Summary:
OPLEV EX laser glithes were reported to be causing DARM glitches by detchar (e.g. alog 31810) even though EY is no worse or maybe somewhat glitchier (1st attachment, this is a week's worth of trend before Jason made an adjustment). It seems like this is due to a hidden design feature that there's a fixed 0.1uf cap across transimpedance resistor, which gives different pole for different resistor.
As a quick fix I'll insert digital zp(160,16) for all EX segments, as it seems to me that we cannot change the transimpedance without accessing the receiver module on oplev pyron.
Details:
T1600085 tells us that the effective transimpedance and the whitening gain of ETMX and ETMY are [2*10k,18dB] and [2*100k,0dB] (a factor of 2 comes from the differential drive, physical transimpedance is either 10k or 100k). They both use two stages of whitening filters too.
D1100290 shows that there's a 0.1u cap across the transimpedance resistor. The pole formed by 0.1uF cap and [10k, 100k] resistor is [160Hz, 16Hz].
Analog signal of each segment including whitening gain is
ETMX ~ 8*2*10k*zp([],160Hz)*photocurrent (a factor of 8 due to 18db whitening gain)
ETMY ~ 2*100k*zp([],16Hz)*photocurrent.
DC level is about the same but ETMX has an effective zp(16,160) whitening relative to Y. Look at RIN (second attachment) and PIT and YAW signals (third).
Residual power glitch that isn't canceled out by power scaling goes into the oplev PIT, and the effect is worse for ETMX due to this difference because X and Y has the same damping filter with similar gain. For f>300Hz the PIT signal is limited by the electronics noise, so the power glitch at this frequency directly goes to oplev damping without any cancellation.
Inserting zp(160,16) for all EX segments will, for the moment, make EY and EX about the same.
Added FM10 (named "alog32115") to H1:SUS-ETMX_L3_OPLEV_SEG1, SEG2, SEG3 and SEG4.
Some old ISIINF channels were still defined in the foton file but not in the frontend, JeffK confirmed that all of them were safe to remove from the file. These channels were automatically purged from the foton file when we saved it.
Filter coefficients of the SUS-ETMX model were reloaded, I enabled the new filters, and they were accepted in SDF safe.
This introduced a new 16Hz pole (and 160Hz zero) only to ETMX oplev damping, which should have made EX and EY oplev plant identical (or more similar to each other) which is a good thing.
But there might be some impact on ASC. If this makes HARD (or SOFT though less likely) unstable, turn FM10 off for the moment.
Note that I didn't make similar changes to ITMs yet, though they both use 10kHz transimpedance resistors as these are not of immediate concern. Longer lever arm of ITM oplevs means that the equivalent angle noise of the oplev sensing electronics for ITMs are roughly a factor of 6 smaller than ETMs.
J. Kissel I've checked the open loop gain transfer functions of the ETMX optical lever pitch loops after this change. As expected from the frequency content of the new filter, there is no significant difference in the OLG TF with these new filters ON vs OFF. I attach the OLGTFs: - RED: With Keita's change OFF - BLUE: With Keita's change ON - BLACK: Same transfer function for ETMY (which has had no change) The template lives here: /ligo/home/jeffrey.kissel/Templates/ ETMX_L2_OLDAMP_P_OLGTF.xml ETMY_L2_OLDAMP_P_OLGTF.xml
[Travis, Jenne]
We have also accepted these FM10s in the Observe SDF file.