The CAL channels concerned by this belong to columns 5,6 and 7

XtalEnergy vs GemDeltaEventTime histograms for the 3 columns:

column 5

column 6
example: tower 8

column 7

An ASCII list in the format (tower,layer,column) of channels having the noise:

1 Comment

  1. Alexandre Chekhtman

    I've done the complete study of shaped readout noise in 2006 which was presented at Instrument analysis workshop Feb 27 2006. In this study I used periodic trigger events, which allowed to see complete picture of shaped readout noise for each channel, which is in study hidden by zero suppression. On the slide 9 of my presentation there two plots showing the amplitude of shaped readout noise for all LEX8 channels versus 10*col+twr for positive and negative crystal faces and on slide 10 - 1d histograms of noise amplitudes (see these plots below). Main conclusions:

    • Noise amplitude is asymmetric: if it is big at one end it is small at another end
    • There are channels with big NEGATIVE noise (layer 7, col 4 and 7), whch you don't see in you r study (because of zero suppression)
    • I have the full list of channels with readout noise parameters which I have attached to this page - it contains twr, lyr, col, face, pedstal bias and noise amplitude at 26.5 us after previous avant in LEX adc units
    • The list of channels which we want to exclude from position measurement depends on the threshold we set.
      • mean noise is 24 LEX8 units, we probably should exclude channels with readout noise > 50 adc units and <-50.
      • It is not that difficult to correct for this effect - may be it is a better solution ? The exclusion of channels will also require the modification of CalXtalResponse - may be we just subtract the noise, calculated by formula ronois_corr = ampl*exp(-(dt-tdead)/tshp) where ampl is taken from the attached table, dt - GemDeltaEvntTime in microseconds, tshp=4.2 us (that's what I got from my study).
      • For position measurement of carbon trajectory 1mm bias corresponds to 0.002 * 500 MeV  = 1MeV of noise amplitude.(30 LEX8 units).
      • I never implemented this correction mainly because considered it not very important. But now, when we want to get poition measurement aprecise t down to mm scale, probably this correction could be justified.
      • There is similar effect in HEX8 range, maximum amplitudes are also ~10 MeV and similar pattern, but individual amplitudes are different from LEX8. At the beginnig of HEX8 range (1 GeV per crystal) it corresponds in worst case to 1% effect and gives position bias 5 mm - may be there is some sense to take this into account for channels with biggest noise as well.