Calorimeter Status: Preparing RUN II & LS2 Pascal Perret LPC Clermont 24 February 2015
Dec 13, 2015
3
Reopening of the Calorimeter …
CALO opening (January 28) Few problems to fix:
PS: 1 VFE board A side (64 channels)• Back before to change it
ECAL: 1 PMT line (20 PMT) + 2 PMT non operational cells at the A side, because of 3 dead CW• + 4 PMT C-side
Pascal Perret - LPC Clermont24/02/2015
access
ECAL LEDs
• The 4th problem seems to be related to the FEB and not to the PMT
4
Reopening of the Calorimeter … HCAL: benefit of the access to:
Fix one dead cell side A New shielding for few PMT:
• the idea was to try to reproduce the ECAL situation in electrostatic configuration – decouple the PMT glass from ground
• add insulator between the steel and the µ-metal shields
• remove rubber O-rings• in order to study the PMT
degradation problem, modify magnetic shielding in a part of PMTs (66 pc, only new ones, installed in 2014)
Need to re-run the 137Cs calibration before physics data taking
Pascal Perret - LPC Clermont24/02/2015
GND
O-rings(conductive)
µ-metal shieldcontacts to ground
GND
conductive O-ringsremoved
wrapped with plastic tapeno contact to gnd
5
Calorimeter LS1 Activities
ECAL and HCAL ECS ECAL:
Both DAQ and HV are configured via the new SPECS • works well, with both SpecsServer and CAT – a great success!!!• Patrick and Clara continue debugging, to achieve perfection
HCALDAQ and HV work via old SPECS. the 137Cs source control and readout of integrators are migrated to
linux (V. Kudryavtsev, BINP)• integrated into DCS (the HCDCS project was created by Clara)• fully operational (used for the source run on 06/12/2014)• there is room for improvement of interface etc• in case of migration to new SPECS masters, the 137Cs control may
require a dedicated SPECS node (to be tested).
Pascal Perret - LPC Clermont24/02/2015
6
Calorimeter LS1 Activities
CALO opening (January 28) Closed Wednesday 11/02
Last minute: Water leak on demi water circuit?
~3l lost since ~1 month … some drops identified on a manifold
last week• Laurent/Cédric are investigating …
Pascal Perret - LPC Clermont24/02/2015
7
SPD status
VFE 24 noisy cells Two noisy cells were found in one of the replaced VFEs during
cosmic data taking:
Pascal Perret - LPC Clermont24/02/2015
Solution: higher thresholds were configured.• 735: 1 MIP• 794: 0.7 MIPS• To be checked
CellIDs = 735, 794
Dec. 2014
8
SPD status Check done during operation week (01/26/2015)
Cosmics data taken: looks ok!
Pascal Perret - LPC Clermont24/02/2015
ECS software improvements done
9
Calorimeter LS1 Activities: summary
A lot of movements/activities: ~275m
Pascal Perret - LPC Clermont24/02/2015
+ Maraton maintenance, Addition of TELL1, etc.
