleeper opportunity is non-mains applications such as low-voltage lighting,
vehicles, and indeed anything with a power cord.
Power, and data, to the people! I
Tom Cantrell has been working on chip, board, and systems
design and marketing for several years. You may reach him
by e-mail at [email protected].
REFERENCES
[1] J. Leydon, “Powerline Networking Firm Suffers
Brownout,” The Register, 2002, www.theregister.co.uk/
2002/03/28/power_line_networking_firm_suffers/.
[2] V. Oksman and S. Galli, “G.hn: The New ITU-T
Home Networking Standard,” IEEE Communications
Magazine, 2009.
[3] J. Hushley, “Cypress Powerline Communications
Board Design Analysis,” 001-55427, Cypress Semiconductor, 2009, www.cypress.com/?docID=18145.
SOURCE
ONE WIRE WORLD
When it comes to mains powerline communication,
the Cypress solution is definitely a step up from something like X10. Yes, it isn’t perfect, but thanks to the
CY8CPLC10, CY8CPLC20, and CY8CLED16P01
Powerline communication chips
Cypress Semiconductor | www.cypress.com
NEED-TO-KNOW INFO
—
Power Line Modems Meet Home Control
by Brian Millier
A poorly designed control system prevented Brian’s
heat recovery ventilators from being effective. So, he
tried powerline tech. Topics: Powerline, FSK, Control
•
CIRCUIT CELLAR®
covered in Tom Cantrell’s Issue 240 article, the
Circuit Cellar editorial staff highly recommends
the following content:
—
Power Analysis Primer
From Powerline Measurements to PFC
by Robert Lacoste
Familiarize yourself with PFC. Robert explains
how to measure the power consumption of linepowered devices. Topics: Powerline, PFC, Load,
Power Measurement
73
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Across
74
3.
8.
9.
11.
13.
16.
17.
18.
19.
20.
21.
MIPS means millions of these per second
Brightness
QAM, FM, AM
.MI
A lux is what per square meter?
Mega binary
Java app for your server
Valueless character
1 billion FLOP
MIT, 186
Code parse
Down
1.
2.
4.
5.
6.
7.
10.
12.
14.
15.
8-GB USB flash drive holding every issue if Circuit Cellar
VM [two words]
EQ
Off the motherboard, but connected
Tera binary
Bps [two words]
Ground pi
Group larger than a sector
IEEE 139
Unavailable, outdated
The answers will be available in the next issue and at
www.circuitcellar.com/crossword.
CIRCUIT CELLAR®
•
I
DEA
BOX
THE DIRECTORY OF
PRODUCTS AND SERVICES
AD FORMAT: Advertisers must furnish digital files that meet our specifications (www.circuitcellar.com/advertise). ALL TEXT AND OTHER ELEMENTS MUST
FIT WITHIN A 2" x 3" FORMAT. E-mail [email protected] with your file or send it to IDEA BOX, Circuit Cellar, PO Box 180, Vernon, CT 06066.
For current rates, deadlines, and more information contact Peter Wostrel at 800.454.3741, 978.281.7708 or [email protected].
The Vendor Directory at www.circuitcellar.com/vendor/
is your guide to a variety of engineering products and services.
I/O-Rich SBC Supports
Embedded Ethernet/WiFi
PDQ Board with
The PDQ Board Ethernet module
a low cost 2.5x4
C-programmable single-board computer
Fast 16-bit Freescale HCS12 processor
Ideal for data acquisition and control
Embedded dynamic web server
Email program-controlled messages from
your instrument
Plug-in modules for any kind of I/O,
including GPS, A/D, D/A, UART, USB,
AC and DC relays, and many more...
Mosaic Industries Inc.
tel: 510-790-1255 fax: 510-790-0925
www.mosaic-industries.com
•
CIRCUIT CELLAR®
75
Rowley CrossWorks
C/C++ Compiler, IDE, Debugger,
and JTAG programming tools in
a single professional package
Board and CPU support for NXP,
STMicroelectronics, Atmel, TI, and
many other microcontrollers
Includes royalty-free tasking library
Come and see what we have!
76
Mention Circuit Cellar when you
