If I compare the average chamber position as found using the ASPD P4 positions (from CMS-SG-UR-0103) Z4 with the chamber position as found using the coded targets (from CMS-SG-USR-0058) ZTARG, and combine with the difference between the Z of Celso's fit and mine for the chamber centers dZ I get the following picture:
Although the statistics are small, it seems as though adding the dZ (Link fit minus my fit) to the difference between the P4 estimated chamber center and the coded target chamber center reduces both mean and Rms. This is odd, because I'd expect the opposite: the Link-My difference subtracted from another Link-My fit should largely cancel.
The chamber center estimates for X and Y I take from the alignment pins. The Z of the alignment pins may be discounted. The Z of the chamber I estimate in two ways: by the coded target and by the average of the (ASPD P4 target minus the appropriate ASPD mount height). The Z's are relative to the disk center, as usual.
|chamber||X||Y||Z from P4||Z from coded||Zcms from P4||Zcms from coded|
Shifting to the appropriate SLM coordinate systems gives for the P4 estimates:
And from the coded targets
The result of the change is that my Z chamber center estimates are now even farther away from Celso's. I am using the post-Craft PG estimate of the disk center of 7565.57, which is .57mm out from the ideal. My chamber Z fits are (now) .49mm closer in than Celso's.
Comparing ASPD positions is instructive:
|name||Celso X||Celso Y||Celso Z||Link SDF X||Link SDF Y||Link SDF Z||My X||My Y||My Z|
The results look both promising and odd. I can reproduce Celso's X and Y, but not his Z, using the SDF file he gave me. However, I can reproduce his Z's at the 44 micron level, though not his X and Y, using my ME+1 SDF file.
|-Celso X||-Celso Y||-Celso Z||-Celso X||-Celso Y||-Celso Z|
Celso points out that the file he gave me was for the phase-1 fit, and that the link-disk position is shifted in the final fit. When I compare the final fit to this value the answer is (surprise!)
So I can replicate his fit with his file (correcting for the phase-3 fit), and get his Z's (roughly) with my own file. That leaves the X and Y differences for the ASPD positions, and of course the Z differences for the chamber centers. Which currently are (his minus mine):
I don't have any good explanation yet for why we should agree on ASPD positions but not Z. Rotations are the next suspect.
The rotations for the chambers about the chamber axis (between alignment pins) is, over 400 mm
Clearly the rotations resulting from my fits aren't going to account for the 500micron level changes shown in the previous table.
For Celso's fits the results are different. The width is 554mm, for starters, and the results are much larger:
In order to determine what the shift would be at the center of the chamber, just divide these by 2. Notice something odd about the list: there are duplicates, just with the opposite sign.
If I subtract the chamber tilt component from the difference between Celso's numbers and mine I get a rather interesting plot. Everything shows up in two distinct regions, most at about .65 or so. The other two are from HSLM4 and 6.
Modified 27-March-2009 at 10:16
|Previous notes||Next notes||Main slide directory|