Group EB R&D meeting, Feb 10, 2005
Tests of SLAC CFD-TAC-S/H boards in the PMT Scanning Setup
We use the same SLAC ADCs in all tests:- #1 for channels 1-32,
- #2 for channels 33-64.
By swapping the CFD input between the ADCs we verified that both ADCs measure consistent results.
Using the analog output from the CFD part of the board allows us to compare performance of the TAC-S/H part of the board to another TDC, in our case we used a Philips 7186 (16ch, 25ps/count).
Last report last November. Since then much progress.
- Much improved linearity over wider range (FET voltage).
- Improved signal shape and stability after many tests with S/H gate delay and width, yielding stable operating point.
- Biggest single improvement: raised CFD zero-crossing threshold.
Linearity:
Calibration of ps/count:
MCP pad scans:
-
Place red PiLas laser at center of each pad, take 100,000 triggers
in single-photon mode, read MCP #16 with SLAC TDC/ADC, fit single
Gaussian plus constant to peak.
- status Nov/Dec - we sometimes saw satellite peaks or double-peaks
- Nov 16, 2004 CFD 1
- Nov 16, 2004 CFD 2
- Nov 23, 2004 CFD 2
- Dec 16, 2004 CFD 10
- after FET fix:
- Jan 28, 2005 CFD 5 - good resolution
- Jan 28, 2005 CFD 10 -
many bad channels
- Jan 31, 2005 CFD 10, pad 44 - double-peak from SLAC ADC
- Jan 31, 2005 CFD 10, pad 44 - double-peak from Philips
- after zero-crossing threshold fix:
- Feb 9, 2005 CFD 5 - good resolution, 1 dead channel - best board so far;
- Feb 9, 2005 CFD 10 - good resolution, few borderline channels;
- Feb 9, 2005 CFD 2 - good resolution, few borderline channels;
- Feb 10, 2005 CFD 4 - good resolution, 1 dead channel;
- Feb 10, 2005 CFD 11 - good resolution, 2 bad channel, few borderline channels;
- Feb 11, 2005 CFD 1 - good resolution on "normal" channels.
Plots assume 25.5ps/count calibration.
Jumps in timing:
-
Take 100,000 triggers every 30 minutes for several days,
always in the same location on pad 24.
- Jan 21, 2005
drift and fluctuation of mean cause bad signal resolution - Jan 29, 2005 after FET fix
drift and fluctuation of mean seem to no longer affect resolution as much (confirmed by Philips)