Look at the effects of the ADC voltage range on the ADC noise floor.
ADC input was terminated with 50 ohms. We then looked at the channel with DTT. This was at +/- 10 V range. We used C1:SUS-PRM_SDSEN_IN1 as the test channel.
The map.c file (in /opt/rtcds/caltech/c1/core/advLigoRTS/src/fe/ ) then had two lines added at line 766.
//JCB temporary 2.5V test, remove me
adcPtr[devNum]->BCR &= 0x84240;
This hard coded the 2.5 V range (we default to the 10 V range at the moment).
We then rebuilt the c1x02 model and reran the test.
Finally, we reverted the code change to map.c and rebuilt c1x02.
I've attached the DTT output of the two tests.
It appears the ADC is limited by 1.6 uV/rtHz. Hence the increase in noise in counts by a factor of 4 when we drop to +/- 2.5 V from +/- 10 V.
We changed the range of the two SUS AA boards in the corner from +/-2 V to +/-10 V by changing the supply voltage from +/-5 V to +/-15 V. The change was made by switching the AA power feed wires on the cross connect. The max supply according to the spec of DRV134/INA134 is +/-18 V.
We checked the new range by applying the voltage to the input of AA and measuring the output going to the ADCs. The local damping MC1,2,3 appears to work.
Now that we have increased the range of the AA to +/- 10 V I have increased the PRM side OSEM transimpedance from 29 kV/A to 161 kV/A by changing the R64 in the satellite box. The first attached plot shows the ADC input spectrum before and after the change with analog whitening turned off. The PD voltage readback went up from 0.75 to 4.2 V. The second attached plot shows the sensor, ADC, and projected shot noise with analog whitening turned on and compensated digitally. The ADC calibration is 20 V/ 32768 cts. The PRM damping loops are currently disabled.
I checked for oscillation by looking at the monitor point at the whitening board. There was no obvious oscillation on a scope - the signal was 20 mV p-p on 1 us scale which was very similar to the LL channel.
We missed a factor of 2 in the ADC calibration: the differential 16 bit ADC with +/-10 V input has 20 V per 32768 counts (1 bit is for the sign). I confirmed this calibration by directly measuring ADC counts per V.
So the ADC input voltage noise with +/-10V range around 100 Hz is 6.5e-3 cts/rtHz x 20V/32768cts = 4.0 uV/rtHz. Bummer.
The ADC quantization noise limit is 1/sqrt(12 fs/2)=1.6e-3 cts/rtHz. Where the ADC internal sampling frequency is fs=64 kHz. If this would be the limiting digitization noise source then the equivalent ADC input voltage noise would be 1 uV/rtHz with +/-10 V range.
We realized that the SRM sensors are connected to the readout but just sitting on the BS in vacuum table with no magnets and therefore no shadows in them. We swapped the inputs to the SRM and PRM satellite boxes to use the higher transimpedance gain of the PRM side sensor. The attached plot shows the current spectrum in this configuration. The PD readback voltage was 9.5 V. Since this is close to the rail we put a slightly higher voltage into the AA of this channel to test that we can read out more ADC counts to make sure we are not saturating. The margin was 15800 vs 15400 counts with p-p of 5 counts on the dataviewer 1 second trend. We returned all cables to nominal configuration.
The calibration from A to m is 59 uA/1 mm.
I measured the SRM OSEM (no magnets at the moment) noise out of the satellite box with a SRS785 spectrum analyzer. I inserted a break out board into the cable going from the satellite box to the whitening board. The transimpedances of the SRM OSEMs are still 29.2 kOhm. The DC voltages out of the SRM satellite box are about 1.7 V. The signal was AC coupled using SR560 with two poles at 0.03 Hz and a gain of 10.
The noise is consistent with the one measured by the ADC except for the 3 Hz peak which does not show up in the ADC spectrum from Sunday. The peak appears in several channels I looked at. The instrument noise floor was measured by terminating the SR560 with 50 Ohm.
I recommend to change all OSEM transimpedance gains from 29 to 161 kV/A. Beyond this gain one will rail the AA filter module when the magnet is fully out of the OSEM.
The OSEM noise at 1 Hz is about factor of 10 above the shot noise. The damping loops impress this noise on the optics around the pendulum resonance frequency. Also the total contribution to the MC cavity length is sqrt(12) time the single sensor as there are 12 OSEMs contributing to MC length. The ADC noise is currently close but never the less not limiting the OSEM noise below 100 Hz. It can be further reduced by getting an extra factor of 2-3 in whitening gain above ~0.3 Hz. The rms of the ADC input of the modified PRM SD (R64 = 161 kOhm) channel is 10-20 cts during the day with damping loop off and whitening on.
The transimpedance amplifier LT1125CS is also not supposed to be limiting the noise. At 1 Hz the 1/f part of the noise: In<1pA/rtHz and Vn<20nV/rtHz.