1. Calorimeter and its Upgrade Pascal Perret LPC Clermont 8 December 2015 OUTLINE  Detector status  Calibration  Upgrade

2. 2 DETECTOR STATUS 08/12/2015 Pascal Perret - LPC Clermont

3. 3 Pb-Pb event  PS, ECAL, HCAL: Nominal PMT gain; SPD: Gain/2 Pascal Perret - LPC Clermont 08/12/2015

4. 4 SPD & PS during Pb-Pb collisions Pascal Perret - LPC Clermont 08/12/2015 SPD gain/2 (equivalent threshold~1 MIP)~300 PrsAdccounts/SPD hit PbPb, Run 168523, 26 Nov ~50 PrsAdccounts/SPD hit 13 TeVpp, Run 167116, 2 Nov

5. 5 SPD during Pb-Pb collisions No saturation, nice correlation # SPD hits vs VELO Pascal Perret - LPC Clermont 08/12/2015 Julien talk’s

6. 6 SPD  First look at SPD efficiency during run 2 (July 10 th – 20 th ) Reminder: some drop observed at the end of 2012 Pascal Perret - LPC Clermont 08/12/2015 APRIL 2012OCTOBER 2012 July 2015: The efficiency drop pattern is still observable (as expected) Some problems coming from PS Requirement on coincidence between PS and SPD cells for efficiency computation Thresholds updated on Nov. 10 th To be checked with 5.TeV data

7. 7 SPD  Look at spillover with 2015 data: http://indico.cern.ch/event/455324/contribution/3/attachments/1177017/1702172/SPD_Prs_spillover_27Oct2015.pdf TAE data taken the13th of August 2015 The fraction of signal observed after the current BX is explained by The time response of the detector (the subtractor corrects this one). Very slow secondaries produced mainly in interactions of initial particles with the detector material. Decay of long lived particles (neutrons, muons, pions and kaons). Backsplash particles produced in the material of ECAL and HCAL. Next1/T0 occupancy ratio was 1.3(3)% for inner cells, 3(5)% for middle and 6(10)% for outer in 2010 data (14 TeV MC). Pascal Perret - LPC Clermont 08/12/2015

8. 8 SPD: spillover Pascal Perret - LPC Clermont 08/12/2015

9. 9 SPD & PS: spillover Pascal Perret - LPC Clermont 08/12/2015

10. 10 PS  PS status Pascal Perret - LPC Clermont 08/12/2015 PS pedestals refreshed on Oct. 22 New HV settings for two PS tubes PS2FEB8 still not working properly VFE + MAPMT to be changed during YETS

11. 11 ECAL ECAL status 11 dead and unstable cells (bad contacts, to be fixed during YETS). 3 dead LED 7 dead LEDs were replaced in TS3 All LEDs will be replaced during YETS to (more) reliable ones. Pascal Perret - LPC Clermont 08/12/2015

12. 12 ECAL ECAL LED status Pascal Perret - LPC Clermont 08/12/2015 The high LED noise problem was understood: in 8 cases out of 9 this was LEDTSB instability, cured by incrementing their delays by 1 ns In one case this is unstable LED driver, to be changed during YETS Before TS3: very high RMS in few LEDs (mostly compensated in LED/PIN) LED After TS3: one (C125#1) left (bad LED driver)

13. 13 ECAL  Timing problem in one crate? Fill 4530 / Fill 4528 was pointed out when looking at occupancy ratio Pascal Perret - LPC Clermont 08/12/2015 Looking at TAE taken in September: Crate 20 seems to be off by more than 3ns ??  To be understood Crate 19 Crate 20

14. 14 HCAL  HCAL status 1 dead cell Pascal Perret - LPC Clermont 08/12/2015

15. 15 HCAL Pascal Perret - LPC Clermont 08/12/2015 But, no degradation is seen in PMTs replaced in 2011 and 2014 (total of 183 PMTs were replaced). 137 Cs source calibration results HCAL dark currents: The PMT degradation continues: during LS1 all the PMTs with dark >10 nA: 163 PMTs >50 nA: 59 PMTs >100 nA: 31 PMTs current >50 nA were replaced to new ones. Order of 120 PMTs would need replacement in the winter shutdown…  Additional PMT to order!

