Reports 1-1 of 1 Clear search Modify search
takahiro.yamamoto - 2:04 Wednesday 11 May 2022 (20682) Print this report
ADC noise measurement on ETMY

[Ushiba, YamaT]

This is same work as on ITMX (klog#20667) and on ITMY (klog#20658).


- Although we measured ADC noise with combinations of four ADC boards and three AA chassis, no combination in which ADC noise was nominal level was found.
- Finally, ADC1 was replaced from S1402719 to S1402710.
- Strange noise still remains, but it's better combination than others for now.


From the experience on ITMX, and ITMY, glitches and strange structure on spectra don't come from only ADC or AA.
So we tried some combinations and compared spectra of ADC1 noise in each combination.
Used instruments are as follows.
ADC board:
- S1402719: Originally installed as ADC1 at ETMY
- S1402710: Originally installed at ITMY. Finally installed as ADC1 at ETMY.
- S2214047: Another ADC (old product lot; revB)
- S2113341: A new product lot (revC). Tested also on ITMY

AA chassis:
- S1706962: Originally and finally installed as AA1 at ETMY
- S2113203: Another AA (no modification)
- S2213719: Modified jumper pins (See also klog#19452)

Because we didn't replace instruments in ADC0 and ADC2 today, only one combination was measured for ADC0 and ADC2.
All results are put as /users/DGS/measurements/ADC/K1EY1/*.xml.

replacement of AA chassis
In the original situation, strange peak was able to be seen on many channels in addition to low-frequency excesses as shown in Fig.1 ~ Fig.4.
These behavior was not removed by replacing AA chassis. But frequency of peaks seemed to change by AA chassis replacement.
Fig.5 ~ Fig.8 and Fig.9 ~ Fig.12 shows the spectra with spare AA and with AA on which jumper pins were modified, respectively.
References in all figures represent a spectra in original situation.
We cannot see benefits by modifying jumper pins. But so many circuits are installed in EYV racks.
So we may need to take care about jumper pins on all circuits in order to see some benefit.

replacement of ADC board
By replacing ADC board, strange peaks around 10Hz vanished in all other combinations.
Fig.13 and Fig.14 shows the CH0-8 of ADC1 with another spare ADC and with ADC used in ITMY.
On the other hand, low-frequency excesses remained and channel number of contaminated channel depends on ADC boards we installed.
Unfortunately, there was no combination in which all 32 channels became healthy.

Finally, we installed ADC board which had been used in ITMY and the original AA chassis today.
The ADC noise spectra of current situation are show in Fig.15 ~ Fig.26.

knowledge, thoughts, etc.
Spectra in Fig.14 were measured with the ADC which had been used in ITMY.
Channel #2 had been assigned to ITMY_IP_ACC_H2 which was glitchy one (See also Fig.2 in klog#20636).
But it seemed to be healthy according to the measurements on the test bench reported as klog#20682.

Current test bench may not be enough for checking our problem
because the situation of power supply and ground connection was probably different so much from one in the mine.
When I measured ADC noise on the test bench, only one ADC board and one AA chassis were installed.
I plan to test ADCs and AAs which were brought back to Mozumi today with multi-ADC, and -AA installation.

Images attached to this report
Comments to this report:
takahiro.yamamoto - 22:04 Tuesday 17 May 2022 (20762) Print this report
[Ushiba, Tamaki, YamaT]


- We tried to install a dedicated power supply unit for AA chassis in EYV1 rack because it worked well on ITMX reported as klog#20715.
- CH1 of ADC1 which was the most noisy one became slightly quiet. But it didn't worked well such as the case on ITMX.
- ETMY situation seems to be something different from ITMX.
- Finally, we uninstalled the dedicated power supply unit and reverted all situations.


As I did on ITMX reported as klog#20715, we tried to install a dedicated power supply unit for AA chassis.
Because of the mount space, the power supply unit was installed in the EYV2 rack and was connected to AAs in the EYV1 rack with 10m cables.

As shown in Fig.1, CH1 of ADC which was the most loud channel got slightly quiet (Cyan -> Blue). But some channels e.g. CH0 of ADC0 got louder than the original situation (Magenta -> Red).

In the case on ITMX, all channels became quiet (~1e-2ct/rHz level).
So it seems to be different between ETMY and ITMX situations...

We also found that all screws for mounting AA chassis connected to the dedicated power supply unit must be loosened to see the effect of the GND separation for AA chassis.
If some screws remain, GND of the dedicated power supply unit and the regular power supply unit are connected.
(And also, GND of AC seems to be connected to one of DC18V unfortunately...)

Finally, we reverted the power supply situation as the original one because there is no effect with tightened screws and it's difficult to leave such a situation.
Images attached to this comment
Search Help