Int_t rdo = St_tpcRDOMapC ::instance()->rdo(sector,row, pad);
Yuri Fisyak
STAR Phone: +1 631 344 3913
Brookhaven National Laboratory Fax: +1 631 344 4206
510A/1-161
http://www.star.bnl.gov/~fisyak E-mail: fisyak AT bnl.gov
From:
star-tpc-l-request AT lists.bnl.gov <star-tpc-l-request AT lists.bnl.gov> on behalf of Van Buren, Gene <gene AT bnl.gov>
Date: Friday, October 18, 2024 at 5:00 PM
To: Videbaek, Flemming <videbaek AT bnl.gov>
Cc: Tonko Ljubicic <tonko.ljubicic AT gmail.com>, Ray, Robert L <ray AT physics.utexas.edu>, Star-tpc L <Star-tpc-l AT lists.bnl.gov>
Subject: Re: [[Star-tpc-l] ] sector 24 RDO 4 issues.
Hi, all
Flemming asked me to look for any observable event-by-event differences, so I looked at 6000 events from run 25292005 (taken this morning), file st_physics_25292005_raw_2000014. I then counted hits, sampling from 4 patches:
Patches 1 & 2: rows (25-28) by 9 pads in RDO 3 and RDO 4
Patches 3 & 4: rows (37-40-) by 19 pads in RDO 3 and RDO 4
I was conservative in my patch sizes because there are lots of dead areas in both of these RDOs. Unfortunately, there's no function in StEvent that says what RDO a hit was in (probably because hits can span RDOs for some padrows). What I see:
First: no significant spans of events without hits in any of the RDO patches. It should be noted that the events in the DAQ files are however not immediately consecutive.
========= Comparisons on patches 1 & 2 =========
Of all the 207607 hits found in these patches, 111835 (53.9%) are in the RDO 3 patch and 95772 (46.1%) are in the RDO 4 patch.
Event-wise....
1058 of 6000 events (17.6%) only had hits in the RDO 3 patch, none in the RDO 4 patch
...and of these 1058, 182 (17.2%) had two events in a row like this
691 of 6000 events (11.5%) only had hits in the RDO 4 patch, none in the RDO 3 patch
...and of these 691, 74 (10.7%) had two events in a row like this
3983 of 6000 events (66.4%) had hits in both patches, and in those events the number of hits in the two patches was roughly 52.4% in RDO 3 and 47.6% in RDO 4
========= Comparisons on patches 3 & 4 =========
Of all the 229001 hits found in these patches, 156177 (68.2%) are in the RDO 3 patch and 72824 (31.4%) are in the RDO 4 patch.
Event-wise....
1015 of 6000 events (16.9%) only had hits in the RDO 3 patch, none in the RDO 4 patch
...and of these 1015, 169 (16.7%) had two events in a row like this
619 of 6000 events (10.3%) only had hits in the RDO 4 patch, none in the RDO 3 patch
...and of these 619, 61 (9.9%) had two events in a row like this
4119 of 6000 events (68.7%) had hits in both patches, and in those events the number of hits in the two patches was roughly 51.5% in RDO 3 and 48.5% in RDO 4
========= Conclusions =========
1) The probabilities for seeing zero hits in RDO 4 patches is significantly higher than for RDO 3! This appears to be a real difference.
2) The probabilities for two events in a row without hits in any of the patches is reasonably consistent with their single event probabilities repeating randomly, i.e. there is no clear sign in this data of RDO 4 being dead for spans of time.
3) While the strongest bias appears to be seeing no hits in a patch in RDO 4 vs. RDO 3, there is also a weaker bias towards more hits in RDO 3 when there are hits in both RDOs.
4) My over-arching conclusion is that it doesn't appear to be that RDO 4 is dead for certain events, but rather that RDO 4's efficiency for finding hits is simply lower overall, for all events. That would be consistent with the cut-off for zero suppression
for RDO 4 being wrong by a small amount for the entire run.
