These are transfer plate positions (approximately the SLM reference DCOPS position) as fit by Cocoa vs Z in CMS coordinates. Clearly something is deeply wrong.
For SLM 1 (plotting transfer lines 1 and 4): the minus points average promisingly.
For SLM 2 (plotting transfer lines 2 and 5): +3 and -3 are different and not plotted.
For SLM 3 (plotting transfer lines 3 and 6): the radius increases as you go inward.
The displacements in the tangent and vertical directions are odd. The tangents scatter quite a bit, but are otherwise reasonable.
The radial displacements are huge, and of two distinct types: 1,3,5 have one type and 2,4,6 have another (opposite) pattern.
This is reflected in angle shifts of the transfer plate reference positions, which are close to the SLM reference DCOPS. Note that the transfer lines themselves are supposed to be at 14.052, 74.099, 135.901, 195.948, 254.052, and 314.099 degrees respectively).
The lasers are at slightly approximately the right positions wrt the SLM angles:
In transfer plate coordinates this looks different. Note that abs(y) varies by less than 5 microns.
The raw intersection points from Cocoa (a collection from the last few iterations). Transfer Lines 3 and 4 are plainly rubbish.
Using a better datafile (0T) reduces the xfit-xpg rms from 6mm to 3.77mm. Still not good, but a great improvement. The rms of the "normal" x fit positions is now 3.5mm (pg is 1.45mm), which in quadrature is 3.78mm; and a scatter plot shows random scatter. It doesn't vary much by transfer line.
Modified 18-September-2009 at 09:27
|Previous notes||Next notes||Main slide directory|