OPEAnalysis analyses data from the OccupancyPerEvent Data using OPETool.
The cold run data from [Run 2180] Scan 0 now has been analysed. This run is the most complete run for NoiseOccupancyTest since it has complete data from range -40 to +20mV in threshold (relative to TrimTarget). In this data set, 9 modules have been tagged as having "high badness" chips.
New: Run 2180 Scan 0 was taken before this analysis was available. Now the old-analyzed files have been replaced with newly analyzed files and are available on [here on web]. This show the maximum "badness" for a module over all chips.
Here is a list of these modules. In many cases, cross-checking with the Oxford Reception test data and cold test data from module production sites revealed that s-curve dicontinuities had been seen previously. There are two different modes of "badness":
- Badness at high threshold
20220040200324 LMT21 z-6 bad at 20mV over 1fC at chip S11 hint of s-curve wiggles at cold test 20220330200353 LMT11 z+6 bad at 15mV over 1fC at chip S12 not previously known to have had s-curve problems
- Badness at low threshold at bad at -40mV below the TrimTarget. All of the following have more than one chip that has high badness -- always at the link 1 side.
20220380200107 LMT29 z-5 s-curve wiggles at Oxford Reception test 20220380200136 LMT28 z+4 s-curve wiggles at Oxford Reception test 20220380200114 LMT29 z+1 s-curve wiggles at cold test 20220380200081 LMT28 z-5 s-curve wiggles at cold test 20220330200078 LMT26 z+3 s-curve wiggles at cold test 20220170200210 LMT30 z+5 s-curve wiggles at Oxford Reception test 20220040200070 LMT07 z+1 not previously known to have had s-curve problems
The plots for these runs are available at [OPEData]. There are three plots for each module:
- Raw.2180.0.modulenumber_1.eps: contains Occupancy information for each link. The x-axis is the channel number and the y-axis is Threshold minus the TrimTarget in mV. For those two modules with badness at high threshold, this plot is very interesting because you can see the cluster of hits at high threshold. Note that the number of triggers is less at lower thresholds. This was done to speed up the readout. When there are more hits in an event, it takes longer to read it out.
- Raw.2180.0.modulenumber_OPE_1.eps : contains histograms for each chip. The x-axis represents the number of channels firing in any single event. The y-axis is the Threshold minus the TrimTarget in mV. The color scheme is logarithmic. For each threshold bin, the distribution of the number of hits for event is a Binomial distribution. (For a 3D example from the sector run for thresholds between +20 and -80mV, see [plot].) For the Binomial distribution, the mean determines the variance directly.
- 2180.0.modulenumber_chipNO_2.eps : contains plots for each chip. The x-axis is the Threshold minus the TrimTarget in mV. THe y-axis is the calculated "badness". Badness is defined as the ratio of the variance from data to the variance as expected from the mean of the distribution. Ideally, badness should be equal to 1. However, the gain from channel to channel varies a bit and makes the distribution of number of hits wider slightly. Anything above badness of 2 is flagged to be "bad". For the distribution of badness over all of Barrel 3 chips, take a look at [histbad.eps] or for a log plot, [histbadlog.eps].
Note: NoiseOccupancyTest starts taking data at threshold of 20mV above TrimTarget and then go lower. The following runs aborted often due to data errors at lower thresholds. They are not complete data sets but, they were analyzed for checking the results above.
- Run 2179 Scan 0 -- Slot 14 failed (Z+ LMT1-8) and caused abort at the end of the scan with threshold -35. So there is no data for threshold -40. However, there is still one module with high badness: 20220380200107 at threshold -35. This is one of the modules from the list above.
- Run 2088 Scan 8 -- Z- only data. Again aborted at the end of the scan with threshold -35. There is no data for threshold -40. No badness detected at higher threshold.
- Run 2185 Scan 22 -- aborted at threshold -20. No badness detected at higher threshold.
Analysis of warm runs will follow shortly.
Changes for Barrel 6
Barrel 6 contains lots (and lots) of modules with s-curve wiggles. So with the above criteria, nearly half of the modules on B6 fail the OPEanalysis. So the cuts have been loosened. We want to pick out modules on B6 which have a bad OPE at high thresholds, close to 1fC which will be used for running. Now, thresholds below -20 are not taken into account while calculating badness. Also, there is now a new defect "BAD_OPE" which shows up as PROBLEM=TRUE in the TestResult of NoiseOccupancy?.Here is a list of OPE related Badnesses on Barrel6 RawScanResult.5874.0.20220380200057.gz : You can see a Badness in Chip 1. This is due to the fact that the PINI got higher when the BOC cooling was turned on. There was data errors on this run as well. This was not reproducible.
RawScanResult.5852.4.20220170200878.gz : shows a High OPE. The test was repeated and the bad OPE was not reproducible.
Barrel 4 summary
9459.7.20220170200659 chip S2 : Noise Occupancy run, LMT16 z-4: One event with 5-9 hits.