40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  PSL  Not logged in ELOG logo
Entry  Mon Jan 31 20:24:45 2011, frank, tara, Notes, BEAT, beat signal down to 12.5 mHz beat_2011_01_30_lowf.png
    Reply  Tue Feb 1 22:22:46 2011, frank, tara, Notes, BEAT, beat signal down to 12.5 mHz PLL_FC_good.png2011_02_01_log3.pngac_highpass.png2011_02_01_dat.matcode_2011_02_01.m
Message ID: 463     Entry time: Tue Feb 1 22:22:46 2011     In reply to: 459
Author: frank, tara 
Type: Notes 
Category: BEAT 
Subject: beat signal down to 12.5 mHz 

We measured beat noise from the frequency counter(FC) instead of the feedback from PLL. The result is plotted below

 

We switch to use the FC because we hope it will allow us to measure the signal down to lower frequency (~10 mHz) which is not quite possible

for PLL because of its small input frequency range for the acceptable phase noise level.

 

The FC can measure the relative frequency from the chosen center frequency and give a voltage output (from 0 to 8 V.)

For example if we set the center frequency to be 160 MHz, with 1MHz/V gain, signals at 156 MHz and 164 MHz will correspond to 0 V and 8 V respectively.

Since the LIGO VCO range is ~ 10 MHz, we tried measuring the signal with 500 kHz/V setup which is equal to 4 MHz range. However the noise is too high,

so we have to choose a new gain setup, and  20kHz/V setup is still acceptable for our signal compared to the signal from PLL.

 (It turns out that data from 50kHz/V is too noisy)

Attachment 1:

Blue and Green show the beat noise measurements from PLL and FC. They agree well. The gain for FC is set to 2kHz/V.

When the temperature became more stable, the data could be acquired to down to ~ 3mHz (red).

 

We used AC coupling for FFT measurement, so the TF for AC coupling is measured. There is a 160 mHZ high pass, and the data is corrected accordingly.

Attachment2:

Time series for ACAV's temperature, RCAV's temperature, VCO mon, and Frequency count Vout during the FFT measurement.

 

Attachment3: show TF measurement for AC highpass and fit with 160mHz high pass.

The plot is flipped because chA is AC couple and ChB is DC couple during the measurement, and

The output is B/A.

Quote:

This morning I was able to measure the beat noise down to 12.5 mHz. So I plot it together with the noise budget here.

We also added channel for frequency counter,C3:PSL-FSS_FREQCOUNT which will allows longer data acquisition time for lower frequency.

Then we will be able to see the temperature effect at lower frequency(~ 10 mHz.)

 

Fig1: beat signal at 12.5 mHz to 10 Hz

 

We are also working on PID thermal control for refcav.

Perl script for PID thermal control won't work on Solaris because it doesnt have ezcaread/write command, we will get that from op40m machine at the 40m.

(We can't run Perl script on Linux because it complains that it fails to read / write data from C3:PSL-FSS_HEATER.)

 

 

Attachment 1: PLL_FC_good.png  51 kB  Uploaded Tue Feb 1 23:42:39 2011  | Hide | Hide all
PLL_FC_good.png
Attachment 2: 2011_02_01_log3.png  42 kB  Uploaded Wed Feb 2 00:12:06 2011  | Hide | Hide all
2011_02_01_log3.png
Attachment 3: ac_highpass.png  31 kB  Uploaded Wed Feb 2 00:21:18 2011  | Hide | Hide all
ac_highpass.png
Attachment 4: 2011_02_01_dat.mat  68 kB  Uploaded Wed Feb 2 00:24:26 2011
Attachment 5: code_2011_02_01.m  470 Bytes  Uploaded Wed Feb 2 00:24:36 2011  | Hide | Hide all
load 2011_02_01_dat.mat

fc=fc_20kHz_22db_gd;
%correct for 160 mHz high pass
fc(:,3)=fc(:,2).* sqrt( (1 + (0.16./fc(:,1)).^2 ));

loglog(fc_2kHzv(:,1),fc_2kHzv(:,2)*2e3,...
        beat5Hz(:,1),beat5Hz(:,2)*71e3,...
        t2kHzv(:,1),t2kHzv(:,2)*2e3,...
... 5 more lines ...
ELOG V3.1.3-