Commissioning and operation of the CDF Silicon detector

Transcription

1 Commissioning and operation of the CDF Silicon detector Saverio D Auria On behalf of the CDF collaboration International conference on Particle Physics and Advanced Technology, Como, Italy, October 2001 Como Int. Conf., 15/10/2001 p. 1

2 Commissioning and Operation of the CDF Silicon detector Outline: Introduction Installation Commissioning Integration and Operations First Data Como Int. Conf., 15/10/2001 p. 2

3 Introduction: CDF Silicon Detector = 3 subsystems: L00, SVX, ISL strips, 5644 chips, 5.8 m 2 of Silicon. 1 L00 tube, 1 layer, 3 SVX barrels in z with 5 layers each, 1 ISL space tube with 1 central and 2 outer layers All sub-systems use: the same chip The same control card ( portcard ) Same DAQ system. Slightly different power supplies. Cross section Beam Pipe Como Int. Conf., 15/10/2001 p. 3

4 Detector description Layer Inner/Outer Radii [cm] Axial Pitch [m] Stereo Angle Stereo Pitch [m] / / / / / / Forward 19.7/ Central 22.6/ Forward 28.6/ modules, RO channels Layer 00 silicon detector is 15 times larger than in Run1! Como Int. Conf., 15/10/2001 p. 4

5 The SVX-3D chip overview The SVX-3D chip 128 channels 42 cell analog pipeline Double correlated sample Dynamical common-mode noise rejection. (aka Dynamical Pedestal Subtraction) Deadtimeless operation Data sparsification with threshold, Programmable 197 bits Bandwidth, Signal polarity, Readout mode (all, sparse, NN) Pedestal mode Calibration mask. Como Int. Conf., 15/10/2001 p. 5

6 The readout system See Mary Bishai s talk. Como Int. Conf., 15/10/2001 p. 6

7 Wire bonds in the system: Non critical bonds (single channels): (strips) Chip critical bonds: (chip wire bonds) Ladder (module) critical bonds: (bias lines, control lines, optical links) Wedge-critical: 816 (control lines) Como Int. Conf., 15/10/2001 p. 7

8 Installation All components (ladders) have been extensively tested before assembly. Quality control tests passed ladders with no readout error and less than 2% noisy or disconnected channels. During assembly a second test procedure was implemented: Electrical tests Functional tests About 10 % of the ladders developed defects: replaced or repaired. Excess noise developed on Small-angle stereo detectors ( grassy ladders ): 11 ladders out of 360 have been found to have additional number of noisy channels in excess of 2%. Origin is buildup of oxide charge. A number of potential problems turned out to be no problem at all: Cross-talk between ladders: needed no action: readout was not influenced. System grounding was reasonably good at first try. Como Int. Conf., 15/10/2001 p. 8

9 Installation: noisy ladders Pedestal # before Noise ADC After ADC Pedestal and noise vs. strip number after assembly It only affects layer 2 and 4 Noise distribution for a single chip before and after development of grassy-ness Como Int. Conf., 15/10/2001 p. 9

10 Tests before installation All portcards were functionally tested before installation: Test of functionality (complete readout) against a limited set of available receivers (OFTM) performed at 15 C All FIBs and Optical Fiber Transition Module (OFTM) were tested on bench with a Data Emulator card and a limited number of transmitters All power Supplies were tested with passive load and burned-in All installation cables have been tested. Como Int. Conf., 15/10/2001 p. 10

11 Commissioning/Test Due to accelerator schedule there was no time to test before commissioning. Test and Commissioning occurred at the same time. Initial cabling, then we started commissioning a fraction of detector, then cabling and testing finished during 1 month shutdown. Collision-Hall tests: Power supply and voltage cable tests with passive load - Correct voltages are on the right pin - Correct currents - Test of overload protections and interlocks Functional tests with test devices (Wedge in a box). Cabling the real wedge, full functional test, debug. Como Int. Conf., 15/10/2001 p. 11

12 Wedge-in-a-box 5 ladders + 1 portcard Solid-state-cooling Como Int. Conf., 15/10/2001 p. 12

13 Commissioning/Test 3 Voltage cables, 5 Optical cables, 3 command cables (out of 114) had to be replaced. 35 power supplies out of 102 had developed problems and were replaced/repaired. 3 FIB modules (out of 57) had to be replaced A large number of OFTM had to be replaced/reseated. Main Difficulties: Optical link to front-end: Rx/Tx light level mismatch. Power supply delivery time: being received while commissioning. Cooling interlock system stability and tuning, especially annoying during detector debugging phase: the interlock system was being commissioned at the same time. Lesson learned: Functional tests may be not adequate if components have to be matched (e.g. Rx/Tx) and some crucial component parameter has not a large safety margin. Unexpected no problem: No big interaction with the accelerator RF, noise about the same as at fabrication (SVX, ISL), no interference with the Central Chamber (COT). Grounding reasonably good at first try, needed only small adjustment. Como Int. Conf., 15/10/2001 p. 13