16. 16 HCAL Rate effect On average, rate effect is moderate. It depends on lumi Pascal Perret - LPC Clermont 08/12/2015 special fills (elastic and diffractive), low lumi (HV updates OFF) in some cells the rate effect is high 1% 10%

17. 17 HCAL Rate effect Pascal Perret - LPC Clermont 08/12/2015 The measure of rate effect is a ratio of LED amplitudes at the end and beginning of a long fill (4569). High positive rate effect accompanies high dark current. For normal PMTs, rate effect is negative, ~ 0…-2%

18. 18 CALIBRATION 08/12/2015 Pascal Perret - LPC Clermont

19. 19 HCAL Cs calibration The HCAL light yield degradation plot, adding 2013 and 2015 137 Cs runs: 17/06, 02/09 & 11/11 Pascal Perret - LPC Clermont 08/12/2015 2013 and 2015 Fastest ageing at the startup 2011 2012 TS1 TS2 29/08 TS3 Now the ageing continues at ~ same speed (not completely correct: in rad. dose, 1 pb -1 @13 TeV is equivalent to ~2 pb -1 @8 TeV)

20. 20 HCAL Cs calibration HV modification are implemented in the PHYSICS recipe Pascal Perret - LPC Clermont 08/12/2015 The calibration was performed on 11-Nov-2015 evening; the PHYSICS recipe updated 12-Nov Smaller gain change thanks to LED following. Significant gain modification for the unstable cells – even with LED following. HCAL HV updates with LED in automatic mode since Oct. 4 th

21. 21 HCAL LED Calibration Pascal Perret - LPC Clermont 08/12/2015 ref 2, Cs calib 11-Nov TS2 TS3 start auto LED updates ref1, Cs calib 02-Sep special fills (elastic and diffractive), low lumi, LED updates OFF Pb-Pb, LED updates ON 5 TeV, LED updates OFF Pb-Pb, LED updates OFF start LED updates 1%

22. 22 HCAL LED Calibration  L0 Hadron/Electron rate As a reminder (July 2012) HV corrections based on LED each ~100pb -1 Pascal Perret - LPC Clermont 08/12/2015 July 2012  Now: Removal of very HCAL central cells Automatic HV corrections based on LED

23. 23 HCAL LED Calibration Pascal Perret - LPC Clermont 08/12/2015 December 2015 L0 Hadron/L0 Muon+Dimuon (HLT) L0DU Electron/L0DU Muon (L0DU) XCAL HV update New TCK ox10600A2 ox10600A3 TS2 ECAL HV update  L0 Hadron/Electron rate: HCAL HV updated after each fill

24. 24 ECAL Calibration Pascal Perret - LPC Clermont 08/12/2015 for ECAL: still in a “semi- automatic” mode (generating automatically, applying after verification by a human) Fine calibration + new reference Special fills HV update First inclusion of outer mask 06/09 08/07 07/10 03/11 No correction applied! New CondDB

25. 25 ECAL Calibration  Online calibration with LED: LED are doing the job!  Online calibration with π 0 Some additional improvements in the process: Reconstruction step split not only between the 1700 nodes, but also between ~20-30 cores on each node Lead to a good speed-up of the procedure The whole sample was reconstructed in ~2 hours (~8×10 7 events) Was ~6 hours for ~5×10 7 events before (+ ~7 hours for the fit itself) Very few badly (over) calibrated cells (fit doesn’t converge) Pascal Perret - LPC Clermont 08/12/2015 Where are π 0 ?

26. 26 Plans for YETS  SPD: - (SPD CB test setup to ressurect)  PS: Replace 1 MAPMT block (PS2FEB8)  ECAL maintenance: Replace all old type LEDs Fix dead/unstable cells and LED driver(s)  HCAL maintenance: Replace PMTs with instability and dark current (~100)  Varia: Maraton maintenance Check ATI connectors at the input of ECAL and HCAL FEBs 137 Cs source annual inspection (15-Dec) HCAL 137 Cs calibration before the end of the YETS Pascal Perret - LPC Clermont 08/12/2015

27. 27 UPGRADE NEWS Pascal Perret - LPC Clermont 08/12/2015

28. 28 Test beam @ CERN Pascal Perret - LPC Clermont 08/12/2015 Lecroy integratio n FEB prototyp e Trigger (coincidence ) Measurements laptop ECAL Module in the zone Faisceau Purpose was to check the performances of the analog electronics Electrons up to 120 GeV Check of the linearity Comparison of two acquisitions Fast integration based on the ICECAL electronics Measurement based on a Lecroy integrator

29. 29 Spill-over: Signal in preceding (T-1) and following clock cycles (T1, T2, T3, …) ASIC Pole Zero permits to reduce the Spill-over Tested with different PMT CW bases Pascal Perret - LPC Clermont 08/12/2015 Test beam @ CERN Spill over at different energies:Spill over with different PMT+base:

30. 30 Measured noise in the cavern and with the present system Nominal conditions of gain Minimum analog gain (1/2 of nominal gain) Only dynamic pedestal subtracted values shown Effects of Cockcroft-Walton HV sources and cables increase noise To be understood: Noise increase when detector connected not seen in current electronics Pascal Perret - LPC Clermont 08/12/2015 Electronics noise Nominal gain noise (MeV) ~½ gain noise (MeV) Not connected3,8 (1,6 LSB)4,5 (1,1 LSB) ECAL typical6,3 (2,6 LSB)7,0 (1,7 LSB) ECAL worst case6,9 (2,9 LSB)7,4 (1,8 LSB) HCAL typical5,2 (2,2 LSB)6,0 (1,5 LSB) HCAL worst case6,9 (2,9 LSB)7,5 (1,9 LSB) Calibrations Nominal gain: 2,38 MeV/LSB ~1/2 gain: 4,06 MeV/LSB

31. 31 Electronics Readout Problem discovered in the processing of the TELL40 True bandwidth is 100Gbits/s (80Gbits/s + margin) In fact, it is more 2x50 Gbits/s … Several solutions envisaged Increase the number of TELL40 → very expensive Increase the FPGA needed resources: Not sure that they would work Reduce the bandwidth by removing most of the empty-empty This should be a collaboration wide decision Is there any reason why it should not be implemented ? Calorimeter fibres A first meeting took place 10 days ago Discussion on the type of fibres to be used for the calorimeter Plan is to have a small distribution box at the level of each rack A proposal of architecture should be presented soon Pascal Perret - LPC Clermont 08/12/2015

32. 32  Front-end board Design is almost done The implementation of the 8-channel blocks is achieved From the connector, the ICECAL, the ADC, the FPGA to the GBTX Optical emitters, master FPGA, GBT-SCA, clock distribution, serializer (LLT) parts to be achieved soon. Still some pending issues: We would like to test that the power supplies can be easily adapted to our needs for the upgrade Plan is to use a spare PS Adapt and test it 2 prototypes FEB foreseen A test beam is planned for Spring 2016 Pascal Perret - LPC Clermont 08/12/2015 Electronics

33. 33 Summary  A lot of progress Recently the upgrade group performed a new test beam Check the performances of the analog electronics Noise study performed in the cavern Several meetings concerning Readout Fibre PS/SPD dismantling Tuesday meeting session on the extension of the ET range of the ECAL Agenda at https://indico.cern.ch/event/450022/ https://indico.cern.ch/event/450022/ Monday 7 dec. Improvements of the calorimeter design during the LS3 Pascal Perret - LPC Clermont 08/12/2015

34. 34 Summary & Conclusion  CALO detectors are running smoothly (even with Pb-Pb) Thanks to all the piquets who take part to the operation Online calibration at work: π 0 method + LED following allow good ECAL calibration 1-2% Few fills needed beginning of 2016 data taking period for a 1 st fast calibration HCAL (Cs +) LED Occupancy implemented and running; needs stable conditions YETS: a lot of activities (as usual)  Upgrade: A lot of progress 2 prototypes FEB expected beginning of 2016  THANK YOU! Pascal Perret - LPC Clermont 08/12/2015

35. 35 SPARES Pascal Perret - LPC Clermont 08/12/2015

36. 36 Pb-Pb event  Asymmetry in PS Pascal Perret - LPC Clermont

37. 37 Pb-Pb event  Asymmetry in PS Mostly disappeared after changing the L0TCK Pascal Perret - LPC Clermont

38. 38 ECAL ECAL LED instability issue The reason is understood: this is an unfortunately wrong choice of the LED model done in 2013. the chosen LED type (Multicomp Mega Bright OVL-5523) have very good light yield but low (lower than usual) reverse breakdown voltage (8-12 V) the design of the LED driver is such that between pulses the reverse voltage of up to ~15 V can be applied to the LED the lower is required flash magnitude, the higher is reverse voltage therefore the Outer area suffers most Now working on selecting a proper LED CREE LC503FBL1 is a good candidate: also good brightness, also cheap: ~ 0.15 euro/pc selection is necessary: only 60% have UBREAK>20V, minimum is ~12 V Total number of LEDs is 456 (however we can choose to replace LEDs only in the Outer zone – 168): 1000 LED ordered Pascal Perret - LPC Clermont 08/12/2015