buy and well gladly send you a
JTAG adapterfor free!
RowleyAssociates.com
CIRCUIT CELLAR®
•
XL- MaxSonar
Ultrasonic Ranging is EZ
XL-MaxSonar Products
• High acoustic power • Low cost
• Low power, 3V-5.5V, (< 4mA avg.)
• 1 cm resolution • Serial, pulse
width, & analog voltage outputs
• Real-time auto calibration with
noise rejection • No dead zone
XL-MaxSonar-EZ
• Choice of beam patterns
• Tiny size (<1 cubic inch)
• Light weight (<6 grams)
XL-MaxSonar-WR (IP67)
• Industrial packaging
• Weather resistant
• Standard ¾” fitting
• Quality narrow beam
www.maxbotix.com
•
CIRCUIT CELLAR®
77
! "
)! "
%
)
$ "*
# $%$
Tools for Embedded Development
CROSSWARE
®
$& ' ($! " ( $ %! ®
ARM
8051
COLDFIRE
®
Three quality tool suites
C/C++*
G Code Wizards
G Debugging
G Simulation
G Support
G
* Embedded C++ for ARM and
ColdFire only
Advanced software
tools since 1984
www.crossware.com
[email protected]
CROSSWORD ANSWERS from Issue 239
78
Across
1. OBFUSCATE—Purposely obscure
your code to protect it
4. UNPLUG—Disconnect, both literally
and figuratively
5. POSITIVE—Cation charge
6. HEATSINK—Prevents overheating
10.FIELDPROGRAMMABLE—Can be
tweaked post-manufacture [two words]
12.PORT—Point of I/O
13.RELAY—Closes and opens contacts
15.LEGACY—Old equipment
16.TESLA—Flowing water, Niagara Falls
19.SERVICEPACKAGE—A bundle of
software updates [two words]
20.TESTMODESELECT—TMS pin [three
words]
Down
2. SANJOSE—Embedded Systems Conference, April 2010
[two words]
3. OXIDE—Flux removes
7. IMPURITY—Dopant
8. DUKE—Java’s character
9. TAPE—Plastic strip for data
11. EXTRINSIC—Semiconductor
with impurities
14.ANEMOMETER—For wind
velocity
17.ACIDIC—pH < 7
18.VOLT—Common EMF Unit
1
2
O B F U S
A
4
U N
J
5
P O
S
6
H E
10
F
12
P O
16
T
19
S E R V
20
T
3
C A T E
O
X
P L U G
I
D
S I T I V E
7
A T S I N K
M
P
U
11
I E L D P R O G R A M M
X
I
13
R T
T
R
15
R
L E G A C Y
I
N
E S L A
17
I
A
I C E P A C K A G E
I
E S T M O D E S E L E C
I
C
CIRCUIT CELLAR®
•
8
D
U
9
T
K
A B L E
P
14
E L A Y
N
E
M
O
18
V
M
O
E
L
T
T
E
R
Page
The Index of Advertisers with links to their web sites is
located at www.circuitcellar.com under the current issue.
Page
Page
Page
76
AAG Electronica, LLC
31
EMAC, Inc.
13, 32
JK microsystems, Inc.
51
PCB-Pool
31
AP Circuits
39
Electronica 2010
75, 78
JK microsystems, Inc.
C4
Parallax, Inc.
75
All Electronics Corp.
77
76
52, 53
Elektor
23
Jameco
10
PoLabs
Apex Embedded Systems
72
Elektor
30
Jeffrey Kerr, LLC
68
Pololu Corp.
ATRIA Technologies, Inc.
65
Embedded Developer
11
Keil Software
76
Rowley Associates
CC Gold
25
ExpressPCB
77
Lawicel AB
35
Saelig
33
CWAV
75
FlexiPanel Ltd.
13
Lemos International Co. Inc.
51
ScanTool.net, LLC
12
Cleverscope
32
Grid Connect, Inc.
77
MCC (Micro Computer Control)
21
Comfile Technology, Inc.
69
HAMEG
77
Maxbotix, Inc.
78
Crossware Products, Inc.
30
HobbyLab, LLC
77
Mental Automation
77
Tern, Inc.
76
Custom Computer Services, Inc.
ICbank, Inc.
76
microEngineering Labs, Inc.
57
Total Phase, Inc.
75
Decade Engineering
Imagineering, Inc.
75
Mosaic Industries, Inc.
76
Trace Systems, Inc.
77
DesignNotes
30
Intuitive Circuits LLC
19
Mouser Electronics
43
Triangle Research Int’l, Inc.
C3
Digi International
75
Ironwood Electronics
C2
NetBurner
73
V-Module
7
P
61, 63
1
REVIEW
of August Issue 241
2, 3
Technologic Systems
The 4-to-20-mA Current Loop
Build an X10 Controller (Part 2): The Controller Program and Utility
THE CONSUMMATE ENGINEER Embedded Safety
THE DARKER SIDE A Tour of the Lab (Part 1): Time Domain Measurement Equipment
ABOVE THE GROUND PLANE Crystal Properties: Circuit Models, Measurement, & Conversion
FROM THE BENCH Application Communication with USB (Part 3): Assembly Code Finale
SILICON UPDATE MIPS and More
