Title: Pixel Commissioning
1Pixel Commissioning
- Claudia Gemme
- CERN/INFN-Genova
- IDSG 21 May 08
2Before the sign-off
- The full detector was tested
- Fiber sign-off individual channel power
measurements to check light transmission in both
directions. - Connectivity Test
- optolink tuning at room temperature to allow data
transmission, - pixel threshold scans (1W/module, 270 pixels
injected in FE0). - We had a full map of problems at module level
(not all FE chips tested nor bump connection).
- Problems in BOC TXs (off-detector)
- less light than expected
- individual dead channels (and keep dying! 7 ch in
5 TX since 13Apr 22ch in 17 TX overall ) - Few ch.s send Ck but no commands
- Problems on detector
- Eight modules with no HV
- L0_B03_S2_A7 M3
- L1_B14_S2_C6 M5
- D3A_B02_S1 M6
- D3A_B01_S2 M5
- D3A_B03_S2 M6
- D3A_B02_S2 M5
- D2A_B01_S1 M4
- L1_B18_S1_C7 M3 (found in SR1)
- One module that can not be correctly configured
in L2 (to be debugged). - Three modules under investigation (no HV? But
R_module at PP4 is fine) - Since the integration
- One module VDD/Ck short
- One barrel module as disk 7th module
End-cap A is clearly more affected and in
particular the face on which HV bondings were
not reinforced during the End-cap integration
(M4,M5,M6 lay on the same disk face)
3Pixel Sign-off Procedure
- Cooling commissioning per quadrant detector OFF
- Cooling Operation/commissioning of one loop at
the time - -gt Steve for problems
- Check mapping between loop, detector and pixel
pipes sensor -gtOK - Cooling/Detector commissioning per quadrant
- Optoboard cooling circuits first operating
cooling, optoheaters and optoboards at the same
time. - Detector operation, one loop at the time
- Use the detector as a flexible heat load
- Leave cooling and modules ON
- Run threshold scans on a full quadrant after
individual loops commissioning.
4Pixel Sign-off Summary
- () detector can not operate stably in standard
configuration - Summary worksheet in detector elog
https//atlpix01.cern.ch/elog/Detector/439 - Q1, Q2, Q4 completely tested few problematic
circuits in Q4 (octant7). Stop testing in Q3
while we were starting detector/cooling
commissioning of loops in Octant 6.
51) Optoboards operation
- One loop ? one quadrant (8 loops total)
- Per loop, Simultaneous running of cooling
(power50), up to 36 optoboards and six
optoheaters (40-50) (see next slide). - Operation was stable except for cooling loop 50
(opto A56). - Optoheaters operation stable (except we have
ELMB communication problems probably fixed with
CAN correct termination). - Optoboards temperature is rather uniform (2-3C
difference for six optoboards) but we have
observed at least one optoboards being 7C higher
then the average. - Feedback for optoheater operation is done by
optoheater NTC that is located on one edge of the
optoheater strip optimal setting of Tset still
to be calculated, especially when optoboard array
temperature is not uniform (also due to spare
boards- see next slide). - Data analysis is on-going but it will be
- uncomplete as archiving of
- optoheater project dP was disabled.
A12 heater
6A56 opto loop
HeaterNTC
HeaterNTC
HeaterNTC
HeaterNTC
HeaterNTC
HeaterNTC
72) Modules loops, power measurement
- Procedure
- Cooling on, modules off (0W)
- Switch on modules, uncfg (1W/module)
- Config modules with standard cfg (4W/module)
- Increase low voltages ( 5W/module)
- change FE DACs setting (5.6 W/module)
- Back to standard settings
- Used only data
- from travellers
- (shifter dependent).
- Q1,Q2 only.
- More in Markustalk
- (http//indico.cern.ch/
- conferenceDisplay.py?
- confId33569)
0W
4W
4W
6W
8Heater Power vs. Detector Power
- Mostly linear behaviour
- Full loop power
- Two classes clearly visible discs (12 modules)
and barrel (26)
9Temperature vs. Power
All but L2
L2
- Very similar behaviour for all local supports
(L0/L1/discs). - Full loop power
- Temperature averaged over full loop (bistave /
bisector) - Much larger spread for L2.
