[Evan, Nancy]
Evan was trying to look for a signal for BEACON sensing (drive PCAL (at say f1), drive OMC ASC dither (at say f2), demodulate OMC ASC signals at f2-f1 and f2+f1 to obtain the sensing error signal). Similarly, in the demodulated the ASC signals at f2, ne would expect to see a signal at f1. Evan was not able to see this signal, so we tried to troubleshoot why that was. While we still haven't found the signal or the reason why it's not there, here's just an account of what we tried.
1. PCAL X was driven at f1 = 8.703 Hz with 10000 counts. The top left panel in the first screenshot shows the DARM signal with this PCAL line (for now, just look at the black, no offset line). DARM without PCAL is shown in the second screenshot, upper panel.
2. OMC ASC was dithered at the nominal frequencies (f2) are shown in the third screenshot.
3. These lines can be seen in DARM in the fourth screenshot, again, for now look at the "no offset" trace.
Now, one would expect a signal at f2+/- f1, so for example at the two vertical cursors in the fourth picture. As can be seen in the spectrum, there is no signal at those beatnote frequencies, so it's not surprising that they don't show up in the demodulated signals.
4. Now, given that both the PCAL and OMC dither lines can be seen above the noise, one reason one might not see the beatnote lines is if the angular coupling from OMC to DARM was small, which can happen if the OMC is perfectly aligned. For this reason, OMC was also misaligned to check if that would make the lines show up. All these settings are shown in the fifth screenshot. Now, one should look at the misaligned traces in the fourth screenshot - misaligned in POS X in the top panel, and misaligned in ANG X in the bottom panel, by offsets of 1 to 3. As we see, even misaligning the OMC does not make the beat frequency show up in DARM. Also shown in the second screenshot is the fact that PCAL line didn't change with the OMC misaligned.
The absence of beatnotes even with misaligned OMC led me to suspect that perhaps OMC dithers are completely uncoupled from DARM, and maybe that's why dither alignment efforts have also not worked. To verify this, I checked the heights of the dither lines with different levels of OMC misalignment in the fourth screenshot - these are without changing the dither drive counts, so the peak heights change represent the coupling between dither and DARM, which can be seen is non-negligible. For instance, the height of the 2175.1 Hz line changes from 5.63e-6 with the OMC aligned to 3.18 e-5 with the OMC misaligned with an offset of 3 in POSX. Similarly, the 2200.1 Hz line changes from 8.9e-7 with the OMC aligned to 3.68e-6 with the OMC misaligned by an offset of 3 in ANGX.
5. Finally, I suspected that perhaps the low SNR of the PCAL line was the suspect, so I tried a pcal line around 20 Hz, and modified the dither lines to allow separation of expected beat frequencies. The settings for that are shown in the sixth screenshot. This didn't show anything new either, so I didn't bother making nice plots for that.
More notes and thoughts on this later, but for now just wanted to put the measurements in here.
We also briefly tried out the deacon infrastructure. We were able to excite the EX drumhead mode by a factor of a hundred or so over its quiescent amplitude over the course of several tens of seconds while the interferometer was acquiring lock, but seemingly not when it was in nominal low noise. I suspect (but did not verify) this is because the EX ESD amplifier switches from high to low voltage near the end of the acquisition sequence. (That would be consistent with D1500016, which shows the PI signal is summed into the ESD amplifier chain before the high/low-voltage selector stage.)
The drumhead amplitude control screen suggests that between the OMC-ASC_BEACON_C_BANDPASS and OMC-ASC_BEACON_C_CLOCK_RECOVER filter modules there should be a saturation feature so that a constant-envelope sinusoid is sent to CLOCK_RECOVER, but I did not observe this behavior, so I (ab)used the limit button in CLOCK_RECOVER to mimic it.
I cannot say we got a robust amplitude lock of the drumhead amplitude going, even in the high-range ESD state, but that is probably just a matter of tuning the loop shape more carefully. The larger question is whether there's enough range in the low-noise state to actuate on the mode amplitude.
Our deacon testing occured between 2022-07-27 18:45:00 and 2022-07-27 22:00:00 UTC. To set the drumhead amplitude setpoint, we were adjusting the offset in OMC-ASC_BEACON_C_AMPCTRL.
Keita pointed out that unlike what I said previously, the saturation behavior does seem to be present between the BANDPASS and CLOCK_RECOVER.
We also tried adjusting the feedback phase, though trending through SWSTATs I cannot figure out exactly how. We may have been adjusting the zpk coefficients in FM2 of SUS_ETMX_PI_UPCONV_UC8_INP, which was a (probably redundant) attempt at filtering away the square wave behavior of the saturator.