September Issue 242
Deadlines
Space Close: July 14
Material Close: July 21
Advanced USB Design Debugging
•
SparkFun Electronics
A TTENTION A DVERTISERS
Theme: Embedded Development
5
Theme
Data Acquisition
Bonus Distribution
PCB West; Embedded Systems
Conference East
Call Peter Wostrel
now to reserve your space!
800.454.3741 or 978.281.7708
e-mail: [email protected]
CIRCUIT CELLAR®
I
NDEX OF
ADVERTISERS
79
PINTERRUPT
RIORITY
by Steve Ciarcia, Founder and Editorial Director
I
80
72 Flavors and More
don’t wish to be overly critical all the time—well, actually, I’m not losing any sleep over being a curmudgeon ;-)—but have
you tried to figure out what the heck they are doing over at Intel these days? Certainly, the people in the marketing department
who came up with their latest processor numbering system must be smoking something stronger than tobacco. I’m sure I’m
not the first who thinks that selecting a new Intel chip is a very daunting process.
I’ve been sitting on the fence for a long time when it comes to an operating system. I’ve been using XP Pro for years and
saw no logical reason to “downgrade” performance to Vista when Microsoft tossed that lemon in the ring a few years ago.
Instead, I continued upgrading hardware, even going so far as having a quad-core laptop, but still using XP Pro as my OS.
Now, however, I think Windows 7 may be a reasonable next step even without Microsoft’s continued threats of XP abandonment. But, unlike the cars that I tend to purchase with no more than 10 minutes of procrastination, I find myself researching
computer selection to the point that it might take months. It must be the difference between a purely emotional decision and
overly intellectual analysis.
The first consideration in buying a new computer these days is deciding whether to choose Core i3, i5, or i7 (and I heard
there is an i9 in the pipe someplace). The logical assumption is that the higher the family number the more overall performance it has and that the numbers following the family designation indicate a capability increment in that family (such as a Core
i5 670 benchmarking faster than an i5 650). The answer is: Well, sorta, but not necessarily. It turns out that many of the processors within the same family have different features and, in the end, what you get for performance depends a lot on what kind
of software you are trying to run!
What might have been three simple family designations is confusing the heck out of me since the numbering scheme incorporates three different design processes, dual, quad, and hex cores, hyper-threading or not, turbo-boost or not, on-die graphics or not, and various cache quantities and memory bandwidths. Figuring out all the players is a game in itself.
The Core i3 5xx series dual-core (32-nm Clarkdale) processors are considered “basic level” but aren’t slouches. They incorporate on-die graphics and hyper-threading, and they’re considered equivalent in processing power to the then top-of-the-line
quad-core Q6600 I bought a few years ago.
The Core i5 6xx series dual core (32-nm Clarkdale) processors seem a bit faster. They have on-die graphics and hyperthreading, but they also include something called turbo-boost. This is a process that automatically runs the cores faster than
the base operating frequency if the die operating power, current, and temperature are below specification limits. These processors appear equivalent to current midrange Core 2 quads and high-end Phenom II processors.
From then on, it is straight downhill. You’d think that the Core i5 7xx series should be an incremental upgrade—but, no, it’s a
