Reports 1-1 of 1 Clear search Modify search
AdV-ISC (Sensing and control implementation)
bersanetti, heitmann, ruggi, masserot - 23:57 Monday 15 May 2017 (37611) Print this report
SSFS shift

Today's shift on SSFS was mainly split into two activities:

  • Alain did some tests on the SSFS at 1 MHz, but it wasn't completely working as expected, we leave the details on these tests for a separate entry;
  • then, with the SSFS working at 500 kHz as usual, we moved the roll-off of the boost filter from 20 kHz to 50 kHz, with an increased Q of 1.5:
    • this should have made us gain a gain margin of about ~1.5; we could see it as we could increase the SSFS gain from the standard 900 up to 2000, without getting into any fatal instability, even if some peaking is visible in the high frequency region (see Figure 1); however, we could not benefit from this in any way, as the increased gain made no difference for the error signal suppression;
    • looking at the SSFS_B4_Corr_DS signal, we could see some structures appearing at ~ 20kHz, but this did not spoil or saturate the correction in any way (see Figure 2);
    • with a higher roll-off we decided to try and move the boost filter's zeros to higher frequencies, from 500 Hz and 1000 Hz to 750 Hz and 1500 Hz; however, once we modified the boost filter on the MC DSP card the interferometer unlocked repeatedly already when moving CARM to the MC, which is something that should not have been affected at all by modifying the boost filter; we thought about some non-saved changes on the MC card which were discarded once we Compiled & Downloaded the DSP, but moving back the boost filter to the old configuration made also the misbehaviour disappear, so something else (which is not clear to us) must have happened;

In order to prepare some test with both the B4 signals (the SPRB and the SSFS one) we restored the phase tuning for both SPRB and SSFS, and saved the correct demodulation phases in Metatron's ini files, although the lock acquisition still uses only the SSFS signals; the correct phase for SPRB's B4 was off by only -0.12 rad and, as usual, the phase offset for SSFS was adjusted twice as much, in the opposite direction. Something that we noticed during this last test, and that we do not understand yet, is that while the Q signals ("MICH") are basically the same in SPRB and SSFS (aside from a dephasing at high frequency), the I signals (frequency noise) are quite different up to 100 Hz, and the MICH line is clearly visible in the transfer function (see Figure 3). This is something we need to investigate.

 

We leave the ITF in LOW_NOISE_1 for the night.

Images attached to this report
Comments to this report:
bersanetti - 20:02 Tuesday 16 May 2017 (37624) Print this report

Looking at a single lock (in this example, from today at ~12:35 UTC), it seems that the difference between the two B4 signals is not a constant feature; discarding everything before dark fringe (as only the I quadrature is controlled with SSFS while the Q is not used for MICH yet), one can see that there is no difference between the signals in dark fringe with the SSFS with the standard filter (Figure 1), but as soon as we engage the boost filter the structures on the I signal appear (Figure 2).

Images attached to this comment
mwas - 14:43 Monday 22 May 2017 (37677) Print this report
I am not quite sure but the lack of improvement when increased the gain by a factor 2 is probably due to the misleading way the gain is applied in SSFS_Ctrl. The gain is not applied to the control filter (as is usually the case) but to the error signal. So the calibration of the error signal changes by the same number as the gain of the filter.

B4_Err_post = B4 demodulated at 56 MHz / B4 DC power * SSFS_B4_GAIN

On the figure you see that B4_Err_I_unnorm (which should be with the SSFS_B4_GAIN/B4 DC power factor removed) does indeed go down with the higher SSFS gain. And this is also the case for the LSC_B4_56MHz_I channels.
Images attached to this comment
Search Help
×

Warning

×