[Hps-ecal] ecal hv

[Nathan Baltzell]

Dear Everyone,

This evening we realized another problem probably due to the ECal HV supply.

First, we see gains from cosmics that are 30/40% higher on all channels in
certain HV groups, while all channels outside those groups are within few-%.  
Those high-gain groups are isolated to the left half of the calorimeter, on both
top and bottom, and occur every other HV group.

One hypothesis is that it is due to motherboards, since left-top and left-bottom are
from one manufacturer, while right-top and right-bottom are from another.  But I
think we ruled that out as the primary cause out by using LEDs:

Swapping calorimeter channels from one HV mainframe channel to another,
while keeping everything else constant (including, of course, the voltage demand
and readback settings), resulted in large changes in measured LED signal amplitude
consistent with cosmic gains for these low/high-gain groups.

It may also be worth noting that these high-gain HV groups are on mainframe
channels 3,5,7,9,11 for the modules in both slots 0 and 8.  When plotting for ECal
geometry the correspondence with mainframe channel is not obvious, due to
mappings and relocation of one dead channel to another slot.

So it seems the large variations in cosmic gains for certain HV groups must be due
to the HV supply and nothing to do with the calorimeter.  One question is what to do in
the short term for 2014 running.  We can try moving the modules to open slots, or
just reseating them, both possible tomorrow.  Or measure voltages directly at the
calorimeter and adjust accordingly, ignoring mainframe's readback.   Or just correct
with gains, which could require some channel-dependent thresholds for data
acquisition.

- Nathan