[Eg6_analysis] Data Acquisition System

[Dupré Raphaël]

Hello Nathan,

Very nice!

This seems to indicate that the noisy channels do not give very useful
information. This is a bad news, I am afraid we are going to loose a
bunch of pads in this story, but we should keep as much as possible
until we are sure no meaningful data can be obtain from them.

I think the first question now would be, should we use averaged
pedestals for the pads or event by event subtraction? I am not sure
how to test this. Maybe looking at a sample of good events hitting
some noisy pad would show if one or the other method is better (or if
the pad is completely useless).

For the right left asymmetry, it was clear during the run that one
side was more noisy than the other. It is not surprising that
occupancy reflect this and it confirms that the noise remain an issue
even after cutting many bad pads and applying the noise reduction
algorithm.

Best,


On Mon, Feb 4, 2013 at 5:03 PM, Nathan Baltzell <baltzell at anl.gov> wrote:
> Hello Everyone,
>
> Here's a report on RTPC pedastal/noise:
>
> http://clasweb.jlab.org/rungroups/lowq/wiki/index.php/RTPC_Pedestal_Investigation
>
> Let me know your comments and suggestions.
>
> -Nathan
>
>
>
>
> On Thu, 31 Jan 2013 11:16:39 -0600, Dupré Raphaël
> <raphael.dupre at gmail.com> wrote:
>
>> Hello guys,
>>
>> The system of data acquisition is set up such that it stocks the data over
>> a long time. When the trigger signal arrives the DAQ system takes data for
>> 8.5 micro seconds more and keeps values from -1.5 to 8.5 relative to the
>> trigger. TDC slots are of 100 ns and there you have 100 of them with the
>> trigger time at TDC = 15.
>>
>> Pedestals for the ADC values are calculated from a larger window taken
>> from
>> previous data. I do not remember the time window used for these pedestals,
>> but it is probably documented somewhere (it might be interesting to try to
>> find it). Finally the DAQ send some data, the data are sent by clusters of
>> consecutive TDCs. In order to define a cluster you need to have 3 (?)
>> consecutive hits with ADCs higher than 45 (?). Then the cluster is defined
>> by all these consecutive hits plus the 3 before and the 3 after in term of
>> TDC, so you have a minimum size of clusters of 9 (except on the edges of
>> the time sample). These additional 6 hits can be of any ADC value and can
>> eventually be used in order to make an additional pedestal, either by
>> using
>> a mean value of them or the 2 extremes (or any other combination that
>> would
>> be most appropriate) and subtract this to all the hits of the cluster.
>>
>> I am not completely sure of the final values we used in the run, so the
>> numbers followed by a "(?)" should be checked. It should be straight
>> forward to find them by looking into data.  The minimum size of clusters
>> will give the number of consecutive hits needed and the ADC distribution
>> has a discontinuity at the threshold level.
>>
>>
>> Best,



-- 
Raphaël Dupré