filiberto.clara@LIGO.ORG - posted 13:37, Friday 12 February 2016 - last comment - 19:47, Friday 12 February 2016(25516)
OMC and SR3 Sat Amp Units Modified
While investigating the high frequency oscillations seen on SR3, we decided to modified the satellite amp units for OMC and SR3 (Bottom). Two decoupling capacitors were placed on the input of the negative regulator (U503), as implemented at LLO. C602 0.1uf C601 10uf The OMC units modified are S1100129 and S1100127. For the SR3 unit, we removed unit S1000287 and replaced with a modified unit S1100074. High frequency oscillations 1.7KHz and 1.9KHz seem to have cleared. Work to continue next week, as some channels still show some noise on the spectrum.
SR3 should be back (but we cannot prove that until we see that the cage servo doesn't act up with pre-disaster gain.)
In the first attachment, we can see that the SR3 M3 WIT_PMON and YMON seem to have gone back to the old peak-to-peak.
In the second attachment, top right shows the comparison of now (red) and before (blue, green) of SR3 M3 LL OSEM. After the modification (red), 1.7kHz and 1.9kHz lines obvious in blue and green are gone. I'm sure that the SR3 is fully back..
Top left shows the four SR3 M3 OSEMs as of now. None of them show kHz oscillation lines. LL is the noisiest, LR is the most quiet. There are some ugly 60Hz harmonics, and there are also a bunch of peaks that are not the power line harmonics (e.g. 142Hz). I cannot prove it but I suspect that these peaks were there before.
The bottom left shows OSEM outputs of OMC, PR3 and SRM. SRM and OMC don't have ugly power line and whatnot peaks, but PR3 look similar to SR3.
Even though SR3 seems to be back to its old self, fast OSEM is still copied to EPICS without decimation and pollutes cage servo when the oscillation comes back (and the OSEM RMS is anyway dominated by high frequency junk even when it's not oscillating).
Since these are only used for cage servo and as slow references, I put 60Hz 3rd order Butterworth LPF to all of the SR3 M3 OSEMINF (attached). This is a mild one and should not interfere with the cage servo, but reduces the high frequency junk leaking into WIT_PMON that is used by the cage servo.