A flexible stand-alone testbench for characterizing the front-end electronics for the CMS Preshower detector under LHClike timing conditions Paul Aspell1, David Barney1, Yves Beaumont1, Suhas Borkar2, Aruna Borkar2, Jacques Domeniconi1, David Futyan1, Apollo Go3, Suresh Lalwani2, Carmen Palomares1, Remi Prunier4, Serge Reynaud1 1 CERN, Geneva, Switzerland Bhabha Atomic Research Centre, Mumbai, India 3 National Central University, Chung-Li, Taiwan 4 French Cooperant at CERN 2 LECC 2002, Colmar 8-13 September David Barney, CERN Overview of Presentation • • • • The CMS Preshower Requirements of the front-end electronics PACE-2 architecture Test setup – – – – Requirements Hardware Software Results • Future & perspectives LECC 2002, Colmar 8-13 September David Barney, CERN CMS Preshower – physics motivation Main physics goal of CMS is search for SM Higgs If mH < 150 GeV/c2 best chance is through 2g decay But large reducible background from p0 faking single photons Idea of Preshower: Single incident photon LECC 2002, Colmar 8-13 September Two closely-spaced incident photons David Barney, CERN CMS Preshower – physical location LECC 2002, Colmar 8-13 September David Barney, CERN CMS Preshower – mechanical layout 32 strips 63mm Front-end electronics 63mm LECC 2002, Colmar 8-13 September David Barney, CERN PACE-2 Architecture PACE-2 is an assembly of two chips in DMILL 0.8mm BiCMOS technology Delta • Preamp with leakage current compensation • Switched gain shaper. Low gain (0-400 MIPs) for physics, high gain (0-50 MIPs) for calibration • Programmable biasing, modes of operation and calibration pulse PACE-AM • 32 channels, 160 columns analogue pipeline. 3 samples per trigger into a 24-deep FIFO • Programmable biasing, latency, modes of operation • LVDS inputs for 40MHz clock, LV1, ReSynch etc. • I2C input for programming of both PACE-AM and Delta LECC 2002, Colmar 8-13 September David Barney, CERN PACE-2 MCM Hybrid containing PACE2 PACE AM 9.6 x 6.4 mm2 Delta 3.5 x 6 mm2 Most outputs from PACE2 are for test purposes only! LECC 2002, Colmar 8-13 September David Barney, CERN Design Verification Testbench – basic requirements • Flexible – Many digital and analogue functions to test – Capable of tracing the cause of possible malfunctions • Fast - need to simulate LHC-like running conditions – 40 MHz clock – Fast control signals for Delta and PACE-AM via LVDS – Multiple triggers (programmable) • Slow control – I2C interface for register loading etc. • DAQ system on-board • Cheap! Want to replace NIM/CAMAC/VME with a small standalone unit, controllable via LabView – Produce multiple copies of system for different labs, beam tests etc. • Prototype for Production Test system – need to evaluate ~7000 PACE for the Preshower LECC 2002, Colmar 8-13 September David Barney, CERN Motherboard Architecture LECC 2002, Colmar 8-13 September David Barney, CERN PACE Test Motherboard – side 1 Delay lines Mitsubishi M16C mcontroller RS232 PACE-2 Hybrid FIFOs (2 x 8k,8-bit) Altera FLEX 10k FPGA LECC 2002, Colmar 8-13 September AD9042 12-bit 40 MHz ADC David Barney, CERN Programming Environment • Multithreaded application running on M16C microcontroller – Text-string communication from PC via RS232 – Structure of command is: • s<task ID><bytecount><parameters>, e.g. S42D1 – read FLEX control register – User interface built using LabView 6i Task ID Function Task ID Function 0 Scan i2c address 11 Read ADC FIFO status 1 Write to i2c 12 Read 1 trigger block 2 Read from i2c 13 Write to DAC 3 Write to FLEX 14 Read software version 4 Read from FLEX 15 Toggle echo 5 Set one bit in FLEX 16 Set echo to ON 6 Clear one bit in FLEX 17 Set echo to OFF 7 Toggle one bit in FLEX 18 Clear M16C circular buffer 8 Write to delay line 19 Toggle offset to ASCII 9 Read from delay line 20 Set RS232 baud rate 10 General reset LECC 2002, Colmar 8-13 September 21-25 Various tasks for Poisson random trigger generation David Barney, CERN Modes of Operation • Motherboard as a “Master” with internal trigger generation – • Normal operation mode for design evaluation Motherboard as a “Master” with external trigger generation – • Useful for testing system with a silicon sensor attached, stimulated by an IR laser or a radioactive source Motherboard as a “slave” with external trigger generation – Useful to synchronize two motherboards together (e.g. in a beam test) LECC 2002, Colmar 8-13 September David Barney, CERN Internal (Motherboard) Triggering Modes • FLEX FPGA can be programmed to generate a burst of up to 16 triggers (“CalBurst”) by specifying the time (in # of clocks, up to 65535) between triggers • Each “trigger” is actually two pulses: – Calib – telling the PACE to generate an internal electronic injection signal – LV1 – sent <latency> clocks after the Calib • 6 different internal trigger modes: – Mode 0: Single ReSynch, single CalBurst – Mode 1: Free running ReSynch (specifying period), each followed by a single CalBurst – Mode 2: Single ReSynch, free running CalBursts (specifying period between CalBursts) – Mode 3: Free running ReSynch, free running CalBursts – Mode 4: Single ReSynch, CalBurst on demand (by toggling one bit in the FLEX control register) – Mode 5: Free running ReSynch, CalBurst on demand LECC 2002, Colmar 8-13 September David Barney, CERN FLEX FPGA Control Via LabView LECC 2002, Colmar 8-13 September David Barney, CERN PACE Register loading/reading LECC 2002, Colmar 8-13 September David Barney, CERN Design Verification – Digital Part • Test I2C addressing on PACE-AM – If this doesn’t work, stop testing • Load/verify registers on Delta and PACE-AM – Can also read the default register values at power-on • Put PACE into “Run” mode (set one bit in the PACE-AM control register) • Set internal trigger to mode 0 (single ReSynch, single CalBurst) • Examine basic digital signals via connectors on hybrids – – – – ADC FIFO empty flag Multiplexer signals Data-valid signal Loop signals (signify number of columns in the memory that are currently blocked) – Memory addresses of blocked columns (timing of trigger relative to a ReSynch can be used to determine a required column – and thus check that all columns are working) LECC 2002, Colmar 8-13 September David Barney, CERN Test of Skipping Mechanism 1. 