[Dan, Koji]
Last week, the OMC cavity was scanned with the fully locked IFO at 5.22W. In addition to the usual cavity scan analysis (let's say "DC scan"),
the technique called "beacon scan" was tried in order to check the signal content in each carrier higher-order mode.
Here are some preliimnary results.
Motivation
- Dan took OMC scans in a full lock with and without the DARM offset a while ago (LHO ALOG 16747).
- When the DARM offset is zero, basically there is no (or almost no) carrier TEM00.
- No matter how the offset is, there are carrier higher-order modes.
- This means that we are actually not sure how much of the carrier higher-order mode is coming from the arm.
- Aren't we under-estimating the signal mode matching because of the carrier that has no signal?
- We want to characterize the signal mode matching.
- We can use the DARM signal to distinguish the signal carrier and the others
Measurement conditions
- The DRFPMI was locked at 5.22W MC input.
- DARM was locked with RF with an offset. (~10mA for TEM00)
- The OMC HV PZT (PZT2) was swept from 30V to 90V with the ramp speed of 0.6V/sec (i.e. 100s per ramp).
- During the scan, the whitening filters of the DCPDs were turned off in order to prevent from their saturation.
Signal processing (Attachment 1)
- DCPD_SUM and PZT2 Readback were sampled at 16k. All the signal processing have been done offline with MATLAB.
- DCPD_SUM was preconditioned for demodulation by a bandpass with the passband of fmod +/- 20Hz, where fmod is the demodulation frequency.
- The preconditioned signal and sin or cos signals at fmod were multipled.
- The DC and demodulated signals were decimated to 512Hz (the decimation ratio=32).
- The decimation filter is the MATLAB default one. (i.e. fc=0.8*fsample, 0.05dB ripple 8th Cheby LPF)
- The decimated demodulation signals were further LPFed for noise reduction. (f_LPF = 5Hz)
- The RMS of the demodulated signals are calculated to obtain the beacon amplitude.
- The noise reduction filter imposes the phase delay in the pass band. This was approximated as a time delay.
i.e. Negative time delay (or sample shift) of 60 samples were applied. (Dan suggested me to use "filtfilt" function
in order to suppress this kind of phase delay. I'll use it in the future scan.)
Result (Attachment 2)
- The demodulation frequency of 1009.4Hz was used for the beacon scan.
There was a big peak in DARM spectrum due to undamped ITM violin mode.
- As seen in the figure, a big response found at the carrier TEM00 mode. (Success!)
(In fact, the response was too weak at the demod frequency of the calibration lines. That's the reason why the violin peak is used.)
- Not all the carrier amplitudes in the DC scan were explained by the beacon scan result.
This indicates that each higher order modes consists of a signal carrier from the arm and a non-signal carrier.
Note that the peaks in the DC scan and beacon scan are matched up by multiplying the factor of 5800 to the beacon result.
- As the background data, the demod frequency of 700Hz was used. There was no visible peak in the DARM spectrum at 700Hz.
Most of the carrier peaks are above the background level. So this is not an artifact of the noise power in a peak.
Discussion
- The table below is showing the peak heights that were read from the plot. (Note this is not a peak fit result.)
The ratios were estimated from the sum up to 8th order.
Mode DC_scan Beacon_scan
# [mA] [ratio] [ratio]
---------------------------------------
0 9.18 0.589 0.928
1 0.539 0.035 0.010
2 0.675 0.043 0.020
3 1.327 0.085 0.024
4 0.564 0.036 0.004
5 0.821 0.053 0.005
6 1.355 0.087 0.006
7 0.568 0.036 0.001
8 0.559 0.036 0.001
---------------------------------------
- The table indicates that the signal mode matching was 92.8% and was way better than the DC matching.
- The 3rd and 6th order modes (both in DC and beacon) are larger than the others. Is there any reason?
How can we improve?
For better result:
- Slower scan: The cut off of the noise reduction filter is determined by the scan speed. Slower scan will improve the SNR of the measurement.
- Beacon at ETMs: The current beacon isone of the ITM violin mode. I suspect that is the reason why some of the sidebands shows some signal
above the background level. We should use pure DARM signal.
- Better alignment: The alignment of the beam to the OMC was not perfect. We can definitely improve it.
For better understanding of the result:
- Mode healing/harming: What does the SRC do for the peak heights? Are any particular modes enhanced/suppressed by the SRC?
