Yutaro, Masayuki, Tomo, Kiwamu,

In the course of the experiment today, we happened to find mysterious noise in the ALS PDH loop.

However, this noise turned to be likely due to some readout nonlinearlity in some op-amps in the SLOW DAQ path only. We believe that this noise has no impact on the ALS performance.

[Symptom]

In short, noise measured at SLOW DAQ was found to be higher than SLOW MON which should be identical to the former by design.

This noise was found in the configuration where the IR frequency was directly locked to the X arm (7289). The ALS PDH signal was used to evaluate the ALS sensing noise that may be in the ALS part of the system. The thing was that the error signal of the ALS PDH loop without closing the loop at a 00 resonance was found to be lower than the one estimated from the feedback signal (i.e., SLOW DAQ) when the loop was closed.

Later, we learned that the SLOW DAQ channel showed higher noise below approximately 10 Hz than SLOW MON channels only when the ALS loop was closed by a factor of several (this factor seemed to vary as a function of time). For example, when the loop was open by physically disconnecting the cable going to the VCO, these two signals show exactly identical signals.

[High frequency saturation or some sort]

After several tests, we discovered that a similar situation was reproducible by injecting a very high frequency signal to the ALS PDH common mode board. Using Moku Lab, we injected an RF signal at around 30 MHz with an amplitude of about 600 mVp-p. This gave us a similar situation where SLOW DAQ showed higher noise below 10 Hz. So we determined that excess noise in SLOW DAQ is due to some kind of high frequency saturation presumably in some op-amps in the SLOW DAQ path.

[A 100 kHz low pass newly installed for this]

We inserted a 100 kHz lowpass right in front of the PDH common mode board with a hope to reduce this excess noise. Indeed this resulted in lower noise in SLOW DAQ, but it is still too high typically by a factor of no greater than 2 or so below 10 Hz. The attached plot shows noise spectra of SLOW DAQ and SLOW MON after the insertion of the 100 kHz low pass. Note that discrepancy at high frequency is due to the fact that SLOW MON is limited by ADC noise. Since this noise is not in the loop itself, we don't expect any improvement in the ALS though.

Also, the diaggui xml file of the attachement can be found at

/users/Commissioning/ALS/PLL/ALSPLL_strange_slowsignals_20181211.xml

Even though this noise shouldn't impact on the actual performance, some of the past measurement may need to be revisited such as the one in 7256.