HCAL ECAL PS LEAD SPD HCAL ECAL PS LEAD SPD19/03/2013 5 5 4 4 6 6 5 4,5 4,5 5 O02/07/2013 1 C03/09/2013 3,8 4,8 4 4 4 C26/09/2013 3,5 3 4,5 C30/10/2013 4,5 3,5 C29/11/2013 1 0,5 0,5 0,5 C16/01/2014 4,8 O19/02/2014 4,6 O25/02/2014 4 O04/03/2014 4,8 4 4,5 C, OO19/06/2014 4,5 4,5 4,5 O25/06/2014 9 C/O22/07/2014 4,8 4 3,5 4 O29/07/2014 4,8 9,6 O,C,O04/11/2014 8 4,8 4,8 O/C, C06/11/2014 4,5 4,5 4,5 4 3,5 4 C10/11/2014 4,8 O13/11/2014 4,5 4,5 4,8 C28/01/2015 4,5 2 2 2 O02/02/2015 5 O09/02/2015 5 O11/02/2015 5 5 2 2 2 5 C
28 31,5 21 20,5 34 38,6 43,6 16,5 19,5 21,5 274,7
A C
ECAL Fibre installation
XCAL PMT
SPD VFECERN Open Days
PS VFE
Muon activities
Passive dosimeters
Closing/Alignment
XCAL PMT
Moving system maintenance
Closing
10
Calibration
An alternative ageing measurement has been developed by Irina and taken over by Marie-Noelle & Jean-François:
it is based on raw occupancy: the method as a reminder
Pascal Perret - LPC Clermont24/02/2015
Precision reached with 1 hour data taking
11
Calibration Occupancy method
Compare occupancy wrt reference sampleRatio of occupancies proportional to changes in hardware
characteristicsRequest HV changes if occupancy ratio passes some criteria
• Deviation > 1%• & deviation > 2-sigma of its resolution
This method was used in November 2012 to update ECAL HVIt will remove the ageing dependence from triggerDepends on the instantaneous luminosity, trigger type, magnet
polarity, etc;• Need proper selection of events and of stable periods
Statistics needed: goal: reach 1% error on the occupancy ratio• 600K : 1% reached for half of channel (80Hz → 168 k/hour)• Most channels 2000K ~ 10 hours
Possible to check the agreement with LED measurement
Pascal Perret - LPC Clermont24/02/2015
12
Calibration scheme 80Hz minimum bias data from ODIN/L0
Sent to a dedicated farm e.g. Calo Calibration farm as CALD007
• Produce occupancy profile histogram each 10 min (50K)• Store on Saveset/ CaloDAQxx with on line Timestamp
From on line luminosity info select stable time interval
Calculate gain change from selected stable luminosity intervalProduce gain change table
Produce new PMT HV settings
First check of ROOT files production made successfully on FEST 2 weeks ago; new tests this week at the pit
Pascal Perret - LPC Clermont24/02/2015
Dat
a ta
king
Inte
rfill
per
iod
LED response could be used as a cross-check and where the statistics is too low (outer part)
13
Calibration: Occupancy method
Further studies ongoing: Sample selection
A sample not tagged on calorimeter could be used instead of MB: Di-muon sample is a good candidate!
Pascal Perret - LPC Clermont24/02/2015
Occupancy comparison:• Factor at least 2• Up to a factor 2.8 in outer region
Difference on the derived corrections MB wrt di-muon: <0.2%, RMS=2%
•Further checks on-going
14
Calibration: Occupancy method
Further studies ongoing: magnet up/down transition
Occupancy change after a polarity transition even with a small ageing• Fill 2266 Magnet up - Fill 3273 Magnet
down• Similar effect for down/up transition
Pascal Perret - LPC Clermont24/02/2015
Idea: to build a map to derive the effect of the transition
Plans:Precise determination of the ratio using all the up/down transition of
2012Use this map on 2012 data → test method accuracy.
• The map can be used at the root level.More study needed to study the mu dependence
15
Calibration
Occupancy method will not give the initial calibration … Fine π0 calibration needed (with FMDST)! Alternative to off-line fine calibration under investigation
Use on line cluster to calculate π0 mass candidate / cell• Uses low multiplicity event : SPD hit selected • Cell without SPD hit matching to highest cell used• Cell with more than 60% of photon cluster energy• Request PS contribution to photon energy <10% of the cluster energy. • Book 1 histo/cell• Fit each histogram• Determine gain to be changed using determination precision from fit
qualityReplace iteration by adjusted HV on the next bunch of events
• Data needed from 8TeV runs are estimated to be ~20 hours of data taking.
• Use of occupancy method to correct from ageing from fill to fill
Pascal Perret - LPC Clermont24/02/2015
17
ASIC
2 concurrent solutions for the analog electronics The ASIC has been chosen Present version very close to the final one. The last version is
being submitted and should be received in June 2015
2 test beams are planned: June 2015: most probably using the current version November 2015: final ASIC soldered on a new FEB
This is an aggressive schedule for the FEB
The EDR of the ASIC should take place soon (April) However, we may wish do the EDR after the test beam
Pascal Perret - LPC Clermont24/02/2015
18
New Front-End Board
CAD design of the new FEB prototype about to start Should be as close to the final version as possible
Use final components32 channels
Components are definedCAD documentation received from CERN (DC-DC, GBT, …)
However, we need to decide on the number of fibres per FEB 2 solutions proposed
5 fibres: 4 for DAQ and 1 for the LLT• No data reduction (full BX id transmission, 4 fibres fully independent, …)
4 fibres: LLT data embedded on the 4 DAQ fibres• Data reduction necessary, more risky, but much more cost efficient !
