Microphone preamp box had a low-pass filter at 2kHz, Ayaka changed it to 20 kHz by replacing 100pF capacitor with a 10pF.
We've measured frequency response of the box. Signal from the microphone was split into two. One path went to the box, while another was amplified by the gain 20 (and bandpass filter 1Hz - 300kHz) and sent to spectrum analyzer. Coherence and frequency response were measured using box output and amplified input. Low-pass filter in the box does not limit our sensitivity.
Acoustic noise significantly decreases at frequencies higher then 2kHz. So we need to modify the circuit by adding whitening filter.
I've plugged in PMC length channel into PEM board CH15 through and amplifier (gain=200) that is AC coupled to avoid ~2.5 DC V coming from PMC servo. I measured coherence with microphone that was located ~30 cm higher. Measurements show contribution of acoustic noise to PMC length in the frequency range 20-50 Hz. In this range PMC length / MC length coherence is ~0.5.
Acoustic noise couples to PMC length in a non-stationary way. 5 minutes after the first measurement I already see much higher contribution. This was already discussed here. I've made C1:X02-MADC3_TP_CH15 a DQ channel at 64kHz. This a fast PMC length channel.
Next step will be to use several microphones located around PMC for acoustic noise cancellation.