It wasn't clear to me whether the distribution of means was a due to Shon's test rig or to the different capacitances of the wires systematically crimped into similar locations on each cable by the manufacturer. On the assumption that it was due to wire to wire variations in Shon's test rig, I optionally "corrected" the times by adding the mean differences.
If we could subtract out wire to wire variations in time, it could clean up our stub finding. Bob Handler has been looking at the time offsets between ganged cells, in order to make our Z measurements more accurate. He says that ASD card swaps do not noticeable change the time offsets for a pair of ganged cells. (The time offset is the correction you must add to T_stackodd - T_stackeven to have the delta-time=0 refer to the vertex end of the chamber. Channel to channel variations keep this from being 0.) In addition, he says he does not see much effect when a pre-amp is swapped out. The TDCs are very uniform, so the variation must be in either the 50-foot cables from the chambers to the ASD, or the 200-foot flop-n-flat cables going upstairs.
Bob has some good numbers for the difference between timing for pairs of cells. So I generated the differences for the same pairs for wires for Shon's data, and compared the two numbers. The upper distributions show the two cases for wire numbering, with the left one being the customary scheme. The lower distributions are the same, but first adding the "corrections" which assume that the pattern of mean distributions was due to Shon's apparatus. In no case do I see clear correlations.
Possibly I have the cable numbering wrong. If so, they are pretty thoroughly mixed up. I looked for matches to a pattern from Bob's list in Shon's. I don't find anything useful--there's a candidate, but with enough variation to make using it problematic. I give up.
Modified 18-Mar-2003 at 10:41
http://hep.physics.wisc.edu/~jnb/imu/19Mar2003