Organize a meeting with the online experts by the end of March The firmware of the FE and Trigger PGA of the new FEB is
almost written
Pascal Perret - LPC Clermont24/02/2015
19
New FEB
Pascal Perret - LPC Clermont
GBT
VTTx
GBT
GBT
GBT
GBT
ICECAL-ADC-FE FPGA (4ch)
VTTx
VTTx
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
ICECAL-ADC-FE FPGA (4ch)
GBTSCA
Trigger FPGA
Old FEB First idea on the topology on the future FEB
24/02/2015
20
Next steps and milestones Choice for the Analog electronics Choice between GBT/IGLOO2 for the serialization Decide on the number of fibres per FEB by the end of March EDR/PRR milestones:
EDR FE Analog → April 2015 (July ?)
PRR FE Analog → Sept. 2015EDR FE board → Sept. 2015PRR FE board → Jan. 2016EDR Ctl board → Jan. 2016EDR Electronics (HV, Monitoring, Calib.)→ Jan. 2016PRR Ctl board → July 2016PPR Electronics (HV, Monitoring, Calib.) → July 2016
Ready for installation → Sept. 2018
Pascal Perret - LPC Clermont24/02/2015
21
Summary
A lot of work achieved during LS1 Detector status
Detector closed: all channels are functionalA small water leak has been detected on demi water system
• Is it on PS/SPD system? To be followed …Maraton maintenance?
CalibrationGood progress made
• First tests successful with FEST • Still some work to do and new ideas on the table
HCAL Cs source scan to be performed Upgrade:
Final ASIC version being submitted and should arrive in summerCAD design of the new FEB proto about to start
Ready to take data!
Pascal Perret - LPC Clermont24/02/2015
23
HCAL status
HCAL PMT problem – follow-up (unstability and large dark current) Discussion with Hamamatsu on-going
Few samples sent
Pascal Perret - LPC Clermont24/02/2015
From HAMAMATSU: light emission shows the place where dark current occurs
Found by HAMAMATSU experts:• the vacuum inside the PMT bulb is
not degraded, no micro cracks in glass
• the problems (dark current etc) are caused by migration of alcali inside the bulb: but why???
• one PMT had been baked for 1.5 hour at 125°C – the behavior improved significantly
• however the HAMAMATSU experts do not expect the improvement to be stable for long term
24
HCAL PMT problem – follow-up
Pascal Perret - LPC Clermont24/02/2015
While the way of degradation was found (migration of alcalis within PMT bulb), the reason why it happens in ECAL and not in ECAL is unclear. We can only make hypotheses.The differences in the HCAL/ECAL operating conditions which can be relevant are only
somewhat higher voltage (900 – 1300 vs 600 – 900)in combination with electrostatic conditions (grounded bulb)
The limit operating voltage for this type of PMTs is 1800 V according to the specs, it is not exceeded. However an additional factor leading to the decrease of allowed voltages can be the special treatment of the dynode system, which HAMAMATSU made for us in order to decrease the rate effectWhen purchasing additional PMTs from HAMAMATSU, we can think of asking them to produce (a significant part of) PMTs without such treatment. The rate effect will of course become larger; however it will be mitigated by the 11 kHz LED pulsing. It can be a good compromise for (most of) HCAL, where the requirements are not very tight (5% will be good enough) – of course if the guess is right and PMTs without treatment will not experience such degradationWe can also try to modify the electrostatic conditions in HCAL, by, e.g., wrapping the µ-metal shield with kapton tape In the meanwhile, we can bake the bad PMTs (we have ~150 for the moment) and see whether it will help; however HAMAMATSU experts do not expect steady improvement