14 Installation Plugging crew: ~30 students and postdocs ~7weeks, 24h/day, 4 people at the time Como Int. Conf., 15/10/2001 p. 14

15 Commissioning/Tests Test Results One whole wedge can not be read out: one vital wire bond broke during transportation/installation (PRD2-PRD2* shorted). Lesson learned: Minimize handling and transportation of sealed detector (if possible!), Allow for redundant (vital) interconnections. Two wedges are under investigation, probably bad cable/connections. Part of a ladder now has dead front-end amplifier. ISL has part of the cooling channels blocked (see F. Palmonari s talk). Not commissioned yet, but all ladders functionally tested during cabling. Como Int. Conf., 15/10/2001 p. 15

16 Commissioning/Integration After commissioning test we performed integration tests: Single-wedge test in stand-alone mode, check for readout errors. Ladders that pass this test are integrated, and are read out with the CDF daq framework. Then the set of integrated ladders are passed to the CDF DAQ Run Control. We integrated and operated 70% of SVX, 35% of ISL due to difficulties with cooling. Main difficulties encountered: Power Supply communication errors: modules spontaneously turning on and off. Mismatch of light output getting worst with cold operation of DOIMs residual infant mortality of FIBs, PS, crates. Managing a system configuration that changes frequently: - Synchronization between daq database and power supply - Synchronization between data and calibration configurations. Interlock tuning: preventing false alarms while keeping the detector safe Como Int. Conf., 15/10/2001 p. 16

17 Operations Luminosity to-date: 4.5 pb -1 Physics quality on tape. Silicon Detector operated in DPS-on mode at fixed (low) threshold, no optimization of chip parameters has been applied yet. Common mode noise is not an issue (SVX, ISL). Calibration: Data-mode (Intrinsically Slow Feedback): Data taken, analyzed off-line, constant table made of - noise, - pedestal, - noisy and dead channels calibration constants are loaded in database and associated with a set of runs. Calibration: X-mode (Fast Feedback): Analysis takes place at the VME crate. Noise, pedestals and noisy and dead channel lists are read-out and loaded in the database. Como Int. Conf., 15/10/2001 p. 17

18 Data Quality Monitoring On-line checks running on-line consumers on a subset of data. On-line plots with occupancy, cluster charge, track distribution. On-line check of Silicon functionality and error rate Off-line data quality monitoring on a larger data set, using the right calibration constants. Landau peak monitoring, (extremely useful) Occupancy monitoring (not much looked at, yet) p/n side pulse amplitude ratio monitoring. Automatic detection of mis-behaving modules: parameters used are KS test, Landau fit, Number of clusters on track/number of all clusters Como Int. Conf., 15/10/2001 p. 18

19 Commissioning/Data SVX, before optimization Como Int. Conf., 15/10/2001 p. 19

20 Commissioning/Data First task was to synchronize the Silicon clock with the beam crossing: maximize signal vs. delay at fixed BW. All detectors show peak at the same delay value. ISL SVX n-side 90 o p-side n-side Sas Como Int. Conf., 15/10/2001 p. 20

21 Commissioning/Data First task was to synchronize the Silicon clock with the beam crossing: maximize signal vs. delay at fixed BW. All detectors show peak at the same delay value. Como Int. Conf., 15/10/2001 p. 21

22 ON-Line Landau plots Como Int. Conf., 15/10/2001 p. 22

23 Signal Summary L0 L1 Tuning the detector parameters for the best S/N L2 L3 L4 Unbiased Distribution of the peak positions of the charge distribution for individual ladders Red is n-side, Blue is p-side Como Int. Conf., 15/10/2001 p. 23

24 First Physics applications. J/O with Silicon track Como Int. Conf., 15/10/2001 p. 24

25 Conclusions The CDF Silicon tracker is well on his way to be fully commissioned. 70% of SVX and 35 % of ISL integrated with CDF. We already make use of Silicon for physics calibration studies. A 6-week shutdown occurring now will allow us to address the remaining difficulties. We shall have the detector fully operational for Physics by January Como Int. Conf., 15/10/2001 p. 25