OK, so what I did

1) Compare the sensing noise between ACAV and RcAV loop.

    Measure sensing noise from RCAV loop. it is about 30e-9 [V/rt Hz] flat which is much lower than ACAV sensing noise (1e-6 [V/rtHz].)

    I checked RCAV noise at error point(loop closed). It is ~ 50e-9 [V/rt Hz] which is higher than the sensing noise.

    I haven't measure the input referred noise for RCAV servo yet.

2) so I switch the RFPDs to check if ACAV's RFPD is broken or not, and measure the sensing noise again.

    It got the same result ~ 1e-6 [V/rtHz] flat, so both RFPDs give the same noise level.

3) I realize that we have an amplifier for ACAV's RFPD before the mixer, so I remove the amp and measure the sensing noise, the noise level

   is ~ 30 e-9 [V/rt Hz] flat. So this shows that the high sensing noise comes from the amplifer

4) I check the beat measurement again, now the gain for ACAV loop is set to 7.2 (instead of 1.5), where the noise at the error point is smallest.

It starts to get higher around 7.5. The beat noise does not change between having the amp or not. The data is valid upto a few kHz.

5) The Noise at error point for ACAV loop after removing the amplifier, with sensing noise and input referred noise is still weird. RFPD noise

is still higher than the error noise, I'll think about this.

first, the TF of the PDH box, gain 7.2.

then, the ,sensing noise, input referred noise and noise at error point.