Without further ado, here is the current gyro noise spectrum:
The traces are:
- The gyro signal from the AOM actuation signal
- The gyro signal from the PLL actuation signal
- The "spillover" noise, generated by calibrating the primary error signal into frequency noise via the optical gain and then into rotation rate noise using the standard factor
- The displacement noise coupling through the FSR modulation pathway, generated by calibrating the primary actuation signal into displacement, and then calculating the rotation rate noise that results from the change in the cavity length/area. In the interest of openness, I have actually scaled this contribution up by a factor of ~2 so that it matches up with what we see in the PLL. I justify this by saying that it is only an estimate in the first place, and that looking at the primary control signal does not tell us about all the types of cavity mirror motion that contribute to FSR noise.
- The sensing requirement
A few interesting things:
- We are now clearly in the region where we see the improvement in noise from looking at the transmission vs. the AOM actuation
- While the spillover noise again explains the high-frequency part of the gyro noise as measured in the AOM signal (as it did in the last gyro budget), it is absent from the transmitted signal. We thought we would see this noise coupling in here, too, but I guess there is some rejection/suppression. It may pay to go back and look at our noise/signal diagram to see if this is consistent after all.
- We are very close to being dominated by the "unavoidable" displacement noise coupling (in fact, we are in a few narrow bands). From here, using this gyro setup, we can only improve the sensitivity by actually reducing the coupling of environmental noise to the gyro components. This is essentially the "design sensitivity" of the mGyro.
- We are now only a factor of ~50 away from the sensitivity curve from ~200-400 mHz. This is an improvement of about a factor of 20,000 compared to when we first put the PLL together in August---something we should all be proud of!
- Below 200 mHz, we get farther away from the requirement (because the requirement goes down and our sensitivity gets a bit worse). Of course, we are now in the regime where we might begin to see some legitimate seismic activity; we need to do some real analysis with the seismometer data we took a while back to see if this could be the case. In fact, the peak at just below 100 mHz looks like the standard microseism, only I think this is typically a little higher in frequency (though perhaps not significantly??).
- I found that a lot of our low-frequency noise was caused by occasional glitching from narrowly margined loops. Reducing the gain of both loops by a small amount caused them to go away for the most part, and this improved our sensitivity below 1 Hz noticeably. It is possible that this can be done a bit more to some degree.
Here is a picture of the enclosure Alastair and I fashioned for the transmission demod setup and another picture of the one Alastair made up for the two input steering mirrors that did not fit in the IO enclosure: