[Oshino, YamaT]
 

Abstract

- This work is related to klog#20636 and klog#20642.
- ADC0 in K1IY1 was replaced from S2113341 to S2214044.
- Timing error was solved by replacing ADC board as a old product lot. But glitches on ACC H2 were also came back.
- Finally glitches was able to removed by using another ADC board which was also a old product lot.
- Small glitches exists on SEG1 of TM LEN and TM TILT Oplev which are assigned on ADC1.
 

Details

After replacing ADC board which is a new product lot, k1iopiy1 had been had a timing error in exchange for removing glitches (See also S1402710) back as ADC0.
Then the timing synchronization came back with the glitches on ITMY_ACCINF_H2. Apparently the glitch had not been caused by a bad contact.

Next, we replaced another ADC board (S2214044) which is a old product lot and then glitches vanished.
Fig.1 shows the ADC noise on ACC channels with S1402710 (bottom right) and S2214044 (top right).
We need to test the timing synchronization situation with a new product lot and CH1 glitches on S1402710 in the test bench.

We also measured ADC noise in all channels at this time.
All results are shown in Fig.2 ~ Fig.13 and stored as /users/DGS/measurements/ADC/K1IY1/IY1_ADC?_CH??-CH??.xml
These spectra were measured in SAFE state and all Dsub port on AA chassis were opened.
(It may be better to re-measure with terminating Dsub port.)

Though detailed inspection one-by-one haven't been done yet, there is no channel on which noise level is a few tens times worse than others.
On the other hand, there are some channels on which noise level is a few times worse e.g CH16 and CH20 on ADC1 shown in Fig.8.

Noise on these channels became quiet by unplugging the SCSI cable between ADC and AA chassis.
(REF in Fig.14 shows the spectra with plugging the SCSI cable and others show ones with unplugging.)
It may be a problem on the electrical connection in EY1 rack not a problem only on the ADC board itself.

So we tried to measure the noise level with turning off all circuits except AA and AI. But situation didn't change.
Because we didn't remove all cable and circuits were still mounted on 19-inch rack, electrical connection seemed to still exist.
(I want to try measuring spectra with removing all cables and circuits from 19-inch rack, but it's too tough work...)

By the way, CH16 and CH20 on ADC1 are used as SEG1 of length and tilt OpLev, respectively.
We cannot see any glitches in raw ADC noise on these channels. But we can see something in lowpass-ed ADC noise.
Fig.15 and Fig.16 shows the time series signal on OpLev channels with unplugging the Dsub cable at the input of AA chassis.
IN1 is raw (but decimated from 65kHz to 16kHz) ADC noise signal and it's difficult to notice that some noise exists on these channel.
But on OUT channel which is after de-whitened by 'z10p1' (now it works as lowpass filter because whitening filter chassis was unconnected at this time.)
some noise which amplitude is less than 1 count exists.
If these noise become a cause of OpLev glitches, we may have to replace also ADC1.