Wrong initialisation values for gain selection and charge integration for data

[FrancaCassol]

Hi,
looking at the dl0_dl1.py script, I see two issues concerning the initialisation values for data:

1- The threshold of the gain selection for data is hardcoded 4095, I am afraid that this means the LG is never selected (the thereshold value is checked with a "grater" not "equal or grater")

 if not is_simu:
                    combine_channels(event, telescope_id, 4095)

2- The charge integrator in the config file is NeighborPeakWindowSum, but the configuration in the code is hardcoded for the LocalPeakWindowSum

charge_config = Config({"LocalPeakWindowSum":{"window_shift": 5,"window_width":12}})

so this initialisation will never be used (by the way this window is good for flat field data, most probably too large for shower data)

I am in discussion with @rlopezcoto for integrating the new calibrator in the script (see #248). We will add there a solution to these problems, but we need to start also a reflection on how to improve the way we handle the initialisation values in the reconstruction scripts.

[vuillaut]

Hi @FrancaCassol

I think you are right on both points.

Indeed for the gain selection, the value should be 4094 since it's only >.

[contrera]

Just a comment. Presently some times (a few) ADC values which we read from R0 are > 4096 because of the artificial baseline that is added in the DRS4 correction. Is that an issue? I do not know enough about the calibration to judge.

[moralejo]

I think (Julien/Pawel/whoever please correct me if I am wrong) that the values corrected at the EVB get an added shift of, I think, 400 counts. This is after correcting each baseline to 0 with their average value. So some pixels will indeed go above 4095, others will never arrive at 4094. I suggest that as a simple solution we put the cut at a safe value of 3500, which should mean all pixels get to use the low gain sometimes, and it is a high enough charge so that the low gain has a clear signal.

…

On Thu, 19 Dec 2019 at 09:33, contrera ***@***.***> wrote: Just a comment. Presently some times (a few) ADC values which we read from R0 are > 4096 because of the artificial baseline that is added in the DRS4 correction. Is that an issue? I do not know enough about the calibration to judge. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#252?email_source=notifications&email_token=ADMCCOXZPK6Y2YXTYUDPVUTQZMWUNA5CNFSM4J32J3CKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHI3ABA#issuecomment-567390212>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADMCCORTK7VM2QNXGULZO7LQZMWUNANCNFSM4J32J3CA> .

-- Abelardo Moralejo Olaizola Institut de Física d'Altes Energies Tel : +34 931641662 Fax: +34 935811938 Avís - Aviso - Legal Notice - (LOPD) - http://legal.ifae.es

-- Avís - Aviso - Legal Notice - (LOPD) - http://legal.ifae.es <http://legal.ifae.es/>

[FrancaCassol]

Hi @moralejo, @contrera ,

actually this is perhaps a good reason for using the R1 waveform as reference for the threshold (R1 is already baseline subtracted ) because otherwise we should put as threshold 3500 for the data and (3500-R0 offset) for the MC in order to set a comparable threshold value.

So, I would suggest to use R1 (before calibration) and to put as default 3500, what do you think?

[moralejo]

Hi @FrancaCassol, @contrera,

So, I would suggest to use R1 (before calibration) and to put as default 3500, what do you think?

the problem is that in data with pedestal baseline corrected by EVB we do not have those raw numbers. So in a way it cannot be properly done on data produced by EVB with "no gain selection" and "baseline correction".

On the other hand, I think it is currently not at all critical that the gain switch is not made exactly at the same point in data and MC.

[FrancaCassol]

Hi @moralejo

the problem is that in data with pedestal baseline corrected by EVB we do not have those raw numbers. So in a way it cannot be properly done on data produced by EVB with "no gain selection" and "baseline correction".

I am not sure to understand, in the EvB Julien can perform the gain selection on the baseline corrected r0 before adding the offset (isn't it?), and in lstchain the R1 corresponds to the baseline corrected R0 with offset subtraction, I part some other higher level correction the two waveforms are the same

On the other hand, I think it is currently not at all critical that the gain switch is not made exactly at the same point in data and MC.

Probably, but making things closer is always better ;-)

[moralejo]

the problem is that in data with pedestal baseline corrected by EVB we do not have those raw numbers. So in a way it cannot be properly done on data produced by EVB with "no gain selection" and "baseline correction".

I am not sure to understand, in the EvB Julien can perform the gain selection on the baseline corrected r0 before adding the offset (isn't it?),

Should not be the case. EVB has access to the raw values, so the gain selection should be done on the raw HG value. One might argue in favour of having the switch, for all channels, at the same calibrated charge, but that cannot be made at EVB. And in any case I do not think it is important at all, as long as HG and LG are well cross-calibrated.

To be clear, when I suggested 3500 I was referring to the values for which the baseline is at 400.

[rlopezcoto]

Solved in #248