Executive summary:

Three QT32C channels had pedestals of zero ADC.  (strike through doesnt' display - We find that the mean of the pedestal is different for these channels when the vped is set so that the voltage out of the Rx is 0, than when the CAMAC is off, which is puzzling.)   This puzzle has been solved - we were not looking at the channel we thought we were.

Next Steps:
1) Use a scope to tune vped so that 0 mV is coming out of the Rx.  Plug the Rx back into the QT and take a pedestal.
2) Turn off the CAMAC and take a pedestal
3) unplug the cables from the QT and take a pedestal
4) Swap a "good" and "bad" channel at the Rx and take a pedestal.
5) Measure the input voltage across the QT with a voltmeter.

Details:

This is a continuation of drupal.star.bnl.gov/STAR/blog/rjreed/epd-pedestal-issue

3 EPD channels were having pedestal ADCs of 0.  These were:
EQ2 BD10 CH16 - EW0 PP7 TT6 - Board 7 in pedestal files - CAMC slot 14 ch 3
EQ2 BD13 CH17 - EW0 PP3 TT3 - Board 9 in pedestal files - CAMAC slot 5 ch 1
EQ2 BD14 CH10 - EW0 PP6 TT4 - Board 10 in pedestal files - CAMAC slot 12 ch 2
Where I count slot 1 as the very first slot on the left and ch 0 as the very first channel on the bottom

The vped and bias setting prior to any changes as a result of this issue were:
EW0PP3TT3 - Vped 9, Vbias 59.4579
EW0PP6TT4 - Vped 4. Vbias 58.8095
EW0PP7TT6 - Vped 12, Vbias 56.7277

The CAMAC off run was - drupal.star.bnl.gov/STAR/system/files/eq2.mean_.21168009.txt
There was a time difference of 20 - 39 minutes between when the CAMAC was turned off and the data was recorded.
EW0PP7TT6 - mean = 29, sigma = 0.54348
EW0PP3TT3 - mean = 33, sigma =  0.53627
EW0PP6TT4 - mean = 38, sigma =  0.53990

These values all seem nice and normal.  For run 21170028 I tried to put the vped for all 3 to 100.  This caused the TAC to not fire.  From 21170034 onward I changed the vped of these 3 channels to 20.  This has given a non-zero pedestal for all 3 since then.

Checking Run 21172027 I find:
EW0PP7TT6 - mean = 13, sigma =  1.62285
EW0PP3TT3 - mean = 14, sigma =  2.87764
EW0PP6TT4 - mean = 28, sigma =  2.45365

For the access on June 22 we had:  Mike, Skipper, Tim, Akio and Rosi
We thought we were looking at EW0 PP3 TT3 (slot 5 - channel 1 but Tim notes he counted from the right so this would be slot 20 channel 1 (EW0 PP10 TT2, or EQ3 BD4 CH8)l:
vped = -80, scope = 28 mV
vped = +20, scope = -14 mV
vped = +80, scope = -40.8 mV

For this channel (which changed since last year) my calculations indicate the vped value should be -18 looking at the ADC pedestal which is close to what I would guess if I did a linear fit (-13 below).

Checking a good channel (which we thought was slot 1 channel 0 - EW0PP1TT1 - EQ3 BD3 Ch8 - listed in pedestal file as board 2) This would be slot 24 Ch 0 which EW 0 - PP12 and associated with the "zero" tile.

vped = -80, scope = 40 mV
vped = +20, scope = -2.8 mV
vped = +80, scope = -28.4 mV
vped = nominal (-2), scope = 0.4mV
The CAMAC off pedestal for this channel was mean = 85 and sigma = 0.47386

Ranges seemed similar though values were not.  To convert -
(5mV/50 ohm)x75 ns = 7.5 pC = 30 ADC

(Strike through For the nominal setting of vped, we would expect a scope reading of 0 mV.  For this bad channel, if I look at the scope settings, I would have guess a vped of about -13.  However, a setting of that value would give a 0 adc pedestal.  The first non-zero value for the pedestal was with a vped of 20.

This means it is impossible to choose a vped that gives a non-zero adc that would result in no net DC offset.)

We took 3 pedestal runs:
21174038 - Nominal pedestals
21174040 - All ch to vped 80 (some errors, but I think as a result of this large value)
21174041 - All ch to vepd -80

Unfortunately the repair of the chilled water meant that the VME crates were shut off at this point.  Once that repair was completed they were sweeping for the beam so we could not look at this any further.

It is still not clear what the error is.  Mike believes it is the QT board.  But it could be the FEE or diffRx.   I have my suspicions on the connector - the output to the scope seems reasonable for these channels, and the camac off seems reasonable.  But connected one gets something unreasonable.

We will think on this further.