1. Update on Noise Analysis of the DHCal Test Beam at Fermilab
Lei Xia

2. CALICE collaboration meeting @ CERN
Outline
‘Dead’ ASIC’s
Temperature monitoring of Apr – May run
Update on RPC noise
Update on correlated noise
Update on noise ‘hot’ spot
CALICE collaboration CERN

3. CALICE collaboration meeting @ CERN
‘Dead’ ASIC’s
We reported at the ALCPG’11 conference that a tiny fraction of the FE asic’s will not give any data in any run type
We identify them in noise runs
For Oct. and Jan. – Feb. runs, average fraction is only 0.27%
Things are more stable in Apr. – May run
For most part, we miss 16 asic’s (max 27, min 13)
This is only ~0.22%
Max: 0.44%
16 missing
ASIC’s
(~0.22%)
Apr.— May run
Oct. and Jan.– Feb. runs
CALICE collaboration CERN

4. CALICE collaboration meeting @ CERN
Noise monitoring
The DHCal prototype (1m3 and TCMT) is capable of setting the FE freely running
All FE signals are recorded, up to a certain rate limit (not likely to reach without beam)
Perfect running mode to measure noise in RPC
We use this running mode to monitor RPC noise during test beam
2+ noise runs per day
Important data quality monitoring tool
CALICE collaboration CERN

5. Temperature/noise record for Oct. run
Air conditioning fixed
Air conditioning briefly down
CALICE collaboration CERN

6. Temperature/noise record for Jan.– Feb. run
Gas flow reduced to 150cc/min, from 300cc/min
CALICE collaboration CERN

7. Temperature/noise record for Apr.– May run
Didn’t run
due to high T
Portable AC
in place
CALICE collaboration CERN

8. CALICE collaboration meeting @ CERN
RPC noise estimate
Using the measured noise rate, we can estimate the expected noise level in triggered beam data
Assume all measured noise in self-triggered runs is from RPC itself (not exactly true)
Total number of channels in 1m3 + TCMT (52 layers) is 96x96x52 = 480K
Not including any possible correlated noise
Noise contribution to triggered beam events is extremely small (almost negligible), even when T is unusually high
TCMT noise level, Apr.-May run
1m3 noise level, Apr.-May run
RPC Noise rate
(Hz/cm2)
0.1
0.5
1.0
2.0
4.0
Nnoise/evt
200ns gate
0.0094
0.047
0.094
0.19
0.38
700ns gate
0.033
0.165
0.33
0.66
1.32
Expected noise level for Oct. and Jan. runs
Expected for a ‘cool’ DHCal stack
CALICE collaboration CERN

9. Update on correlated noise
By comparing self-triggered (trigger-less) noise data with randomly triggered noise data, we found correlated noise in the system
It is grounding related
Possibly behaving differently with/without trigger
May be different for noise/beam data? – need study
Relation to (different) beam condition? – need study
Careful study on-going
Define characteristics for correlated noise
Identify/eliminate correlated noise in beam data
Estimate the remaining correlated noise in data (if any)
Comparison between two data sets
Trigger-less vs randomly triggered
Green: two data sets show consistent noise level
Red: statistically inconsistent noise level, that suggests
Existence of correlated noise
Deferent behavior in different running conditions
CALICE collaboration CERN

10. Basic “features” of the correlated noise
Hits at RPC ground connector
Hits on the boarder of FE board
A lot of hits on one FE board (up to all 1536 pads)
Light-up single ASIC (all or almost all pads firing on one ASIC)
Events from trigger-less noise run
CALICE collaboration CERN

11. Correlated noise with beam event: type 1
Extra hits on RPC ground connectors
Easy to identify, easy to eliminate extra hits
CALICE collaboration CERN

12. Correlated noise with beam event: type 2
A lot of extra hits in a few FE boards
Easy to identify, events can only be discarded
CALICE collaboration CERN

13. Correlated noise with beam event: type 3
Extra hits on FEB boarders
Not easy to identify, events probably need to be discarded
CALICE collaboration CERN

14. Correlated noise with beam event: others
Type 4: events with ‘hot’ asic
Type 5: any combination of 1 – 4
Study is on-going
Overall, it affects a very small fraction of the data sample (limits to be given)
CALICE collaboration CERN

15. CALICE collaboration meeting @ CERN
Noise ‘hot’ spot
Noise ‘hot’ spots are seen in all three test beam periods
Nearly no visible effect on beam data (a little bit on multiplicity)
Varies with time, temperature, gas flow rate, etc.
NOT seen in the ‘cooler’ tail catcher
Believed to be related to insufficient glass surface cleaning
Showing up ONLY with elevated temperature
CALICE collaboration CERN

16. More evidence for T correlation: layer 14
Run
Last noise run in Feb. run
Two pronounced hot regions
Run
1st few noise runs in Apr.
when stack is still cool
Hot regions disappeared
CALICE collaboration CERN

17. More evidence for T correlation: layer 14
Run
After stack reached stable
temperature (pretty high)
Hot regions came back
and slightly worse
Run
Just started using portable
air conditioning on 1m3
Hot regions shrank
CALICE collaboration CERN

18. More evidence for T correlation: layer 14
Run
1m3 further cooled down
Hot regions shrank further
Similar trend seen in other layers as well
Temperature is indeed the KEY factor here
CALICE collaboration CERN

19. CALICE collaboration meeting @ CERN
Summary
Number of ‘dead’ asics is very small, and stable
RPC’s are in good shape after three beam tests
Average noise level is stable
Absolute noise level is lower due to improved cooling
RPC noise contribution to triggered beam data is extremely small (<0.1 hit/event)
RPC contribute negligible noise hits to beam data
Better understanding of correlated noise
However, more study is needed and in progress
Need to define its contribution to beam data
Noise ‘hot spots’ were due to unclean surface
Strong correlation with elevated T
CALICE collaboration CERN