1. ECAL TIMING

2. 20/04/092 Ratios’ Method Basics Position of pulse maximum parameterized using the ratio of two consecutive samples, i.e., R = A(t)/A(t+1) R fitted with polynomial function Each ratio provides an independent extraction of pulse maximum position (T MAX ) Final T MAX as weighted (with expected error) average of the different measurements

3. Ultimate Resolution at Test Beam Use of high energy electrons at test beam Compared two neighboring crystals in the same electron cluster to reduce systematic effects and extract ultimate resolution of the algorithm Single crystal time resolution can be parameterized as a function of effective amplitude 1/A eff 2 = 1/A 1 2 + 1/A 2 2 Constant term ~0.1ns both for endcap and barrel PLOT: y-axis: time difference spread (sigma of a Gaussian from fit) between two neighboring crystals. x-axis: amplitude over noise (or crystal energy) noise term constant term Endcap

4. Impact of Time Synch. On Resolution Impact of offline synchronization between different channels verified with test beam electrons Compare two neighboring crystals in the same electron cluster Resolution from the difference of time measurements with and without synchronization of crystals O(1ns) improvement. Optimal hardware synchronization together with offline synchronization crucial to reach ultimate performance in time measurement PLOT. Time difference between two consecutive crystals fitted with a Gaussian After synchronizing (7.5-10) GeV testbeam electrons  =0.75ns  =0.21ns

5. Linearity and Resolution with Cosmics Check of time measurement linearity using two clusters associated to same muon (top-bottom). Expected time of flight compared with measured time. Compared time measurement between crystals in the same muon cluster. Offline detector synchronization not yet taken into account. Photodetector gain x4 the LHC conditions. Time measurement performance verified with full ECAL barrel detector in CMS using cosmics. Cosmic clusters selection refined associating muon track to cluster position. y-axis: time diff. spread (sigma of a gaussian from fit) between two crystals. x-axis: amplitude over noise (or energy of crystal). y-axis: mean of measured time difference. x-axis: expected time of flight of the muon between the two cluster positions Barrel

6. +Beam-Beam muons Time measurement is verified with full ECAL detector in CMS using muons from beam splash events. ‘Nominal’ is the expected ECAL readout schema, which reads out the detector based on the assumption of time of flight from the interaction point (0,0,0); ‘nominal’ also includes the plane wave assumption of muons traversing ECAL. Time synchronization is derived from laser pulses which are delivered by fibers of the same length within a module. This produces the structure with a modularity of 4-5  intervals. These measured variations have been used to synchronize the post-splashes data-taking. ECAL Splash Timing, η profiles PLOT: mean of timing for crystals in the same  ring versus  index. Left: beam coming from negative direction. Right: beam coming from positive direction. Reconstructed time after correcting for inter-supermodules phases (red) and the expected ECAL time (blue) are shown.

7. Time measurement is verified with full ECAL barrel in CMS using muons from beam splash events. ‘Nominal’ is the expected ECAL readout delay which reads out the detector based on an assumed time of flight from the interaction point (0,0,0); ‘nominal’ also includes the plane wave assumption of muons traversing ECAL. Time synchronization is derived from laser pulses which are delivered by fibers of the same length within a module. This produces structure with a modularity of 4-5  intervals. ECAL Splash Timing, η profiles PLOT. y-axis: residuals of timing for crystals in the same  ring with respect to the expected time Red: beam coming from positive direction. Red: beam coming from positive direction.

8. PLOT. y-axis: mean of timing for crystals with the same FED #. x-axis: FED #. Red: beam coming from negative direction. Black: beam coming from positive direction. Time measurement is verified with full ECAL detector in CMS using muons from beam splash events. Here the variations are shown as a function of the FED number (one FED is a readout unit: a supermodule in EB or a sector in EE). O(2ns) variations are expected from length of the readout lines. EE+ (FED: 646-654) was expected to have larger uncertainty for some towers which were timed with HV off. Also the beam from the positive and negative  side are compared showing a small effect (<0.5ns). These measured variations have been used to synchronize the post-splash data-taking. ECAL Splash Timing, FED profile

9. Ratio Method PLOT. Spread of the difference of crystal times from the expected time, as a function of amplitude over noise (or energy of crystal). The expected time is taken from the time-of-. flight, assuming that the particles hit the middle of each crystal. Fit vs. Ratios: Beam Splashes in the Barrel

