35t FEMB Status and Noise 20160224 - 35t - B. Kirby
TPC Readout Noise - Main Issues ● Two main problems with TPC readout noise (in order of importance): ○ Noise levels are not stable: all FEMB channels have some chance to jump into a “high-noise” state that can only be removed by power-cycling the ASICs ○ Noise levels are high: when the readout is not in a bad state the noise levels are still high 2
Pedestal ENC > 20000 e- “High Noise” State Affects entire detector, never occurs on single FEMB RMS >100 ADC Oscillatory signal 3
Many tests, main goals were: -check if high noise caused by an external source -check if high noise caused by single board Recent Noise Tests -check if high noise caused by FEMB power on/reconfiguration Date ELOGs Test Feb 17 5917 Power off everything non-TPC related Feb 18 6004-6030 Change FE ASIC baseline to 200mV, vary shaping time settings Feb 19 6055-6066 Power off everything non-TPC related for extended time Feb 19-20 6067-6112 Overnight stability test with everything off Feb 21 6148 Run with only one FEMB with ASICs on, all other FEMB ASICs off Feb 21 6149-6153 Turn on FEMB ASICs sequentially Feb 22 6207-6273 Run with one FEMB with ASICs on a time, vary shaping time settings Feb 22-23 6275-6308 Alternate wire-bias and drift field on Feb 23 Turn on FEMB ASICs sequentially, correlate power draw with high noise 4
Feb 17 “Everything Off” Test Results Run 11481 Run 11504 Run State Collection Induction Collection Noise Induction Noise Noise (ENC) Noise (ENC) 11kHz Subtracted 11kHz Subtracted 11481 Everything Off 2310 e- 4060 e- 1310 e- 2660 e- 11504 Everything On 2940 e- 4780 e- 1990 e- 3570 e- ● APA3 channels started producing realistic noise following power-cycle ● Clearly see more noise when everything on vs everything off (~1500e- ENC) ● Did not see “high-noise” state when everything off, however test was too short 5
Feb 18 FE-ASIC Configuration Scan ● Changed FEMB channel baseline settings on all channels to 200mV to evaluate effect on noise ● Varied shaping time settings to see if any setting was more or less susceptible to high noise state ○ At this point non-TPC related systems were turned back on ○ All FEMB ASICs were kept ON throughout the test ● High-noise state was observed on all configurations tried ○ High-noise occurred randomly, removed via power-cycling or wire-bias trick ● Conclusion: FE ASIC settings do not affect whether “high noise” occurs 6
Feb 19 “Turn Everything Off” Test - Powering Down Run # ELOG# System State 6055 Lakeshore, thermo cables, RTDs unplugged, SSPs off, LCM off Drift HV off, wire bias off, cameras + pumps + purity monitors on 11765 6066 All ASICs power-cycled, everything off, “high noise” state ● Turned off all non-TPC readout related systems to see if “high noise” state occurred in the absence of potential external noise sources ○ “High noise” detected run 11765 ● Conclusion: “high noise” can occur when non-TPC related systems are off 7
Feb 19-20 Overnight Stability Test ● Wanted to measure how long FEMB could run without changing settings until high-noise issue occurred ○ This was done with the HV and wire-bias OFF ● FEMBs took data without high noise for ~24 hours Run State Collection Induction Collection Noise Induction Noise Noise (ENC) Noise (ENC) 11kHz Subtracted 11kHz Subtracted 11768 Drift field + wire 2310 e- 2700 e- 1530 e- 1980 e- 8 bias off
Feb 21, 22 Single FEMB Runs ● Can control LV power supply to ASICs on each FEMB independently ● Turned off ASIC power on all boards, all other LV power channels left on ○ Emphasis: FPGAs were left on, have not been turned off since detector cooldown ● Sequentially turned on ASICs on one FEMB, took data ○ This was done with drift field + wire-bias off ● High-noise state didn’t occur ○ >15 FEMB ASIC power cycles, >45 ASIC reconfigurations ● Suggests high noise doesn’t occur when single FEMB running 9
Alternate Wire-Bias and Drift Field ● Set of runs taken with wire-bias and drift field alternately on to determine if ionization charge can cause high-noise state in FEMBs ○ HV off, wire-bias nominal values: Runs 12453-12517 ○ HV drift on (60kV), wire-bias off: Runs 12538-12691 ● Still analyzing... 10
Feb 21 Sequentially Turn on FEMB ASICs N. Barros ● Lower than normal power-draw on ASIC LV supply correlated with “high noise” runs ● Current draw not logged in plot, but observations suggest drop in power associated with change in current draw 11
