I previously reported results from DARM offset step tests in October and last week. The overall goal here is to figure out what overall effect the power outage has had on the IFO, especially since we have lost 1% optical gain. For example, see this alog about comparing modulation depth tests.

However, after looking closer at the DARM offset measurement results, and especially investigating the effect of whether leaving the OMC ASC on during the measurement matters, I would like to revise the previously reported measurement results.

Some background: this test aims to measure the contrast defect light by changing the DARM offset in mA while injecting strong PCAL lines to capture the DARM optical gain. We expect the relationship between optical gain (mW/pm) and DARM offset power (mW) to be quadratic. We assume by fitting the data to a parabola, the resulting y-intercept can tell us the contrast defect light on the DCPDs, if we had zero DARM offset.

Some considerations:

First question, does leaving the OMC ASC on matter? Jennie has been doing a lot of these measurements, and her experience has been, yes, because the OMC ASC offsets are set at one DARM offset and are not reset at each different offset. We do not expect the OMC alignment to drift significantly during the 15 minutes of this measurement, so the best practice has been to turn the OMC ASC OFF for the measurement. However, I didn't do it the first time I measured last week, but I got significantly better results (using Craig's code) with the ASC ON than I did with the ASC OFF.

Second question, what is the best way to fit this data? This apparently has been a question for a long time, see Gabriele's comment to 30573 in 2016. I believe the model I described above suggests that the vertex of the resulting parabola should be centered around x=0, that is zero optical gain. However, I discovered this week that Craig's code, which I used to report the results in October and last week (see first two links), fits a nonzero vertex (i.e. y = a(x-x0)^2 + b instead of y = ax^2 + b). Of course, that will change the answer significantly! I don't understand the mechanism that would cause us to move the center of this vertex away from x=0, so I refitting the data to follow y = ax^2 + b.

Results:

I am revising the fitted contrast defects to be:

Conclusions:

This measurement shows that the contrast defect is higher now than it was in October. I don't know if we can attribute all of this to the power outage; we lost 1% optical gain comparing kappa c from before and after the vent, giving us at least 2% less optical gain between October and now. It also shows that there is perhaps a small effect related to the OMC ASC being off: slightly higher contrast and a better overall fit.

This also increases our upper limit on the possible homodyne angle- in October I stated that the homodyne angle upper limit would be about 7 degrees, but with this revised value it should have been more like 8.8 degrees. Now it is 9.3 degrees.