FSM

Comments

Transcription

FSM
DCS status and experience in the
system test
Joachim Schultes
University of Wuppertal
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
Outline
1. short description of system
• HW
• SW
• system overview
2. General Experience + Special Issues
3. conclusion
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
Distance from interaction point
The DCS Hardware
Environm.
Module
HV
VDD
T
VDDA
Optoboard
Sensors
VVDC
Regulator
Station
VPin VISet
Data
T
Cover
T
BBIM
BBM
HV-PP4
Iseg
Interlock
LV-PP4
Wiener
System
SC-OLink
Data
Door
BOC
CAN-Open protocol
TCP/IP
CAN-Open protocol
DCS-PCs
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
HW
HV:
2 Iseg crates (1 channel/Detector-module, 1 iseg module/sector)
• 16 modules EHS
• 8 modules EHS + EHQ
LV:
WIENER ( 2 channels/sector)
• 4 crates, 12 channels each
PP4 (½ ELMB/sector)
• 4 crates à 3 ELMBs à 4 Wiener channels
Regulators (1 board/sector)
• 2 crates with 12 Boards each (+ 1 „old“ crate, 2 Boards)
Optolink:
SC-OLink (1 complex channel/sector)
• 2 crates à 4 ELMBs, 4 complex channels per ELMB (Viset, Vpin, Vvdc)
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
HW, II
Environment:
BBM
• 1 crate à 3 ELMBs (2NTC, 1HS)
Interlock:
BBIM (½ ELMB plug/sector T_mod + 1/2 ELMB plug/6 T_opto (+ T_PP2))
• 2 crates à 4 ELMBs à 4 plugs
BOC-I-BOX (5 racks -> up to 2 crates -> 12 BOCs -> 1-4 sectors)
• 1 crate à 1 ELMB
PP1-BOX
• 1 crate à 1 ELMB
LU
• 2 LU à 4 ELMBs (3 Tmod, 1 FPGA (TOpto, TPP2, IPP1, IBOC, IDSS))
IDB-x
• 1 IDB-SC for 5 SC-OLink crates (5 * 4 * 4 = 80 ELMB channels)
• 1 IDB-LV for 6 Wiener-crates (6 * 12 = 72 ELMB channels)
• 1 IDB-HV for 5 iseg slots (5*2*8 = 80 ELMB Channels)
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
SW
FITs: functional order
SIT: geographical order
FSM: hierarchical
DDC:
DAQ-DCS communication
FSM commands + set voltages + Temps
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
overview of systems in SR1, I
dcs1
dcs2
dcs3
_DriverConnection
_DriverConnection
ATLASPixDCS [3]
Peak (iseg)
KVASER (ELMBs)
Data, Event, UI
Iseg OPCServer
CANopen OPCServer
Archiving (RDB)
PVSS: OPCClient (num 3)
(-data dcs3 –event dcs3)
PVSS: OPCClient (num 2)
(-data dcs3 –event dcs3)
Ctrl 10 (ELMB Watchdog)
Subprojects on S:\...
dist -> [40]
_FW [21]
WIENER OPCServer
Data-, Event- Manager
PVSS: OPCClient (num 14)
(-data dcs3 –event dcs3)
User Interface
Sysname:
ATLPIX_3
_FW_DDC [33]
Data, Event, UI
Control Manager
DIM
-num 11 (FSM), 20 (Cooling)
Subprojects on S:\...
UserInterface
(-data dcs3 –event dcs3)
Ctrl 12 (DDC Handling)
dist -> [3], [40]
(SR1EnvironmentCooling_Disc.pnl)
Subprojects on S:\...
Sysname:
dist -> [3], [21]
ATLPIX_FSM_dcs1
Sysname:
ATLPIX_DDC_dcs3
network drive S: ->
network drive S: ->
network drive S: ->
D:\PVSS_Project_Update
\\dcs1.pixel.cern.ch\
\\dcs1.pixel.cern.ch\
PVSS_Project_Update
PVSS_Project_Update
(Subprojects)
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
overview of systems in SR1, II
dcs4
control3
ServiceTest
_CAN-ELMB_FW [4]
_RemoteUI
ServiceTest [99]
KVASER (ELMBs)
Remote User Interface
Data-, Event- Manager
CANopen OPCServer
ATLASPixDCS (dcs3):
SR1Menu.pnl
ServiceTest-MainPanel (GE)
PVSS: OPCClient (num 7)
Data-, Event- Manager
_FSM [26]
User Interface (DEN)
Data-, Event- Manager
Control Manager
User Interface (DEN)
dist -> [3]
dist -> [3], [21]
Sysname:
ATLPIX_4
dist -> [3]
Subprojects local
Sysname:
Subprojects on S:\...
Sysname:
ATLPIX_FSM_CONTROL3
nezwork drive S: ->
\\dcs1.pixel.cern.ch\
PVSS_Project_Update
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
ATLPIX_ServiceTest
PP2
many problems with PP2 crates for endcap
• new hardware slower, was not taken into account in PVSS (FIT and FSM),
commands were sent while Controller was still busy
no reliable execution of commands
-> introduction of very conservative delays
• no ADC reading (no feedback about Vset)
-> missing -15V supply for controller, need PS monitored by PVSS
• problems in ELMB/FPGA code -> already solved by firmware upgrade
• move trimmer to inhibit/uninhibit channel,
• GU when moving trimmer of Board1, Ch1,
• sending inhibit for one channel caused another channel to be inhibited
• Inhibiting/Uninhibiting of Channels 1 when reading T1/T2 after powering up PP2
• „Reset“ (i.e. set all Trimmers to Min) does not work properly in SR1
-> solved with new firmware version
probably was not recognized more clearly, because most of the time PP2 in SR1 was operated
in the way that trimmersteps were not calculated directly, but trimmer was set via moving to
minimum
• no Imon, because new crates were already defined, and monitoring with PDO3 was not
debugged previously with real HW
-> changing to PDO3 now done for PP2_S21, PP2_S22
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
PP2
New Communication Mechanism:
Acknowledgment between each node: FPGA <-> ELMB <-> PVSS
• Commands are acknowledged
(no chance of loosing command without noticing between FPGA and PVSS)
• only as much delay as necessary
• 1000 ms -> 10 ms (Inhibit)
• 3000 ms -> 200 ms (100 Trimmer Steps)
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
PP2
Also:
• Watchdog between ELMB and FPGA
• FPGA can reset ELMB
• FPGA failure is recognized by ELMB
• Reduction of number of commands sent by PVSS to ELMB
• Readout will be stopped/started for commands by ELMB
• PDO will be sent to signal routine mode
(still makes sense to stop routine before long sequence of commands)
• New Commands
• Firmware version of FPGA
• Firmware version of ELMB
• multiple commands reasonable? e.g. „inhibit_Board“? „inhibit_Modules“ ...?
can be implemented on a short timescale in ELMB code
• Number of ADC bits for Monitoring values will be increased (8 bit now)
• faster reading of monitoring values possible
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
iseg
some iseg channels were not switching off, when using „SR1DirectLink“
(while switching off one disk at a time and switching on worked)
• FSM: order of switching on introduced delays (-> PP2) before switching on isegchannels -> not so many commands were sent simultaneously
• introducing delay between commands for modules also helped (but not 100%)
problem was the way PVSS accessed the opc items:
• 1 item for each iseg module with one bit for each channel
• this item is „split“ by OPC client and „distributed“ to the DPEs
(trans.type bitstring, DPE bool)
• old values for first bits caused channels to be turned on again
using now opc item for iseg channel instead of module, and „write only“
• trans.type bool for DPE bool
• no more problems seen till now (switching off 4 times, first week of january)
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
FSM
many problems seen in system test were problems of underlying layer
(PP2, iseg)
missing functionality:
• order for switching on/off should be flexible
• also for example switching on without HV should be possible
• command for reading in of (default) values (->SIT)
• take into account HV for state for modularity 1 (done for 6/7)
• overall status of FSM („idle“, „active“) instead of only on ROgroup level
• SW-interlock implemented but not yet tested
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
SIT
• provides mapping for FSM
• improved xml parsing (mapping and configuration)
• reading in of configuration files
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
DDC
• Setting of Voltages
•
•
•
•
•
•
• supports all FSM commands
•
•
•
•
•
•
SET_VDDA
SET_VDDD
SET_HV
SET_VISet
SET_VPin
SET_VVDC
SWITCH_ON
SWITCH_OFF
RECOVER
RESET
ENABLE
DISABLE
• Publishing of values:
• Set Voltages, measured Voltage, measured Current
• T_Module, T_Opto
• improved monitoring of necessary processes in work (dim, dist, ctl)
• boc monitoring in work (DT)
• message transfer (MT) todo
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
Archiving, Trending
PVSS access to Oracle database?
trending very painful for large amount of DPEs or longer timespan
(trying to access 24 T_Opto-DPEs of ~2 weeks before almost impossible with PVSS
trending, but also problems with recent trends)
-> PVSS dataviewer
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
PVSS Data Viewer
(https://twiki.cern.ch/twiki/bin/view/Atlas/PvssDataViewer)
What's new :
Improved data export
 Possibility to filter values depending on their PVSS status word
 Improved data extracting (faster and can now display deleted elements)

To do :
 Un-/Aliasing of DPEs
 Additional information on requests
 Different display options
Olivier Pisano
Temperature reading (BBIM/BBM)
BBMs were defined as standard ELMB
• Temperatures had to be converted from Voltages manually
• new ELMB-type „BBM_NTC“
Problem with Temp-conversion in OPC-server (BBIM, BBM)
• if disconnected (Vref = VNTC) value cannot be evaluated by OPCserver
no check possible
• if value ok again, does not get updated immediately inside PVSS
wait sometimes minutes for up-to-date value
(looks like combination of „division by zero“ in OPC-Server and PVSS-smoothing)
GQ gets actual value, but not practicable
• no smoothing, Vref = const 2.5V seems to improve this: update with next SYNC
• archiving?
• amount of data without smoothing
• change of value propagated to archive if smoothing applied there?
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch
conclusion
• iseg problem seems solved
• PP2: many updates in the next few days scheduled
• FSM: much additional functionality desired
Pixel Systemtest Workshop, 08-02-2007, Kerstin Lantzsch

Similar documents