39. 39 ECAL  Rate effect in ECAL in ECAL the rate effect is much lower and on average negative Pascal Perret - LPC Clermont 08/12/2015

40. 40 HCAL Cs calibration  Radiation damage of HCAL light yield Pascal Perret - LPC Clermont 08/12/2015 beam PMT spacers WLS fibers light guide master plate scintillators row 0 row 1 row 2 row 3 row 4 row 5 The hadronic shower maximum lays within the tile row 0 (ECAL is ~1.2 λ I ). The relative ageing rates of tile rows agree with the shower profile coefficients. Longitudinal shower profile 0 1 2 3 4 5 HCAL tile row relative ageing rates doses, TDR As at 137 Cs calibration response of every individual tile is measured, the radiation damage of the HCAL light yield of scintillator tiles and fibers in a tile row #i can be determined as a decrease of relative response of this row, (A i /A 5 ), with respect to a reference 137 Cs run at lumi=0:

41. 41 HCAL Cs calibration The HCAL light yield degradation plot, adding 2013 and 2015 137 Cs runs: 17/06, 02/09 & 11/11 Pascal Perret - LPC Clermont 08/12/2015 Fastest ageing at the startup Now the ageing continues at ~ same speed (not completely correct: in rad. dose, 1 pb -1 @13 TeV is equivalent to ~2 pb -1 @8 TeV) 2011 2012 TS1 TS2 TS3 2015 TS1 TS2 TS3

42. 42 HCAL Cs calibration For the unstable cells, the LED following interfered with the LHC intensity ramp: the reference was taken at low luminosity. Pascal Perret - LPC Clermont 08/12/2015 137 Cs calib ref start of LED following

43. 43 HCAL LED calibration  LED behaviour in 2015, the same 44 very central cells Pascal Perret - LPC Clermont 08/12/2015 Start: Cs calib Nov 2014 Cs calib 17-Jun-2015 Cs calib 17-Jun-2015, but new algorithm from this point on: weighting with shower profile Cs calib 02-Sep-2015

44. 44 ECAL Calibration  π 0 mass and L0DUElectron/L0DUMUON rate vs time Pascal Perret - LPC Clermont 08/12/2015 New TCK ECAL HV update

45. 45 ECAL LED calibration Pascal Perret - LPC Clermont 08/12/2015 The last π 0 calibration coefficients were received 19-Nov. Used data taken 08-Oct – 03-Nov. Demonstrated that there is no improvement in π 0 peak after calibration  LED correction works well. No LED-based correction done after TS3 (no significant degradation observed after 5 TeV run; then ions). ref 3 TS2TS3 ref2 π 0 calib 07-Oct ref1 π 0 calib 05-Sep special fills (elastic and diffractive), low lumi Pb-Pb, low lumi 5 TeV

46. 46 Discussion on the Et range in the ECAL reconstruction Having a similar precision at low energies (Et~100MeV, e.g. Bremstrahlung) AND at 60 GeV (Et) would most probably require a major redesign of our electronics Need probably a double gain system with 2 parallel channels The consequences would be Large delay (complete redesign) → difficult to imagine for 2019 Increase of the needed manpower (which is small at present) Increase of the cost Proposal From measurements → estimation of the effect of a gain reduction Ask some contact persons to evaluate the consequences A major modification looks difficult to justify for processes or measurements which can be/have been done by other experiments better or for which the justification comes from a very specific corner of the phase space. The upgrade conditions could be such that the low energy physics is difficult to do anyway and the noise contribution would not be the main penalty Pascal Perret - LPC Clermont 08/12/2015

47. 47 Discussion on the Et range in the ECAL reconstruction Proposal 1. Feel confident that we can do low energy physics and the resolution is degraded at low gain Keep the present configuration Pile-up contribution is such that other effects are negligible at low energies or the noise at low gain is better than expected 20 GeV Et range can be obtained rapidly just by re-configuring the boards (+threshold tuning) Another factor could be obtained with HV reduction for the middle and outer parts The inner part may require the usage of filters (to be checked soon). This could be done during a long shut-down. The filters could also be installed BEFORE run 3 start-up if the pile-up in this region is expected to be too large from the beginning. Notice that 1 (+/- inner filter) could be used up to 10 33 and 2. beyond Pascal Perret - LPC Clermont 08/12/2015