[Halld-tracking-hw] CDC pedestal readout needs

[David Lawrence]

Hi Beni,

   I would still advocate for the combined format Naomi referred to as "D".
At the very least, this would be useful for checking the algorithm is doing
what is expected. I would also push to have data read out in this format
at a highly prescaled rate to in order to give us an additional in 
data-stream
diagnostic should we need it. The prescale factor would be set so as to
have a minor or negligible affect on the data size.

Regards,
-David


On 7/30/13 11:17 AM, Beni Zihlmann wrote:
> Hi Naomi,
>
> 1) A.) can be accommodated by raw event readout. Anything that has to do
> with determining the  proper parameters can/should be done with raw
>       event data and special runs.
>
> 2) C.) is happening all the time because we get pedestal information
> with the ADC integral information (first data word). if a channel is
> never above
>       threshold we never get a signal and never any pedestal information
> as you correctly state but that is as good as bad pedestal information.
> It tells us
>       we have a missing channel, so we catch the problem and can check on
> it (your last point). The oline  monitoring program and shift crew
> should catch that.
>
> 3) I currently see no need to have more than two modes of readout/formats.
>       a) raw readout that handles any initial determination of setup
> parameters and possible routine checks on these parameters every
> day/half-day or your most
>           favourite time scale.
>       b) the format we defined with 2 32bit data words. Where we have all
> the necessary information to check if the channels are healthy and we
> even can
>           correct the ADC integration values if the initial pedestal is
> off because a previous event distorted it. All this is done offline
> anyway and does not
>           need to be done on the FPGA.
>
> I am currently writing down a short explanation on a first attempt of an
> algorithm that satisfies both CDC and FDC detectors in the hopes that this
> algorithm can be improved on and at the same time is independent of
> whether it is used for the FDC or CDC and of course fits onto the FPGA in
> the first place.
>
>
> cheers,
> Beni
>
>
>> Hi all,
>>
>> The last email stream on pedestals etc made me a little dizzy.  Here I've tried to get all the CDC's pedestal measurement needs onto one page, to be sure that they are possible with the planned data formats.   I would guess the FDC would have broadly similar needs.   ?
>>
>>
>> A. Initial setting
>>
>> Several samples with full range 0-4095 of adc value in order to set the pedestal offsets - I collect a number (10-100) of pedestal samples from 50-100 events and take the mean value.  There are several ways to do this (look at less samples, calculate the mean in the flash etc) but it will be easiest with full range readout available.  Also we need to measure the pedestal width which will be needed in order to set the hit thresholds in the flash.
>>
>> B. Event data
>>
>> I need some measure of the pedestal before the hit in order to determine whether it is still high from a previous hit and what is the difference between pedestal and overflow.
>>
>> C. Routine pedestal monitoring
>>
>> Pedestals need to be monitored so that any drifts can be corrected and also to keep an eye on the width (important for hit thresholds).
>>
>> D. Routine fadc health monitoring
>>
>> We should take some event data with full range and full precision sample readout to check that the adcs are ok (no bits missing) and check performance of the flash algorithms.
>>
>>
>>
>> B.  This is in the pulse data format; almost certainly a subset of the full adc range.
>> D.  Would be included in the combined format for pulse data followed by raw sample data as continuation words.
>>
>> If we have dedicated runs with a different data format & different trigger & hit-less data then A&C could use the same raw data format.  (We could also make the flash calculate a mean & width for a different pedestal format).
>>
>> [I thought of monitoring pedestals via the pulse data B or raw data in the combined format for D (if they start before the trigger), but we would only be reading out the channels with hits, so would not see problems if there is a channel without hits because its pedestal has plummeted.]
>>
>>
>> Can anyone see anything that I have forgotten?
>>
>> Naomi.
>> _______________________________________________
>> Halld-tracking-hw mailing list
>> Halld-tracking-hw at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/halld-tracking-hw
> _______________________________________________
> Halld-tracking-hw mailing list
> Halld-tracking-hw at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-tracking-hw