I changed the setpoint for the F1 GAS from -3180 to -2810 so that the MN V Oplev can be zero. The F1 GAS was offloaded with the FR.

[Takano, Ushiba]

Abstract

We characterise the new IMC CMS. First, we measured the transfer function of the new IMC common mode servo from the input to the FAST and SLOW outputs, and the results were as expected. Next, we measured the output for DAQ monitor channels to investigate the source of the large offsets, but nothing strange was found. Finally, we checked the offset level of the input gain stage with different gain settings and concluded that we have to adjust the offset of some gain stages.

Detail

Transfer function measurement in the FAST and SLOW path

To avoid the saturation of the signal for the IMC control during lock acquisition, we designed a new topology for the IMC common mode servo, and a spare one was modified (klog 35396). The transfer function was once measured, but today we measured it again using a Moku:Lab (measurement) and the AEL standalone system (CMS control). The measurement and fitting results are shown in Fig.1. The fitted parameters are as follows:

• High-pass cutoff: 2.32(1) kHz
• Low-pass cutoff: 36.8(1) kHz
• Overall gain: 9.53(5)
• Time delay: 312.2(1) ns

These parameters are slightly different from the expected values but are quite similar (2.26 kHz for the high-pass cutoff, 36.2 kHz for the low-pass cutoff, 10 for the overall gain).

Additionally, it was pointed out from the onsite measurement that the low-pass cutoff in the SLOW path was ~ 10 kHz, not 30 kHz as designed (see p.9 of this document). To make sure that the servo design is correct, the transfer function from the input to the SLOW output was measured. The measurement and the fitted result are shown in Fig.2. The fitted parameters are:

• Low-pass cutoff: 30.6(1) kHz
• Overall gain: 0.954(2)
• Time delay: 262.8(3) ns

Therefore, the cutoff frequency is surely ~ 30 kHz, as designed, and probably something went wrong during the past transfer function measurement.

Investigation of the offset source in DAQ monitor channels

It is known that the FAST_DAQ_OUT and SLOW_DAQ_OUT channels have a large offset (~ O(1 V)), but MIXER_DAQ_OUT and other slow monitor channels (such as SLOW_MON_OUTPUT and FAST_MON_OUTPUT) do not. We suspected that the opamps used for these DAQ channels are oscillating in RF frequencies, so we measured the outputs for these channels using the Moku:Lab. The results are shown in Fig.3. From the power spectra, these signals are unlikely to oscillate at any specific frequency. On the other hand, we confirmed that the offset voltages measured by the Moku:Lab were almost identical to those observed from the CDS, indicating that these offsets indeed exist within the CMS. Tomorrow, we will open the chassis and investigate the voltage on the circuit directly.

Offset voltage of the gain stages

When the offset voltage of each gain stage changes, a large glitch occurs when the gain setting is changed, and it is a potential source of the lock loss. To determine at which stages we need to adjust the offset, we investigated the offset voltages at the MIX_DAQ_OUT channel by varying the gain setting. During the measurement, each side of the input port was terminated with 50 Ω to mimic the output from the I/Q demodulator, as shown in the photo.

The results are summarised in the table below:

From these results, it appears that we don't need to adjust the offset on the input receivers, but we have to adjust some of the gain stages. Our plan is to tune the offset of 4dB, 8dB, and 16dB stages. 

There is another variable gain stage in the FAST path. We checked the peak-to-peak voltage of the glitches when the gain setting was changed and concluded that the offset should be tuned for 2dB, 4dB, 8dB, and 16dB stages.

I adjusted the NBDAMP filters for L1, L2, L3, T1, T2, T3, and P1.

In the last measurement, the systematic error of laser PZT efficiency between forward and backward was quite large 1.71/1.36 -1  ~ 26%.

I measured the laser PZT efficiency again using the same way as klog:35884, but with lower laser power (7.7A to 3A ->  28W to 2.7W). The measurement is very hard since mode matching to the PMC with lower power will be worse, and the amplitude of the higher-order (HOM) mode will be relatively larger than the TEM00 mode of the carrier and sidebands.

We know the relative position of the sideband to the carrier from the measurement of the FSR of PMC, so I put cursors on both sidebands in Fig.1. You can see the one (left) sideband, but the other (right) sideband is almost buried in the HOM. However, for the measurement, it is ok if we know the position of one of the sidebands at least.

Sweeping the laser frequency by laser PZT with 0.5Hz +/- 10V amplitude, we can see the series of peaks of the carrier and one of the sidebands (Fig.2). 

Separation in each pair can be measured using a 15MHz modulation frequency and applied voltage to the PZT (Fig.3).

The results for forward and backward with 10 times average are;

• Forward (10 times average): 1.304+/-0.00081 MHz/V
• Backward (10 times average): 1.282+/-0.0052 MHz/V

Still, some systematic error exists, and it is over the statistical error, but the difference is much smaller: 1.304/1.282 -1  ~ 1.7%.

Averaged efficiency of 2ways is 1.293+/-0.028 MHz/V. Error mainly comes from systematic error of 2ways (average of statistical errors + difference).

Higher resolution from 100kHz to 1MHz.

We adjusted T1-T3 NBDAMP filters.

• T1: Changed from FM3(pink) to FM1(cyan), and adjusted FM1(red)
• T2: Changed from FM3(brown) to FM1(red)
• T3: Copied FM3(cyan) to FM5 and modified FM5(5.336, 271K) for the adjustment.

I changed the setpoint for the F2 GAS from 1160 to 1280 so that the MN V Oplev can be zero. The F2 GAS was offloaded with the FR.

I changed the setpoint for the F2 GAS from -1258 to -1188 so that the MN V Oplev can be zero. The F2 GAS was offloaded with the FR.

I offloaded the BF GAS with the FR.

That was my mistake. I had set lower current on neoLASE amplifre. It is back to28.6W now.

We adjuseted NBDAMP filters (L1,2,3, and 5)

• L1: changed the filter from FM3(pink) to FM1(red)
• L2: changed the filter from FM3(brown) to FM1(red)
• L3: checked FM3(original one) and FM1(300K one). As FM3(brown) looks better than FM1(red), we keep using FM3.
• (L4: not be closed at LOCK_AQUISITION)
• L5: Measured TF had too large gain peaking. So we comment out the line which closes this loop in the guardian code.

After this work, we checked the ETMX suspension can reach the LOCK_AQUISISION state and back to ISOLATED state. (The L5 loop was not closed as we expected.)
The current temperature of ETMX is 270K.

K1:VIS-ITMY_BF_DAMP_GAS_OUT16 is showing a very hight value from 2 days ago.

[Kimura]

 At approximately 19:38 on January 11, continuous residual gas measurements using five Q-mass instruments installed in BS, IXC, IYC, EXC, and EYC were stopped.

I turned off the heater at the top of the inner ratioation shield.

I turned off the heater at the top of the inner ratioation shield.

I turned off the heater at the top of the inner ratioation shield.

I turned off the heater at the top of the inner ratioation shield.

I turned off IM heater.