Barrel 6 Routine Assembly Tests
General - 16 bit histograms would speed up r/o time for most of these. (Exceptions: NoiseOccupancyTest and DoubleTriggerNoise test, best done with higher statistics.)
Select 0
- RX threshold Scan
- 10 -> 250, step 10
- NMask
- 32 ->128, step 32 - (in cvs untested - Alan)
- StrobeDelayTest
- unchanged
- ThreePointGainTest
- including slope analysis, otherwise unchanged
- NoiseOccupancyTest
- including OPE analysis, otherwise unchanged
- what about doing this with synch triggers? Alan
- DoubleTriggerNoise
- hunt for light leaks
- minimum number of bins, range tbc
- can we calculate baseline from NO test?
Select 1
not routinely- need at least one threshold scan to look for slope variations
- suggest 2fC threshold scan, same conditions as 3PG
- need to change select using DDC - but already works
- would be most efficient if redundant mappings can be selected online, otherwise must set up and run StrobeDelayTest first. (????)
- Alan dosent understand :(
Alan notes that some charactersitics, such as
- synchronous triggers
- select line
- number of triggers ... these are really options that we might want to add to most/many tests, so perhaps these should be done in the general way, rather than as a new test-name. The problem is we currently dont have a good way of recording what options were set for that scan.
- note to self - if one gets a scanrequest, then one has access to the actual scan object the api will run. However the same is not true for the TestRequest?, since these generate their own ScanRequest? s on getNextScan :(
New Diagnostic Tests
- DiagnosticTests? (eg binary search of chips and channels)
- Needs feedback between scan to do as expected, but could have set of 7 scans with channels on if they have bit n set in their address, then 4 scans for chips as well
- Wouldn't provide full diagnostics on failures which are dependent on neighbours
- If the configuration setup between Scans was fast enough it would be possible to create a series of scans of one bin with whatever setup you want
- Having this as a goal changes the configuration strategy again, I think it would require the copy primitive, and setting the mask etc on the ROD, which is not possible at the moment either BJG
- Needs feedback between scan to do as expected, but could have set of 7 scans with channels on if they have bit n set in their address, then 4 scans for chips as well
- Binary search of chips can be one scan, binary search of channels can be another.
- Start by binary search of chips, all channels enabled, point of failure indicates suspect chip (or lack thereof). This is basically a variant on the RedundancyTest?, but with chips set to return analogue data (eg 4fC Qinj at 2fC threshold).
- Next run binary search of channels for the "bad" chip configuration. Point of failure indicates suspect channel (or lack thereof). This is basically a variant on the NMaskTest, but with chips set to return analogue data (eg 4fC Qinj at 2fC threshold). PWP