different animal. The Core i5 7xx series is a quad-core (45-nm Lynnfield) processor with turbo-boost but no hyper-threading and
no on-die graphics. Not enough features? Go to the 8xx series and you add turbo-boost and hyper-threading back in the mix.
If you weren’t confused enough, there is the Core i7 8xx and 9xx series quad-core processors (45-nm Bloomfield). They
have similar attributes as the other families with the addition of an unlocked multiplier that allows over-clocking for the power
gamers in the crowd. To make the nomenclature even more confusing, the 9xx series isn’t even consistent within its own designation series. A Core i7 975 is a quad-core, but an i7 980X is six cores. Atta boy, Intel!
CPU benchmarks, worshipped by fringe elements who think a liquid nitrogen tank next to the desktop is a necessary part
of life, are limited. Nonetheless, they are still useful enough to give us some peace of mind in performance/dollar comparisons.
For example, here are the Passmark CPU scores for some that I’ve mentioned: Core 2 Duo T5500 (911), Q6600 (2959), Core
i3 540 (2851), Core i5 750 (4207), Core i7 950 (6288), Core i7 975 (7011), and Core i7 980X (9955). There can be a lot of
architectural overlap, so be very careful of the hype in prepackaged computers.
So, where does that leave me? Better informed, but not any closer to making my next computer acquisition. The last time I
bought a new computer, I approached it the way I buy cars—get something fast and then negotiate driving it slower. The result
was a 12-lb, quad-core laptop with a 300-W power supply! Certainly, I only think of it as a “transportable,” but perhaps the idea
of stuffing 10 lb of processor in a 5-lb laptop has exceeded trendy effectiveness.
After quoting benchmarks to you, I have to admit that they are meaningless for most of the stuff I do anyway. It sounds wonderful to simply toss in an i7 980X—but at $1,000 each, I’d have to be nuts. Buying a computer with any one of these new
processors will instantly upgrade me to Windows 7, but that certainly isn’t going to result in a 3× or 5× speed boost over my
present system, regardless of the benchmarks. I doubt it will do a lot for the same old non-threaded application software I’ve
been running for years. I guess that means I have to think about the road ahead a bit longer.
[email protected]
CIRCUIT CELLAR®
•
XBee
®
7KH0RVW)OH[LEOH=LJ%HH3ODWIRUP
: ZigBee PRO interoperability
: Out-of-the-box RF communications
: Common XBee footprint makes substituting modules easy
NEW! Programmable ZigBee modules
: Optimized for ZigBee application profiles including
Smart Energy
: iDigi™ Energy solution bundle
: Complete solution: Modules, adapters, gateways, services
The industry’s first ZigBee Smart Energy modules
'LJLVMMLYZ[OLPUK\Z[Y`»ZTVZ[JVTWSL[LZL[VMOHYK^HYLHUKZLY]PJLZMVY`V\YAPN)LLLUHISLK
ZVS\[PVU?)LLWYVK\J[ZZ\WWVY[H^PKLYHUNLVMHWWSPJH[PVUZPUJS\KPUN:THY[,ULYN`/VTLHUK
)\PSKPUN(\[VTH[PVU9LTV[L+L]PJL4HUHNLTLU[HUKTVYL>P[OHOVZ[VMTVK\SLZZ[HUKHSVUL
Gateways
HKHW[LYZ07LUHISLKNH[L^H`ZHUKP+PNPKH[HZLY]PJLZ+PNPOHZ[OL[VVSZ[VZLL`V\YWYVQLJ[
[OYV\NOMYVTILNPUUPUN[VLUK
Modules
Get started today
with an XBee
Development Kit
Digi International
Tel. 1-877-912-3444
Adapters
www.digi.com/getXBee

Ubuntu Webcam Server C Language Rules and Assignments USB

Transcription

Similar documents

EMUKEY Product Sheet

What does it measure? The Visioscan® VC 98 USB is a special UVA

TB-CDP02-TX/RX

portable usb dj mixer + scratch controll er + dj software

SC-2300 spec

Low Cost Pure Sine Wave Solar Inverter Circuit

www.globo.com.pl OPTICUM HD 9600 PRIMA id: O9600PRIMA

Technical specification

RIES - Szablon - Réseau Rural Auvergne

Thevenin`s and Norton`s Equivalent Circuit