10dT / dP vs support
L0/L1
L2
Disks
- Linear fit result for L0/L1, L2 and disks
- Power calculated per module
- L2 contains a class of staves with similar
performance as L0/L1 and a class of worse staves
(as expected). - L2 bistaves are built with an inserted stave on
the inlet (due to corrosion problems).
11Difference between Pipes, L2
Inlet stave
Outlet stave
- Difference between thermal performances of inlet
stave (pipe inserted) and outlet.
123) Modules loops, Threshold scans
- When all the loops (at least the stable ones)
were commissioned, we had some time to run a
threshold on a full quadrant - Including in the threshold scans all cold parts
- Enabling max 2 PP0s (out of 4) in Layer2 RODs
(Current ROD DSP software limitation) - Use optolink parameters from Connectivity test
- Start iterative tests
- Run a threshold scan
- BOC tuning on modules (PP0s) failing the scan
(30) - Re-run a threshold scan, etc
- Difficult to judge with so few scans how
stable/critical are the optolink settings.
133) Modules loops, Threshold scans
14Calibration Console
Calibration console
15- Modules distributions (Q4 as example)
- Noise 170e
- Double peak in Threshold mean and RMS
distributions - Correlation between higher threshold and lower
RMS and viceversa - It should come from production sites.
- Modules on the same mechanical support usually
belongs to the same peak - One NEW HV open found (it was fine during the
warm tests) - D3A_B01_S1 M1
16DAQ/calibration
- Data taking
- Transition to tdaq 1.09 and combined run with
ROD-simulated hits next week. Main purpose is to
prove compatibility with ATALS framework and
systematically run online monitoring (GNAM) - Calibration
- ROD DSP code (High priority) first official
release is now under test in SR1 and available
for DAQ developers and users. First systematic
test on digital scans, optolink tuning and
threshold scans. To be compared performances and
timing. - Possibility to run scans and the subsequent
online analysis is there even if we still miss
the full chain for few important scans. Results
are stored in the CalibrationDB where we need to
import also the data coming from the production. - Urgent need to have tools to perform offline
analysis (Analysis Framework) to monitor the
evolution of calibration results, make
differences with reference scans and being able
to put together results from different part of
the detector. We are late on this topic!
17DCS
- Main on-going software activities
- DCS configurationDB is now available (used for
high voltage settings). Some items still missing
(optoheater settings, interaction with DAC for
Viset setting, configuration of FSM value range,
etc) - FSM minor branches in preparation (PP2,
interlock, DAQ crates) - Automatic Handling of many PP0s startup (it is
now impossible to startup the full detector, only
12 PP0s at a time!) - Cooling loop oriented panels (showing status of
detector per loop) - Overview temperature panel (for beam pipe
bake-out mainly) - Panel to display configuration of software
interlocks (important for commissioning period) - Hardware
- Still missing 2 LV Wieners (out of 55)
18May 1st
Cool ON Pix On
Cool ON Pix OFF
1st Plant off
Manual modules switch off
2nd Plant off
- Software cooling interlock will handle automatic
modules switch-off when loop state ON-gt
standby/OFF
19Beam pipe services test
- The beam pipe heaters have been switched on for
few minutes _at_50C (longer on the A side, very
short on the C side). The larger effect is on the
top of the detector - Aside L0B3 M6A increase of 4.5C reaching 24C
- Aside bottom L0 bistaves less than 1C
- Disk top modules increased their temp of less
than 2C - On C side three heaters ON almost at the same
timeL0B3 M6C increased of 5C reaching 26.5C
L0 M6 A side
L0 M6 C side
Heating from the A side
20Conclusions
- Calibration and analysis are still our main weak
point. We are working on this as well as a first
release of the DSP code we absolutely need to be
able to run calibration scans (all modules at a
time, Xtalk, monleak, etc) - We are going to profit of this time to train
people and write documentation to be able to
exploit a larger set of people during the
commissioning. End May/beginning of June first
DCS/DAQ courses. - TX problems (death and low emitted light) will be
subject of a joint SCT-pixel review in a couple
of weeks. - Modules problems we have identified as well as
test of BOC algorithm can keep going even w/o
cooling.
21Cool-down speed
- This is probably the measurement with the highest
influence of personal judgement (Where does the
ramp start / stop? Full ramp or only steepest
part?) - Useful only to give an order of magnitude (1K/s)