DSSD#1 undergoes a breakdown at 90V, two of the three wafers show this
- The adapter PCBs themselves have no breakdown at 100V, indicating the issue is internal to the snout
If we are lucky it may be a loose or misaligned kapton connector inside the snout. If not we will have to remove DSSD#2 and inspect DSSD#1 for damage/lint/etc. If we see nothing it should be replaced
DSSD#2 biases perfectly fine, but the leakage current is unstable with biasing -ve to p+n and gnd (return) to n+n. Leakage current is stable biasing +ve to n+n and gnd to p+n. (https://elog.ph.ed.ac.uk/DESPEC/240311_093933/Downstream_positive_bias_vs._Current_(uA).png very nice curve)
We see the same fluctuations just daisy chaining 3 p+n PCBs togerher with the bias (no DSSSD or gnd links). The fluctuations were reduced by connecting all 8 adapter boards of the DSSD together.
When biasing just *one* PCB in this way, the current is stable
The fluctuations happened changing the HV channel, adapter board PCB and LEMO cables: it doesn't seem to be a defect with a specific thing.
The test today confirmed the HV cables (the only ones used) were correctly isolated from everything (OL = "infinite" resistance to ground)
We saw unstable current indications using the Mesytec MHV-4 to apply voltage to the PCB as well.
The behaviour is odd but doesn't seem to be related to the DSSD, which I suspect is OK. One idea is to try connecting the 3 adapter boards to FEEs and repeating the test, as this introduces another (substantial) path to gnd. Maybe this eliminates the current instability? It's about the only difference left from December.
-
For the next steps I believe the snout must be dismounted to inspect DSSD#1. It would be best to coordinate this with replacing the broken(?) bPlas. If we are lucky we may not have to remove the DSSDs but it is likely we have to remove both DSSDs to swap out #1
After replacing the DSSD(s) we should cover the snout with a clean black bin bag and bias it on the MH table to confirm both detectors work. This saves the effort of carrying it to S4 just to find another problem.
If both detectors make it to 120V we can mount it again more confidently
Once we have two biased detectors we can rearrange some FEEs to get 8 for DSSD#1 to do the noise tests. I suggest we wire them up for the numbering plan in https://elog.ph.ed.ac.uk/DESPEC/532 but in the DHCP renumber 9,10,15,16 to 5,6,7,8 temporarily to allow data to be sent to MBS for the dry run (merger limitation)
When the remaining FEEs are recovered from UK+CRYRING we can instrument DSSD#2 and renumber the FEEs to match the cables |