Project

General

Profile

Meetings Minutes »

Meeting (16012026)

TADC

  • run_number is ok in file name but not in binary data. Bohao should correct it.
  • event_number is not correct in binary files (corrected by gtot but not great). Should be managed by the CS DAQ. Bohao should correct it.
  • event_size ok (depends on the trace length
  • t3_number : We don't know exactly what it is ! Trigger number from t3 maker !
  • first_du, time_seconds, time_nanoseconds : Need to be updated by Bohao . Will be more natural when event builder will be released.
  • event_type, event_version: Should be the same as in Trun (unless we decide to have some mixed runs which would require to define another type of data like mixed data)
  • firmware_version: Bohao will discuss with Xushin to determine from where the 257 is comming from !
  • trigger_position, trigger_flag: Same as in Trunrawvoltage. Bohao
  • atm_temperature, atm_pressure, atm_humidity: Empty because sensor is unplugged. May be updated with next version if sensors are plugged again.
  • acceleration_x/y/z: Same as previous
  • adc_samples_count_ch: Not clear what it is !!! event total sample length (length of the traces) ?
  • fire_single_test_pulse, test_pulse_rate_divider: Not clear what it is exactly !!!
  • gain_correction_ch: Bohao will check with Xushin. Olivier suggests that we could hardcode it to 20 .
  • base_maximum_ch, base_minimum_ch: Not clear what it is !!! Channel properties from hardware ?
  • ioff: Not clear what it is !!! ADC trace offset (header length ?) ? Leave it as it is (because it is used).

How do we manage the need for new "custom" fields/trees ?

New data should go to new root trees deriving from mother_run_trees or mother_event_trees.
Need a process to "validate" the new trees.

Issues with directory format (see slides https://forge.in2p3.fr/attachments/543568 )

If the headers are properly filled we will have the first event time in all binary files... and the date_time will be predictible and the same for all binary files... thus the directory name will be correct whatever the binary file we process first.
If all run data are present in the binary header we should write the run trees only once (at first file processing) with the correct values. The change in gtot should be :
check if directory exists or not --> if directory is not present, create it and write the run trees. If directory exists, then just add event trees and skip run trees creation.
This will require Lech to modify gtot accordingly.
Thus having all correct informations (especially the first trigger time) in all binary files is critical.
For the reprocessing of existing data, we should apply the same mechanism except that at the end (when all the run has been reprocessed) another script should goes to all the created directories and correct names and recalculate run trees (a transitional text file storing all run infos from all bin files may be used to ease the run trees update).

Do we correct du_id values (with new naming for antennas) etc. during reprocessing ?

Pro:
Should allow to compare evolution and data from same antennas/position (which is not the case now because a same antenna can have different du_id over time and the same du_id can correspond to different antenas at different periods if the feb has been moved from an antenna to another)

Con:
This prevent comparison of already processed data with new processing as du_id will be completely different !

We could have a "correcting" gtot version to do it... but do we have time for it ?
We need to think more about it before deciding. François will launch a discussion in mattermost about this.