Barrel 3
Part of Barrel3Analysis:
The only cold run which was analysed by the online software is Run 2145 scan 2. Some of the "funny" plots from this run are under this directory [Barrel3_DT]
Lets define a parameter peak. Peak is the ratio of the highest occupancy in the bins divided by the plateau of the other bins. The cut can be made to be around 4 -- taking into account that the error bars are very large on data points.
Here is the list of the modules with a high peaks
20220040200234 LMT06 Z+6 large peaks of a x10 on the upper side of the module. Electrical pickup from neighbor or from Doric? --> This feature is not apparent in Cern data. Maybe the pickup was from a neighbor which is now in a different trigger group? 20220040200427 LMT15 Z+2 chips 0,3,4,6,8 has small but still visible peaks... 20220170200008 LMT30 Z-6 chip 4 has a small peak. chip E13 has known noise slope which goes away in Select=1. 20220170200017 LMT30 Z-2 chip 6,7,11 has small but visible peaks. 20220330200244 LMT18 Z-3 looks like a small light leak... --> also visible in Cern data 20220330200290 LMT17 Z+4 chips 9 and 11 has small but visible peaks. 20220330200415 LMT09 Z-4 a very large and clear light leak! --> also visible in Cern data 20220330200472 LMT08 Z+1 chips 5 and 6 has small but visible peaks. 20220330200475 LMT08 Z+2 chip 5 has high occupancy and a peak...
For the distribution of peaks over all of Barrel 3 chips, take a look at [histdouble.eps] or for a log plot, [histdoublelog.eps].
We now think that some of these are due to light leaks from the optopackages. Unfortunately, we can not check this with Barrel 3. However, we looked for optopackages which might leak light out on Barrel 5 and 6 which are still at Oxford. We have found about 10 cases which we think might indeed leak light: [optoleak1], [optoleak2], [optoleak3] There are now fixed by either gluing the original aluminum foil back or gluing a new aluminum foil on the original one with Araldite 2011.
- Note about DoubleTriggerNoise Runs:
2145 scan 2 From GUI. "correct trigger" shows light leaks run incomplete. 369 out of 384 modules. 2176 scan 0 From script. unknown no light leaks 2177 scan 0 From script. unknown no light leaks 2185 scan 13 From script. "wrong trigger" no light leaks 2185 scan 15 From script. "correct trigger" shows light leaks
NEWS: sctGUI displays all the DoubleTriggerNoise variables. For barrel 3 tests at Oxford, the ratio of max to pedestal and the time bin were not in the summary file, and therefore not available.
The former observables in the Summary of DoubleTriggerNoise were:
- Maximum Occupancy Peak
- Pedestal Occupancy
Two observables have been added to the Summary:
- Ratio of Maximum to the Pedestal Occupancy
- The time bin where the Maximum Occupancy Peak occurs
The DoubleTriggerNoise test was repeated for Barrel 3 reception testing at CERN. From Run 1375 Scan 0, there are three clear DTN defects:
- LMT26 z+1 UK74 Chip 12 : found from Cern data. The features look exactly the same as the one case found on Barrel 5. Maybe this DORIC-related noise will go away when the modules is in SELECT=1? Plot here at [lmt26z+1.ps]
- LMT09 Z-4 UK415 Chip 11 : very clear light leak
- LMT18 Z-3 UK244 Chip 11 : small but visible light leak
Also, interesting is that the effect seen on LMT26 z+1 seems to also show up on LMT25 z+1 UK98 slightly.
Barrel 6
Part of Barrel6Analysis:
The analysis was tweaked a bit to report a DoubleTrigger? defect if the peak is 5 sigma away from the baseline or if the peak is higher than 1*10-4. Interesting cases to look are listed here and the plots are [Barrel6_DT]
5874.1.20220040200309 LMT18 z+2 : Looks like a noise slope on chip S13. But still there is some time structure, and bad OPEs. Maybe electrical pickup? Noise slope shows up in Noise Occupancy test from run 5852. Maybe we should try running this one in select=1?
5874.1.20220330200358 LMT08 z+5 : This is the module that lost bias on one side of the module. Only interesting as a textbook example of what happens when a module loses the bias. Having no bias, it is prone to more electrical pickup. Well, one should also note that historically, this test is when we realised that this module had lost bias.
5882.0.20220040200481 LMT03 z+3 : I have no clue what's going on here. This feature does not show up in Noise Occupancy on chip M1.
5882.0.20220040200137 also an LGS candidate LMT40 z-5 : Seems like chip S11 has LGS and seems to pickup some noise as well. Maybe curable?
No obvious light leaks observed. :)
Here is a list of the optopackages which were fixed:
B5 LMT5 P10 (z+4) LMT8 P3 (z-4) LMT19 P1 (z-6) LMT36 P8 (z+2) LMT40 P4 (z-3) & P11 (z+5) LMT44 P9 (z+3) B6 LMT9 P6 (z-1) - extra layer of foil glued on top LMT29 P5 (z-2) LMT39 P8 (z+2) - extra layer of foil glued on top LMT40 P12 (z+6)
Barrel 5
The analysis has found one very large peak on Barrel 5. This is the first time we had the time to investigate such a big feature carefully so I will here try to document what we did.
LMT10 Z-4, Jap 522, (from run 8137) shows something in the DTN noise, plots can be found here at [Barrel5_DT]. Chip C13 has an occupancy of 3e-3 for some time bins. But the time structure is not the "11101" expected from pickup of the event data being broadcast to the ROD, rather "1100110011".
1) we put the module in clock/2 mode and did not see a difference in leak current which would indicate a light leak
2) we tried reducing the VCSEL to 3V to reduce the light output did not have any effect either.
3) we switched off the power to all other modules. This did not have an effect either and rules out pickup from other modules.
4) when we put it in SELECT=1, the effect went away. Hence is not pickup from the module's own outputs
5) going back to SELECT=0, we tried reducing the TX current to reduce the amplitude of the optical BPM signal, but this does not attenuate the pickup. Increase the amplitude does not have an effect either.
6) We tried the MSR test which may be broken in release4-updated...
Our best guess then, is that this pickup is from an electrical signal coming from DORIC, perhaps ringing from an unbalanced LVDS command signal. Since it goes away when SELECT=1, maybe we don't worry?
Given the feature on Barrel 3 looks the same as this one and seems to effect its neighbor, could this also effect its neighbor on LMT 09? We can not test this since they are on different cooling loops...
Barrel 4
Part of Barrel4Analysis
- LMT12 Z+2 DTN noise slope in last chip
- LMT17 Z-6 DTN feature due to low level light leak
- LMT23 Z-2 DTN vague stripes
- LMT25 Z-6 DTN structured noise slope in last 2 chips
- LMT34 Z-3 DTN stripes
- LMT35 Z-3 DTN small blip in E13 ~ bin 130
- LMT38 Z-6 DTN stripe in chip S10 ~ bin 126
- LMT40 Z-6 3PG/DTN/NO offset/noise slope in last 2 chips, not a property of the module as it has aready been replaced
- LMT40 Z-6 DTN stripe in chip S10 ~ bin 126
- LMT40 Z-5 DTN noise slope in last chip