1. 1 some high rate measurements at IC Mark Raymond m.raymond@imperial.ac.uk attempt to re-produce CERN high-rate behaviour in the lab at IC a few new (I think) observations CMS readout chain (transmission/FED system) can be ruled out module (hybrid) without sensor does not show problem

2. 2 setup UTRI card LV regs, interface buffers CK/T1 I2C CK/T1 and I2C provided by custom VME modules (no CCU) SeqSi digital pattern generator trigger veto output data stream ADC, 128k deep memory digital header detect count up count down trigger veto using simple counter increments on scintillator trigger decrements as APV frames output veto if count reaches 9 (safety margin) source PM/scint. front end experimental technique free run system with random triggers trigger and readout all 128k ADC samples (accommodates ~ 300 data frames at ~ 100 kHz) repeat to build up larger data sets analyse data source generates random triggers

3. 3 reproduction of effect on TEC module noise dependence on channel number (couple of bad channels) noise increases with trigger rate readout interface completely different to CMS system => problem confined to front end module chip 1 chip 2

4. 4 effect on TIB hybrid only no sensor noise unaffected by trigger rate => sensor necessary to reproduce effect? maybe but TIB hybrid layout slightly different to TEC => look at TIB module chip 1 chip 2

5. 5 reproduction of effect on TIB module effect also reproducible in TIB module => confirms sensor required (some bad channels on this module too) chip 1 chip 2

6. 6 problem data frames some “bad” examples note anti-correlation between chip 1/ch127 and chip 2/ch 0 chip 1 chip 2

7. 7 coupling speculation GND V250 GND seems to be only terminal in the right place? disturbance on GND would couple more strongly to chip 1, chan 127, 127, 125, ….. than to chip 2, chan 0, 1, … weak coupling here stronger coupling here? but … anti-phase ch127chip1/ch0chip2 doesn’t make sense to me maybe this is misleading … chip 2 chip 1 127 0 PA

8. 8 coupling speculation … …. maybe this is misleading chip 2/ch0 noise might still be there even if chip 1 not there could confirm by removing power to chip 1 would have to pull bonds to stop all chip 1 digital functionality bit drastic but maybe worth doing? put it another way would chip 2 effect still look like this without chip 1 chip 1chip 2

9. 9 observations a worthwhile exercise I think effect reproduced without conventional CMS readout chain => front end problem – data transmission/FED system not implicated effect not seen on hybrid alone => sensor presence is required shape of effect (data frames) is strange – per APV rather than per hybrid => APV is somehow implicated?

10. 10 extra

11. 11 APV picture

12. 12 APV pads

13. 13 APV front end

14. 14 APV back end