Feb 23 Sequentially Turn on FEMB ASICs N. Barros ● Continuation of sequential FEMB ASIC turn-on tests ○ Turned on RCE15 FEMB ASICs, observed lower power draw then normal ○ Turned on ASICs for FEMBs on RCE14-12 ○ RCE11 FEMB ASICs turned on, ASIC power draw dropped on FEMB 13-15 ○ “High noise” observed in next data run, removed by turning FEMB 14-15 ASICs off ○ Effect is reproducible ● Conclusion: high noise is correlated with abnormal ASIC current draw and 12 can be induced by changes in ASIC current draw on single FEMBs
Feb 23 Sequentially Turn on FEMB ASICs II ● Similar tests were done with FEMBs on different APAs ● Preliminary result: ○ Noise state depended on FEMBs on APA3 ○ If APA3 FEMBs were noisy (ie had abnormal power draw), then FEMBs on APA0 would also be noisy when ASICs were powered on ○ If APA3 FEMBs were normal, FEMB ASICs on APA0 could be powered on and the following run was not noisy ● Suggests one or more FEMBs on APA3 are specifically problematic and induce noise in all other FEMBs, but needs more tests ○ Reminder: before FEMB ASICs were power-cycled in the cold APA3 channels had essentially turned off, electronics response gain was very small ○ Explains why wire-bias “trick” would work when power-cycling would not, APA3 boards have a high chance of entering “high noise” state and inducing problem in rest of system 13
Summary of Noise Tests ● Understanding high-noise state as been focus of recent tests: ○ Demonstrated high noise can occur when most non-TPC related systems are off ○ Choice of FE ASIC settings does not affect incidence ○ System can be stable on ~24 hour timescale, when only single FEMB is running ○ High-noise state correlated with abnormal FEMB ASIC power draw (clear hardware issue) ○ High-noise can be reliably induced by changes in LV current draw on single board ● Noise levels are still generally high ○ Changing baselines to 200mV seems to have helped at cost of induction signal dynamic range ○ Non-TPC related systems seem to introduce ~1500e- ENC in TPC readout 14
Plans ● Focus on understanding how changes in current draw on single FEMB affects other boards, introduces high noise state: ○ Continuation of Feb 21, 23 sequential power on tests ○ Record exactly the ASIC LV current draw at each stage when FEMBs are turned on ○ Explore changes in power-on procedure ○ Demonstrate if ASIC reconfiguration has similar effect ● Identify stable operating configuration: ○ Can we identify a set of FEMBs/power on sequence/reconfiguration process that largely prevents high-noise state from occurring? ○ Determine to what extent (if any) running with drift field and wire-bias on affects incidence of high-noise state 15
Backup 16
35t Recent FEMB History and Noise I ● Initial post-cooldown noise high compared to room temperature, due to increase in 11kHz correlated noise ○ Run 10115-10123 - pedestal runs taken at all FE-ASIC settings ○ Wire bias + HV off through up to 10151 ○ Run 10119 noise: collection wires ~2030e-, induction wires ~2750e- ● Wire-bias turned on (ELOG 4995, Feb 5) ○ Very high noise observed starting run 10164 (ELOG 5018), noise persists after wire-bias turned off ○ Run 10195 noise: collection wires ~21000e-, induction wires ~22000e- ○ Reduced with wire-bias ramping “trick” (ELOG 5089, 5144, 5146, Feb 6-7) ● Drift field ramped up with wire-bias on and trick used (ELOG 5187) ○ Run 10400 taken with wire-bias + HV ON ○ Run 10400 noise: collection wires ~2860e-, induction wires ~3090e- ○ Pulser runs 10422-10431, wire bias ON, HV off (Feb 8), pulse signals look OK ● Noise levels fairly stable from Feb 8th to 11th ○ Run 10788 taken with wire bias on, drift field off before pumps turned on ○ Run 10788 noise: collection wires ~2770e-, induction wires ~3050e- 17
35t Recent FEMB History and Noise II ● LAr Pumps turned on (ELOG 5376, Feb 11) ● FEMB pedestals set back to “default” values (ELOG 5450, Feb 11), Run 10852 ○ Run 10830 noise: collection wires ~2810e-, induction wires ~3220e- ○ Run 10852 noise: collection wires ~3020e-, induction wires ~4450e- ● Drift field ramped up with pumps on (ELOG 5479, Feb 12), Run 10895 ○ Run 10895 average noise: collection wires ~3050e-, induction wires ~4560e- ● FEMB noise increased (ELOG 5574, Feb 12 ~6pm) ○ Started ROUGHLY run 10949, might be earlier ○ High noise has persisted, temporarily reduced by wire-bias ramping trick but still very high ● Changed FE ASIC settings to 4.7mV/fC, 3us (NO ELOG, Feb 13) ● High noise persists on all settings, occasionally increases to extremely high noise (>10000e-) ○ Run 10895 (Feb 12) average noise: collection wires ~6000e-, induction wires ~8000e- ○ APA3 channels have gone “quiet”, relatively low noise and very low gain 18
Recommend
More recommend
