J. Kissel, B. Shapiro, J. Garcia
After some mechanical adjustments on H2 SUS ITMY M0 made by the assembly team,
- F1 flag dis- and re- assembly at TOP stage
- Moving mass forward on UIM
- Recovering UIM and PUM signals, aligning flags,
Jeff G. has taken another set of transfer functions to asses how we did (Last Tuesday, 2011-11-29). In summary, the main chain looks better (but still not great -- Pitch as usual) but not as good as we'd like. We believe there are several issues going on:
- The dynamics are different from the model, because the d's between the top stage and the UIM stage (parameters dn and d1) are not dead on. (yellow flag)
- The large offset in the UIM ballast mass (i.e. having it fully forward (HR side)), is causing that stage's horizontal center of mass to be offset from the center line of the suspension (represented by h1). (yellow flag)
- The overall magnitude of the Pitch transfer functions are a ~50% lower than the model. (yellow flag) This may just be the accuracy of the measurement calibration factor.
- There is an excessive amount of Longitudinal coupling into Pitch. This has been reduced with Travis flag dis- and re- assembly, but some cross coupling still remains. (red flag)
I attached four plots for your perusal. All are M0 TOP to TOP transfer functions.
(1)allquads_111130_H2SUSITMY_ALLM0_TFs.pdf
Comparison between nominal model (BLUE) previous main chain measurements (ORANGE), and current main chain measurements (BLACK). Note that 2011-11-19 measurements of V and R are missing, because the M0RT OSEM had failed. We see hear, as before, that the degrees of freedom (besides Pitch) all line up quite well with the model, implying that the majority of the dynamics in the suspension are free and well.
(2)2011-11-29v2011-11-19_H2SUSITMY_M0_P-P_TF.pdf
Zoom Comparison between nominal Fiber model, the first 2011-11-19 measurement, and the current 2011-11-29 measurement. Here, we see the good news that the severe cross-coupling between L and P has been reduced, but not to what we expect from the model (and from what we've seen on metal builds).
(3)2011-11-29v2011-11-19_H2SUSITMY_M0_P-L_TF.pdf
Model, 2011-11-29, and 2011-11-19 Pitch to Longitudinal cross coupling (compared against 2011-11-29 and 2011-11-19 Pitch to Pitch), quantifying the reduction.
(4)2011-11-29v2011-11-19_H2SUSITMY_M0_P-P_TF_modelcomp.pdf
Comparison between models with parameters varied, in order to try to explain what might be happening with the dynamics and the yellow flags mentioned above. I've tried moving around two parameters that we believe might effect the dynamics as we've seen (motivated by physical differences).
fiber = Nominal Model, with h1 = 0 mm, and both dn and d1 = 1 mm
fiber_h1plus5mm = Modified model, with h1 = +5 mm
fiber_dnd1plus1mm = Modified model, with dn and d1 = 2 mm (break off points are further away from the vertical center of mass, dn increased in +Z, d1 is increased in -Z)
fiber_fiber_h15mm_dnd11mm = Modified model, with both h1 = +5mm, and dn=d1=2mm
Again, we suspect that h1 is offset because there's a good fraction of the ballast mass on the HR side of the UIM, and we suspect the d's concerning the UIM might be off, because they (a) affect the two modes that are the most different in the model, and (b) these d's are adjustable, and are defined by blade tip heights a physical parameter difficult to mail down. One can see that, though both parameters effect the dynamics differently, the combination of the two explain the measurement quite well, specifically offsetting the horizontal center of mass at the UIM explains some of the high-frequency cross-coupled length resonances.
-----------------------
Our best guess up to this point as to where the remaining cross-coupling is originating from is that F1 is not driving as much as we think. The idea being that we intend drive equally in (F2+F3) and F1, and because either the F1 (a) electronics, or (b) the OSEM coil-magnet pair results in less drive in F1, there is an imbalance in the drive, and (F2+F3) ends up driving more, which means that Length is excited more than expected.
Things I'm looking into in order investigate this claim:
- Looking at F1 response in a Length drive, compared against models. The thought is perhaps the monolithic is more susceptible to this particular flaw, so comparisons against metal builds may not be as enlightening, but we can check anyways. If there's a large discrepancy, this would indicate that the sensing part of the OSEM is at fault.
- Comparing Pitch / F1 response to reaction chains. The TOP and UIM masses are identical between the two chains, and should therefore have the same mass. Further, the total mass of the two chains is identical. This means at high frequency, above the resonances, where the transfer function should be just as a free mass (F = m a, or T = I a), the chains should be the same.
Measurements we can do in order to better identify the problem
- Measure the OSEM basis (F1 to F1, F2 to F2, etc) transfer functions, and compare against the model (~2 hours of measurement, 1 days worth of analysis). If the F1 to L check doesn't turn up anything, then if this comparison with model shows something strange, then we know it's the actuator side of the OSEM that's failing.
- Measure the F1 response at DC (at a few different drive levels, using offsets in the COILOUTF banks), then replace the OSEM with a new OSEM from "off the shelf," and perform the same measurement. (1/2-a-day of measurement, 1/2-a-day of analysis) If there's any change, you know it was the bad OSEM. If there's no change, then we know the OSEM coil-magnet pair is OK, and it's some further flaw upstream in the electronics.
- If the OSEM turns out not to the be the problem, we can drive equal amounts of digital signal from the DAC, and measure the response at the mock in-vacuum feedthrough. (1/2-a-day of measurement, 1/2-a-day of analysis).
