We remeasured the input referred noise(IRN) of the UPDH box, and sensing noise (RFPD + mixer +LO) for ACAV loop.
Sensing noise is ~2 order of magnitude higher than UPDH's IRN.
We do this because we want to check if electronic noise in ACAV loop is limiting our beat signal or not.
Data from yesterday is not good because I did not cover the whole frequency for PSD measurement,
and got the wrong TF for UPDH, and wrong sensing noise. So I redo the measurement and plot the noise
in Vrms/rt Hz instead of converting it to frequency noise for debugging purpose.
To measure sensing noise (RFPD+mixer+LO)
1) block the beam on the RFPD and measure the signal after the mixer (after the low pass filter too, of course.)
the noise is flat and roughly around 1 uV/rtHz.
To measure UPDH's IRN
2) terminate the UPDH with 50 ohm (it is self terminated, so just unplug the input) and measure the output of the PDH,
boost off, gain 1.5 on the knob.
3) measure the TF of the UPDH
Both measurements are not limited by SR785's noise floor which is ~ 5 nV/rtHz flat at 100 Hz and above.
3) divide the UPDH signal from (2) by TF from (3) to get IRN.
The IRN (cyan) looks good. At 100k where the gain is 40 dB and at 10k where the gain is 20 dB,
the IRN is lower than the raw data by 2 and 1 order of magnitude respectively, and the crossing happens at UGF (1kHz),
so everything looks ok so far.
However, the noise from the error point (purple) (same point for sensing noise) when the systems is locked does not match the noise we have now
(data from yesterday.) I'll redo the measurement to see if it's the same or not.
I also notice that the sensing noise alone is already higher than the beat. The slope of the error signal is
cavity's FWHM / Vpk-pk = 108 kHz / 428 mV = 0.25 MHz/V.
So either the measurement for RFPD is wrong, or the slope of the error signal I got is wrong. This will be investigated next.
Just for fun, I match the 120Hz peaks on the RFPD noise and the beat, the calibration factor is 3.78e4 Hz/V. Although I did not take the contribution from other electronics into account, the RFPD noise sits a bit below the current beat. If this were true, we will have to fix the RFPD next.