10. SPLASHES IN EE WITH INTERCALIBRATED CRYSTALS

11. ECAL Endcaps response to beam splashes Average energy per crystal in the ECAL Endcaps seen during “beam splashes” collected with beams coming from the EE- side (top) and the EE+ side (bottom). Intercalibration constants are applied White regions are crystals masked in the readout Energy modulations are a combination of the energy flow investing CMS and geometry effects. In particular, the lower energy at large radii in the ECAL Endcap downstream to the beam direction is due to the ECAL barrel shield

12. STABILITY

13. ECAL response sensitive to variations of: Crystal transparency (under irradiation) Temperature: ∂(LY)/∂T, 1/M(∂M/∂T) ~ -2%/K High voltage: 1/M(∂M/∂V) ~ 3%/V Controls and monitoring: Controlled (temperature, CAEN, dark current measurements) ECAL response monitored and corrected with laser data Required performances: Temperature stability at the few 0.01 o C level HV stability at the 10 mV level Laser monitoring of ECAL response at the 2‰ level Introduction LY: light yield M: APD gain

14. stability map in the ( ,  ) plane for barrel laser data RMS (%)   Most channels have a measured response stability below the 1‰ A set of reference channels APD ref is used to normalize the event-by-event laser amplitude variations. One reference APD channel is chosen arbitrarily for each laser monitoring module (100 or 200 channels) For each channel and each laser sequence (600 laser events), the average is employed as monitoring variable “Stability” is defined as the RMS over all laser sequences of normalized Stabilities are computed for each channel on a period with stable laser conditions during the CRAFT (from 60 to 100 laser sequences within 300 hours) APD ref is chosen as a reference because of readout problems with PN reference diodes, which are being fixed White regions either lack statistics (2 supermodules not readout).

15. Mean = 0.3 ‰ RMS = 0.2 ‰ Under stable laser conditions, the ECAL LASER monitoring system is able to monitor the crystal response with a precision < 1‰ This precision is consistent with specifications (2‰) needed to achieve the ECAL design resolution stability for barrel laser data (projection of the previous stability map: one entry per channel) 99.6% of channels with RMS<1‰ 99.9% of channels with RMS<2‰

16. Reminder: Analysis of ECAL Barrel temperatures using 2 independent set of data One crystal out of 10 is equipped with a thermistor, in close thermal contact. The thermistors are read out by detector control units (DCU) Every supermodule is equipped with 10 precision temperature probes (PTM), 2 mounted on the in/out water outlets, and 8 on the mechanical structures (4 on grid, 4 on thermal screen) Hereafter temperatures from the thermistors and the PTM’s are used to assess stability Nominal sensitivity: PTM sensors: 0.01 °C - DCU measurements: 0.012 °C Thermistors fine calibration has been computed by comparing DCU measurements to PTM sensors values, using data from temperature controlled COSMIC Stand. EB-16, documented replaced VFEs, channels with missing calibrations have been re-calibrated with the same technique using P5 data.

17. Map of ECAL Barrel instantaneous temperatures ( °C) as measured by the thermistors located on the APD capsules. In black: 37 missing measurements (broken or not calibrated thermistors). White spots: EB+7 (inactive), 34 thermistors outside range. Temperature spread is within 0.2 °C for 96% of the channels. Thermistors located in ECAL outer borders are on average 0.08 °C warmer. ECAL Barrel temperature map

18. ECAL Barrel Temperatures time evolution during CRaFT, averaging over all the thermistors and PTM probes: thermistors values (black), PTM9 sensors on input cooling water pipes (purple) PTM1 sensors located on the grid of type 1 modules (red), PTM4 sensors located on the grid of type 4 module (green) ECAL Barrel temperature stability

19. ECAL Barrel temperature spread during CRaFT Top: ECAL Barrel Temperatures distribution, measured at the APD capsules level (DCU system). The average temperature is 18.12 ± 0.04 °C, which shows a very good overall temperature stability and homogeneity. In blue, superimposed, are the temperatures belonging to the outer borders of the barrel. The average temperature is 0.09 °C higher than the rest of the Barrel. PTM sensors APD capsules Bottom: ECAL Barrel Temperatures distribution, measured at the SM grid level by the PTM system. The average temperature is 18.1 ± 0.02 °C, in very good agreement with the one measured by the DCU system.

20. Distribution of temperatures RMSs for each of the ECAL Barrel thermistors during CRaFT (logaritmic scale). Missing thermistors are: the broken/not calibrated (37), few masked during CRaFT (LV problems). Peak average RMS is 0.009 ± 0.003 °C. 98% of channels have RMS<0.1 °C. Values in the right-end side tail are channels affected by some read-out problems. APD capsules temperature RMS