Abstract:
I measured spectra of each sensor and found that H2 geophone has 10 times larger noise than the others.
It is necessary to investigate the reason.
Detail:
I measured spectra of each geophone on IP when suspension was ISOLATED state.
Figure 1 shows the result.
As you can see, H2 geophone has larger noise below 1 Hz, especially below 0.1 Hz the noise is 10 times larger.
It is necessary to investigate the reason.
[Takahashi, Sato]
We investigated the H2 geophone. We tried the following cases.
In any case, the noise level of the signal in the H2 channel on the DGS was larger than the signals in other channels. This means the noise source is in the ADC (or AAF).
I also checked the noise situation of Geophone channels of ITMY.
AA seemed to be a cause of this issue but there was no improvement by replacing AA chassis.
This situation is similar to ITMX GAS glitches.
We may need to try replacing ADC or internal cables in IO chassis.
Takahashi-san had already checked the noise behavior with unplugging the signal cable at the output port of the Geophone distributor.
I checked the noise behavior with unplugging the signal cable at the input port of AA chassis, output port of AA chassis and input port of ADC adapter.
Fig.1 and Fig.2 are the time series signals and the spectra of Geophone channel with unplugging at the input port of AA chassis.
In this case, we can see the strange jump and glitches in the time series signal on H2 and the H2 spectrum was something loud same as Takahashi-san's report (See also klog#20323).
I decided that the Dsub cable between Geophone distributor and AA chassis is not a cause of this noise.
As the next step, I checked the situation with unplugging a SCSI cable at the output port of AA chassis.
Then, strange jump and glitches disappeared and H2 spectrum was expected ADC noise level same as H1, H3 and H4 as shown in Fig.3 and Fig.4
And also noise behavior with unplugging SCSI cable at the input of ADC adapter was same.
This noise seemed to be induced by connecting AA chassis to ADC or not.
So I tried to replace the AA chassis as new one. But glitches didn't disappear.
Just in case, I also tried to replace the SCSI cable and confirmed no change in the situation.
This situation is similar to the F3-GAS glitches of ITMX as reported in klog#19212.
Finally, I restored AA and scsi cable as original ones.
We may need to replace the inside of the IO chassis such as ADC, internal cables etc...
[Ushiba, YamaT]
- Similar issue on ITMX F3_GAS will be reported as klog#20639
- Glitches on ACC H2 were removed by replacing the ADC board, the ADC adapter board, and the ADC internal cable.
- Glitch problem was solved.
- On the other hand, ADC timing error is occurred with new ADC.
- It shouldn't affect ITMY local measurement.
- Timing error probably affects measurements with channels on other real-time computer.
- This will be discussed in klog#20642.
According to past investigations reported as the thread of klog#20197,
cause of glitches lies on the downstream of AA chassis. But there was no improvement by replacing AA chassis and the SCSI cable between AA chassis and ADC.
Today, we tried to replace stuffs in the IO chassis.
By replacing ADC (S1402710 => S2113341), signal behavior was drastically changed.
Fig.1 shows the time series signals of ADC noise on Geophone channels (unplugged Dsub cable between LVDT Driver chassis and AA chassis).
before replacing (left panels) and after replacing (right panels).
Vertical axes range is unified as 5-count for all plots. We can see the H2 signal was drastically changed.
Other channels (H1, H3, and H4) were also slightly improved. This seems to come from decreasing the amplitude of 60Hz power line.
Also for spectra, H2 signal became same level of other channels as shown in Fig.2,
though H2 signal had been 100 times louder than other signals below 100Hz before replacement.
After connecting Dsub cable from LVDT Driver chassis, we can also see the improvement of noise level below ~50mHz (See also Fig.3).
Note: due to the Earthquake or other reason, suspension resonances were excited in the plot of "after replacing ADC".
As for the glitch issue, we would say it has been solved.
