WP 6544, ECR E1700107,

Related alogs: 35115, 35139

We have installed the infrastructure in the frontend models to relatively easily produce the DCPD cross correlation spectrum.

It seems to do what it is supposed to do. See the first attachment for demonstration of data acquisition and calibration using DTT.

[Additional model changes]

In addition to what we have reported yesterday (35115), we implemented two additional minor changes today.

• We lowered the data acquisition rate of CAL-DELTAL_A/B to 4096 Hz.
• We added an EPICS monitor for CAL-DELTAL_A/B.

[Front end settings and other settings]

• We disabled the simple bypass for the NULL signal by shutting off the output of OMC-READOUT_ERR_NULL.
• We added a sentence in the ISC_LOCK guardian to update OMC_READOUT_SIMPLE_NULL_GAIN when it updates the counter part for the SUM signal.
• Installed the latest inverse sensing (33025).
• Installed the numerical whitening filter (six zeros at 0.3 Hz and six poles at 30 Hz) (25788)

Also, I have made a DTT template in which the frequency-domain calibration (33161) is applied to all the relevant spectra. The template is saved at

/opt/rtcds/userapps/release/isc/h1/scripts/CrossCorrTemplate.xml

Two MEDM screens are newly made for this infrastructure:

• OMC_NULL_READOUT.adl displays the signal conditioning for the NULL signal in the omc frontend model, which can be accessed from the OMC tab in SITEMAP.
• CAL_CS_CUST_CROSSCORR.adl displays the mixing process of the SUM and NULL signals in the CALCS model. This can be accessed from the CAL tab in SITEMAP.

A screenshot of each MEDM screen is attached as well. They are saved in common medm directories at

/opt/rtcds/userapps/release/omc/common/medm/OMC_NULL_READOUT.adl

/opt/rtcds/userapps/release/cal/common/medm/CAL_CS_CUST_CROSSCORR.adl

[A quick verification: it seems OK]

To check whether the new infrastructure is doing the right thing or not, I exported the spectra from the DTT (the ones shown in the first attachment). I then computed the difference between the ordinary DARM spectrum and the cross correlated spectrum by subtracting one from the other. If things are correct, this leaves only sensing noise which should be dominated by shot noise at most of frequencies. The result is shown in the last attachment-- the difference seems reasonably smooth in its spectral shape as expected at most of frequencies.

The are a few points/regions where the difference deviated from shot noise. The peaks/valleys at 35 Hz, 60 Hz, 350 Hz and 500 Hz and low frequency wall below 30 Hz are reasonably suppressed by at least roughly 20 dB from the ordinary DARM which may be limited by lack of number of averages. On the other hand, the peak at 180 Hz was reduced only by 6 dB or so. I am not sure why. Otherwise, it looks reasonably good.