when monitoring the run 236084237, we noticed the following:
Number of hits in each Cal log end for LEX8:
(FastMon > CAL > Hits per log end(LEX8), FastMon_CalLogEndRangeHitCounter_Range0_TH2_Tower)
significant spikes appear in crystals:
(Tower, Layer, Crystal) = (0,0,6), (2,1,5), (2,7,7), (3,7,5), (7,1,5), (9,0,5), (9,4,5), (10,0,6)
and it's consistent with the list of "hot"/noisy crystals pointed out by Sasha:
Hi, Eric,
To confirm that all "hot" channels are produced by shaped readout noise I
reprocessed conSciOps run, but with selection GemDeltaEventTime <70us. As
you can see on the attached plots, now I do see the selection of 18 crystals
having high rate in lex8 range, unusual pos/neg ratio and/or mean lex8
signal.
The list of "hot" crystals follows.
By the way, it is not clear what to do with this effect. We always said that
if necessary the shaped readout noise could be corrected (we have the
measurment of the noise vs dt), may be it's time to do it. Otherwise we'll
have to artificially increase a dead time upto 40 or 50 us.Best regards,
Sasha.crystals with high readout noise:
twr lyr col
0 0 6
1 1 5
1 4 5
2 1 5
2 7 7
3 7 5
4 1 6
4 4 5
5 0 6
5 1 5
7 0 5
8 1 6
9 0 5
10 0 6
10 1 6
10 4 6
14 3 6
14 4 6
and here are Sasha's plots:
Average number of adc counts per log end, range LEX8:
Rate of adc counts per log end, range LEX8:
Ratio of Positive face / Negative face number of adc counts per crystal, range LEX8:
follow-up of these noisy crystals trough several runs (not done yet...)
As mentioned above, an impact of this noise can be seen on the deadtime distribution, see [Berrie's study| https://confluence.slac.stanford.edu/display/SCIGRPS/2008/07/01/CAL+noise+for+run+r0236535656].