|
Post by jsmith on Feb 16, 2015 12:25:30 GMT -8
The GRIF-16s exhibited variable relative counting rates during the November/December 2014 experiments. We observed this in online data viewed through the Analyzer and it is also visible in the offline data. I've attached two plots that show the hit pattern (number of counts in each channel) as a function of time for run 2369 from the 47Ca data. The top plot shows all channels on a log scale and the bottom plot shows only the GRIF-16 channels (no 4G information) on a linear scale. As you can see, at discrete times during the run, the relative counting rate of the third GRIF-16 changed. My initial thought was that this could be corrected by plotting the number of lost "ChannelId" values over time, but after spending a few minutes trying that, I don't think ChannelId holds the information we need.
|
|
|
Post by r3dunlop on Mar 1, 2015 14:21:44 GMT -8
Does this run contain multiple cycles? It almost looks like that digitizer is not counting at the start of the cycle. I know you have probably thought of this, but are the time stamps corrected for this run?
|
|
|
Post by jsmith on Mar 2, 2015 16:42:37 GMT -8
Yes, this run contains many cycles. You can distinguish them by the green lines on the top plot. It does look like that, but I think that's just coincidence for most of the changes. The last time the third GRIF-16 drops down in count rate (shortly after t=58000), it does so within a single run.
I have corrected the timestamps based on the first subrun in this run. I haven't seen any indications that the timestamps are changing in the middle of a run, but it's probably worth a directed look at the times when these relative rates change.
|
|
|
Post by r3dunlop on Mar 3, 2015 5:57:25 GMT -8
One thing I have noticed is that the system writes the same digitizer in bunches. This makes sense as it is probably dumping a buffer. I could see this being a problem though if we have a network problem at the instant one digitizer is being dumped to disk as we would lose mostly that digitizer. I seem to recall that we also may have seen hints of the collectors preferring certain digitizers. What I mean by that is at high rates we saw DIG 0 > DIG 1 > DIG 2 > DIG 3 in the hit pattern, but at lower rates these evened out. I'm not sure what would be causing this and I haven't investigated it enough to have any more information than this.
|
|