[Halld-online] number of samples for pulse integral and pulse pedestal

[Mark Macrae Dalton]

Hi Richard,

There are a few numbers that need to be tuned to get the required behavior.  The window which I believe is now 55 samples long is merely the place to look for the pulse.  This should be long enough that the pulse itself is separated from the beginning and the end.  There is no major penalty for it being "to long."  The lookback distance (latency in some parlance) should be tuned to put the pulse into the correct (central) part of the window and possible jitter should be taken into account.  All the action happens once there is a threshold crossing.  The location of the threshold crossing is used to determine the pulse region for integration purposes.  Then NSB parameter is the number of samples to include in the integral before the threshold crossing and NSA is the number of samples to include after the threshold crossing.  In your case something like NSA = 10 and NSB = 3 would be sufficient.  Integrating over a few extra samples is generally much more benign that too few.  The number of pedestal samples is fixed at 4 and then “averaged" by bit shifting.

Best,
Mark Macrae Dalton

--
Staff Scientist, Jefferson Lab
office: +1 757 269 6931
mobile: +1 757 849 2929

> On Oct 23, 2014, at 9:22 AM, Richard Jones <richard.t.jones at uconn.edu> wrote:
> 
> Hello David, Sean,
> 
> You have probably seen that I have checked in the first iteration of our online monitoring plugin for the TAGM. 
> 
> On a related issue, are we supposed to tell you how many samples we want to have in the fadc for computing the pulse integral and the pulse pedestal? I see in the rawevent plugin that a default of 55 samples is being used. Perhaps this is just a placeholder, but that is clearly too long for the TAGM pulses. I can send you some raw pulse waveforms for reference if you like, but our rise time (10/90) is 1.5 samples (on the scope it is faster, but I think the FADC receiver amplifier is shaping it with a bandwidth limiter) and the fall time is 8 samples, so 12 samples (50ns) is about right for the pulse integral if you want to capture the entire pulse. At high rates we should probably consider reducing it to more like 8 to reduce pile-up effects. For the pulse pedestal I should think that 3-4 samples would be ok.
> 
> -Richard J.
> _______________________________________________
> Halld-online mailing list
> Halld-online at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-online