Proton Calorimetry/Meetings/2020/11/18

From PBTWiki
Jump to navigation Jump to search

Minutes for UCL Proton Calorimetry Meetings, 18th November

Present

Ruben Saakyan, Raffaella Radogna, Saad Shaikh, Fern Pannell

Saad Shaikh

  • Took and analysed photodiode data at home and at UCL with Raffy.
  • Main conclusions of analysis:
    • Found that the ADC is noisier than it should be by comparison of standard deviation of test mode histogram with that of where photodiodes are unplugged. Any noise introduced by photodiodes is masked.
    • Measurements taken in D27 are likely subject to significant electronic noise – giving strange histograms when the two channels of a photodiode are combined.
      • Ruben suggested to check for bugs in the code, but none have been found yet. Additionally, combined histograms of dark measurements taken at home away from electronic noise make sense. Raffy will continue to work on and debug the plotting macro.
      • Will repeat (dark) measurements with detector but in an office to see if noise persists.
    • Ruben was very dubious about the way charge is split into two channels on the DDC232 board – better ways to do this (transistors?).
      • Behaviour of underflow measurements (where exactly 0 is returned by the DDC232) suggests there is something going wrong: at home, 0 measurements in one channel are paired with 'real' values from the other channel, such that underflow disappears in the combined photodiode histogram.
      • Underflow persists in measurements taken in D27, perhaps due to electronic noise.
  • While ADC is very noisy, might still be worthwhile to do some characterisation with intense LED injected directly into photodiodes.
  • Will take UV torch home during next visit to UCL, to conduct some light measurements at home, away from electronic noise.
  • Will contact Marko to investigate charge splitting and electronic noise problems.
  • Should also contact Texas Instruments to discuss apparent worsening of performance with increasing integration time and decreasing full-scale range. Unclear whether this observed behaviour is something introduced in the board design.
  • Will continue work on Nexys Video USB interface.

Raffaella Radogna

  • More comments from our discussion with Ruben
    • He really almost focused on the single channel distributions suggesting to check for a bug in the way we combint two channels
    • He thinks that a passive splitting is never ideal: usually you end up with 1/2 of the signal keeping the noise. Suggested to double check with Marko iif the passive splitting implemented on the board could cause the ADC noise when compared to Test mode results. Other options are active splitting, or splitting matching impedence usiing a 50 Ohm resistor. If we realise that the splitting is increasing the noise, we should avoid splitting and reduce the light intensity using neutral filters to couple PD and scintillator.
    • One possible interpretation of the observed increased RMS with integration time is due to the large noise at level of single channel plot. Increasing the int time the noise has more time to affect the measument, resulting in a larger distribution.
    • To bernchmark the ADC output against a calibrated signal, Ruben suggested to inject to the ADC input a square pulse signal using a pulse generator and check the ADC output.
    • Another useful comparison could be to translate the intensity from the sensor in charge-equivalent value and compare to the charge measured by the ADC.
  • Asked for an extension for IAA grant. Need to fill and submit an official "changes form" for Innovation and Enterprise.
    • Innovation and Enterprised asked if I need a change in the worktribe to motivate the purchase of the DAQ computer. I replied that a new worktribe is not needed and that the DAQ computer can go under the existing line "consumables - FPGA electronics" where £4k were asked (and about 1k already spent on two Nexys Video development boards)
  • Need to find motivation to change the FLF application to fit the EPSRC open fellow requirements.
  • Will keep looking for a bud in the code to justify spikes iin the PD distributions.

Fern Pannell