========= Offline cluster finder =========
With that final conclusion in mind, I repeated the above analysis running the offline cluster finder on 198 events from file st_physics_adc_25292005_raw_1800005. The poor statistics don't allow for precise conclusions, but the percentages are consistent with
the findings from the online-found clusters shown above: the biases are still very much present.
Hope that helps, and I hope I didn't make any typos with all those numbers!
-Gene
> On Oct 16, 2024, at 5:04 PM, videbaek <videbaek AT bnl.gov> wrote:
>
> Hi tonko for reference the normal sector 24 offline looks like
> the attached plot
>
>
> On 2024-10-16 16:53, videbaek wrote:
>> Hi Lanny
>> I can confirm your observation. The last run in the fill of day 089
>> 289061 and that is bad,
>> but for the next fill as you sya 290008.
>> It turns out to be very hard to see the issue in the online plots
>> attach are the cluster plot for sector 24 in run289061, so we really
>> don't have any tools to see this online, but only after fast offline.
>> Tonko is there anything in the logs?
>> So in both cases the problem was fixed after a beam dump, but not between runs.
>> The short runs done were setRHICclock and emccheck , can this do something?
>> Flemming
>> On 2024-10-16 16:08, Ray, Robert L wrote:
>>> Hi Flemming:
>>> Unfortunately, the low number of hits in sector 24, RDO 4 is
>>> back again starting with run 25289050 for the rest of day 289.
>>> In the few runs I have looked at in today's data (290), the
>>> problem seems to have gone away.
>>> Thanks,
>>> Lanny
>>> -------------------------
>>> From: videbaek <videbaek AT bnl.gov>
>>> Sent: Tuesday, October 15, 2024 1:18 PM
>>> To: Star-tpc L <Star-tpc-l AT lists.bnl.gov>; Ray, Robert L
>>> <ray AT physics.utexas.edu>
>>> Subject: Re: sector 24 RDO 4 issues.
>>> Hi
>>> Tonko pointed out that current runs don't have this problem. So I went
>>> back at looked at the TPC offline QA
>>> and see clearly that the issues on sector 24 itpc 4 RDO was present
>>> over
>>> many daya
>>> first bad run 283030 last bad run 285046 since 286003 all good.
>>> The most peculiar is that nothing special is reported in the elog from
>>> last bad to first good
>>> =no pedestal, no power cycling, no reboot on DAQ.
>>> -single beam run 285047
>>> -blue beam dumped 22:43 10/11
>>> -Run 25286001 - setRHICClock 00:15 10/12
>>> - Run 25286002 - emc-check and 56*56 injected
>>> - run 25286003 Au-Au good run
>>> Was anything done to tpc's during this time? Can one check in log
>>> messages for tpc?
>>> best Flemming
>>> On 2024-10-11 17:03, videbaek wrote:
>>>> In the fast offline QA Lanny pointed out that it seems sector 4 has
>>>> fewer hits and this has been the case since Monday.
>>>> After today’s QA-meeting I got a root file with all the the
>>>> information with various information and looked at
>>>> 1) charge per pad/row
>>>> 2)cluster charge per pad/row
>>>> I focused on row 39 which has two region from rdo 3 and rdo 4. From
>>>> physics point of view they should have same #hits and charge per pad
>>> ,
>>>> in average.
>>>> The file is from run 25283030 in FastOffline,
>>>> The attached file has some additional comments , discussions, and
>>> plots
>>>> Thanks to Gene for providing the complete QA root file.
>>>> best Flemming
>>> --
>>> Flemming Videbaek
>>> senior scientist, emeritus
>>> videbaek @ bnl.gov
>>> Brookhaven National Lab
>>> Physics Department
>>> Bldg 510D
>>> Upton, NY 11973
>>> phone: 631-344-4106
>>> cell : 631-681-1596
>
> --
> Flemming Videbaek
> senior scientist, emeritus
> videbaek @ bnl.gov
> Brookhaven National Lab
> Physics Department
> Bldg 510D
> Upton, NY 11973
>
> phone: 631-344-4106
> cell : 631-681-1596<Screenshot 2024-10-16 at 17.03.17.png>
|
