Saturated PMTs have 0 PE

[bear-is-asleep]

I'm submitting a ...

• [na] bug report
• feature request

Description

When a PMT has a saturated waveform, the PMT stored in the flash has 0 PE. This is a huge issue, as you're unable to identify if the PMT has genuinely 0 PE or is saturated. At minimum, there should be a flag indicating when a PMT is at 0 PE. A heuristic is used to "guess" when PMTs are saturated in OpT0Finder based on the hypothesis.

See docdb 40918 for more info.

Possible Solution (optional)

Options are to -

1. Set the PE of the PMT to a non-zero (maybe negative) value when it's known to be saturated
2. Integrate over the waveform anyways, it's probably better than setting to 0
3. Apply a fit to the saturated component, and integrate this for the PE

How to reproduce

Typically, PMTs are saturated when a particle directly interacts with it (i.e. a cosmic passing through the vacuum chamber). So one could make a particle gun pointing directly at the PMTs to study this effect.

[mrmooney]

I agree this is a significant issue and we should address this ASAP. In the short term, I think the simplest thing to do is option 1 from Bear's list above - that is, set the PE of the PMT to some negative value that allows us to identify it as saturated downstream. It is important to first make sure that this wouldn't lead to any unintended consequences downstream. We should get input from PDS experts on this last point.

[bear-is-asleep]

From @fjnicolas -
the main difficulty is that waveforms are generally dropped after reco1 (both raw and deconvolved), so we loose our capability to check if there was a saturated one. I think the easiest thing would be defining a data product (something like sbn::OpDetSaturation) storing
Channel ID
TimeWindow in which the waveform saturated
That would be very light, so not footprint in the file sizes.
This is the part of the deconvolution code that checks wether it is saturated or not, so we could create instances of that “Saturation” data product that keeps stored in the downstream processing stages.
Then we have two options:
The OpFlash finder reads the sbn::OpDetSaturation objects, and assigns a dummy #PE value if there’s a sautrated channel within the OpFlash integration window
The sbn::OpDetSaturation object is directly read by the OpT0Finder in addition to the OpFlash, so you can have the saturation information as well