Reports 1-1 of 1 Clear search Modify search
takafumi.ushiba - 20:27 Saturday 04 September 2021 (18153) Print this report
Implementation of GAS DC control for ITMX

[YamaT, Ushiba]

We measured the PLANT of GAS filters of ITMX from Miyo-kun's code.
Then, we implemented GAS damping at F0 and GAS DC controls at all stages.
All can be engaged from guardian now.

First, we checked the signs of COILOUTF and found that F0 COILOUTF has opposite sign from the others (see fig 1, 2,3, 4, and 5).
So, we changed the sign of F0 COILOUTF from -1 to 1.
Now, all calibrated GAS LVDT signals (K1:VIS-ITMX_{F0,F1,F2,F3,BF}_LVDTINF_GAS_OUT) increase when applying positive signals at inputs of COILOUTF(K1:VIS-ITMX_{F0,F1,F2,F3,BF}_COILOUTF_GAS_IN1).

To design the DC control filters of GAS, GAS PLANTs were measured by Miyo-kun's code.
Measurement results are stored in /kagra/Dropbox/Measurements/VIS/PLANT/.
File names are PLANT_STANDBY_ITMX_GAS_TEST_{F0,F1,F2,F3,BF}_1189.xml

After the measurement, we designed the filters and implemented at GAS_DAMP filter banks.
For F0, the filter was designed as 0.01Hz UGF DC control and damping (see fig6).
For the other stages (F1, F2, F3, and BF), the filters were designed as 0.01Hz UGF DC control.

After checking the performance of the control, Yamamoto-kun modified the guardian and now all controls can be engaged from the guardian (ENGAGE_GAS).

*We forgot to change the sign of F COILOUTF gain, the phase of measured PLANT of F0 GAS is 180 degrees at DC.
*BF GAS is very stiff compared with the other GAS filters (DC gain is -70 dB for BF GAS while the others are about -50 dB or more, see fig7).
*F2 has a strange behaviour when sweep the F2 GAS position (see fig8). Did it mechanically hit somewhere when LVDT signal of F2 is at 1270 counts?

Images attached to this report
Comments to this report:
takafumi.ushiba - 9:03 Monday 06 September 2021 (18156) Print this report

I checked error and feedback signals of GAS controls during the weekend.
They seems working well for over 1 day.

About - 6 hours, the glitch can be seen in F2 error signals.
I'm not sure but there seems to be an earthquake around that time.
I would like to someone to check the details.

Images attached to this comment
takahiro.yamamoto - 9:38 Monday 06 September 2021 (18159) Print this report
I attached a spectra comparison plot between STANDBY (on 9/3) and ENGAGE_GAS (on 9/4) states.

- Peak at around 0.67Hz is damped well.
- Increasing noise level around 2-3Hz can be seen seen on F0 signal. It may be caused by small phase margin.
- F2 spectrum has large difference. But it may be cause by changing DC position on 9/4.
- Noise excess can be seen on IP signals. Hmm...
I haven't checked the difference in seismic motion situation between 9/3 and 9/4 yet.
So we need to investigate more detailed.
Images attached to this comment
takafumi.ushiba - 11:14 Monday 06 September 2021 (18160) Print this report

I checked the reproducivility of strange behaviour whem moving F2 GAS.
Attached figures show time series data of calibrated LVDT signals and output signal to F2.

Figure 1 shows the data when output to F2 was moving from -20,000 to +30,000 counts (F2 position was +200um to -850um).
Figure 2 shows the data when output to F2 was moving from +30,000 to -20,000 counts (F2 position was -850um to +100um).
As can be seen, the strange behaviour was reproduced.
Also, there is a hysteresis of 100um.

I tentatively put the set point of F2 GAS as -400um, which is between strange behaviour positions.

Images attached to this comment
Search Help