Bias Scan runs taken by Rosi.

No	Run No	Voltages: Vset / Vbias (*)	Prashanth's initial results	Mike's Results
1	18089047	56.5 / 58.3	Bias Scan	drupal.star.bnl.gov/STAR/system/files/BiasScan.pdf See below for analysis
2	18089055	55.5 / 57.3
3	18089063	57.5 / 59.3
4	18089065	54.5 / 56.3
5	18090003	58.5 / 60.3

Root files for above runs are here: rcasxx:/gpfs01/star/subsysg/EPD/sprastar/EPDHists

(*) Bias voltage (Vbias) is 1.8 V above the Vset value one sets with the TUFF box.  The relevant clip of Gerard's mail:
"The actual bias on the SiPM is higher than the setpoint by about 0.9 V for the
VCOMP (temperature compensation) plus about 0.9 V for the preamp input voltage
(the regulator regulates the anode voltage to -(VSET+VCOMP)*(1 +/-1% error) )"

Note that Hamamatsu specs Vbias at about 57.7 V, for this initial batch of SiPMs:
drupal.star.bnl.gov/STAR/system/files/S13360-1325PE%20specs%20initial%20150.pdf

Mike's analysis of the bias scan - 9 April 2017

• We don't want to wander too far away from Hamamatsu's recommendation of Vbias=57.7 V (Vset=55.9 V), since the signal increases linearly with bias voltage (see plots), and the dark current increases exponentially with bias voltage.
• Looking carefully at the 6 FEE groupings (one FEE handles odd or even tiles from a supersector):
  • 4 groupings (PP4 even, PP5 odd, PP6 odd, PP6 even) give MPV ~ 45 ADC counts for Vset ~ 56.5 V
  • PP4 odd gives MPV ~ 60 ADC counts for Vset ~56.5 V: this is because it is read through QT32c which has higher gain
  • PP5 even gives MPV ~ 24 ADC counts for Vset ~ 56.5 V: This may be due to one of the following reasons:
    1. This fee has a lower gain than the others
    2. This QT board has lower gain than the others.
       • This may be checked by swapping inputs with another QT board
    3. The fiber-to-SiPM connection (FSC) is worse than the others.
       • In principle, this could lead to a highter WID/MPV ratio due to reduced photon number and higher Poisson fluctuations, which is not observed.
       • Nevertheless, I kind of suspect #3
• The "gain" is rather linear with Vset, so we can "peak-match" the various channels by adjusting Vset such that the MIP peak is always at the same location.  However, this should only be done within limits.  It makes no sense to lower Vbias a lot, just to "compensate" for the high gain of the QT32c or to raise it a lot to "compensate" for the low gain of PP5 even.
• Therefore, I suggest to proceed as follows:  Set Vset such that
  • MPV = 45 ADC counts for PP4 even, PP5 odd, PP6 odd, PP6 even
  • MPV = 60 ADC counts for PP4 odd
  • MPV = 25 ADC counts for PP5 even
• The Vset values, using this criteria, may be found at drupal.star.bnl.gov/STAR/system/files/VsetValues.txt
  • I set Vset by hand to 56.5 for 3 of the 93 fits that failed.