2. 3. 4. Send N consecutive triggers to block a complete region in memory Send one further trigger such that 2 of the 3 cells are before the blocked region and the 3rd is after the blocked region Test to see if the memory addresses read-out are ok See how many cells can be skipped (before and after irradiation) Skip 3 cells ☺ Test 1 2 1 Test 2 3 1 2 Test 3 4 1 2 3 Test 4 5 1 2 3 4 Test 5 6 1 2 3 4 5 Test 6 7 1 2 3 4 5 6 Test 7 8 1 2 3 4 5 6 Skip 6 cells ☺ Skip 9 cells ☺ Skip 12 cells ☺ Skip 15 cells ☺ Skip 18 cells ☺ 7 Skip 21 cells ☹ Numbers refer to “trigger number” LECC 2002, Colmar 8-13 September David Barney, CERN Design Verification – DAC Optimization • First need to set biases/currents on Delta and PACEAM to default values DAC output (Volts) – Loop over possible DAC values (i.e. set the appropriate registers via I2C) and measure analogue outputs (from test pads) using the ADCs on the M16C – Use plots to determine best DAC settings and apply them 5 4 3 2 1 241 225 209 193 177 161 145 129 113 97 81 65 49 33 17 1 0 Digital value LECC 2002, Colmar 8-13 September David Barney, CERN Design Verification – Analogue Output • For one trigger the analogue output comprises 96 sequential values – multiplexed output for 32 channels for each of three time-samples • Can look at this with an oscilloscope or after the ADC using the PC Charge injection into channel 11, pedestals subtracted LECC 2002, Colmar 8-13 September David Barney, CERN Design Verification + Production Testing • Most remaining design verification tests concern the analogue performance – Injection seen in all channels? – Gain uniformity between channels – Pedestal uniformity through the memory and between channels – Dynamic range (in both gains) – Timing scan to determine signal shape after the preamp/shaper stage • Using the delay lines • All of these tests are also incorporated into our systematic testing procedure…… LECC 2002, Colmar 8-13 September David Barney, CERN Production Testing Front Panel LECC 2002, Colmar 8-13 September David Barney, CERN Testing charge injection on all channels LECC 2002, Colmar 8-13 September David Barney, CERN Timing curves – using all 3 samples Time sample 1 Time sample 3 LECC 2002, Colmar 8-13 September Time sample 2 Result David Barney, CERN Linearity – High Gain, High Precision ~ 0 mips ~ 7 mips ~11 mips ~16 mips LECC 2002, Colmar 8-13 September David Barney, CERN Output (ADC counts) Linearity Curves 4500 High Gain Low Gain High Gain Low Gain High prec. High prec. Low prec. Low prec. 4000 3500 3000 2500 2000 1500 1000 500 0 LECC 2002, Colmar 8-13 September David Barney, CERN Future Production Testing • At present to perform the complete set of tests on one PACE-2 assembly takes ~10 minutes – Speed limited by RS232 interface (115kbits/sec) – Next version of test system may use USB • Speed ultimately limited by settling time for some tests, and by time taken for changing the chip (final version PACE-3 (0.25 mm) will be a packaged assembly of the Delta and PACE-AM) – Perhaps will take ~2 minutes to test one assembly • Need to test ~7000 assemblies (assuming 60% yield) • Should note that currently we have ~65% yield with the DMILL version (~100 PACE-2 assemblies tested), and about half of the failures are due to problems with the I2C interface – In these cases the testing procedure is much shorter! LECC 2002, Colmar 8-13 September David Barney, CERN Other Uses of Testbench – Beam Test • In May 2002 two silicon sensors were placed in a proton/pion beam at PSI – Used PACE-2 as the front-end – Used our test bench as the control/DAQ system – Worked very well, although max trigger rate was only about 20 Hz (limited by ADC FIFOs on the motherboard and RS232 communication speed) LECC 2002, Colmar 8-13 September David Barney, CERN Other Uses of Testbench – Irradiations • Test digital/analogue performance in X-ray beam at CERN LECC 2002, Colmar 8-13 September David Barney, CERN Testing other chips • System is extremely flexible • Changing connectors and adapting programs in the M16C & FPGA (and the LabView user interface) could allow the testing of similar devices – e.g. SCT, APV….. LECC 2002, Colmar 8-13 September David Barney, CERN Summary and Conclusions • A testbench has been developed for the CMS Preshower front-end electronics (PACE) • It is sufficiently flexible to allow testing of all analogue and digital functionality • Software has been developed to allow systematic testing of large quantities of chips • Next version of the testbench will incorporate a higher-speed communication to a PC, allowing production testing of ~7000 PACE assemblies • Further information, including system documentation and schematics can be found on the CMS Preshower web site: – http://cmsdoc.cern.ch/cms/ECAL/preshower LECC 2002, Colmar 8-13 September David Barney, CERN Spare Slides LECC 2002, Colmar 8-13 September David Barney, CERN PACE-II Test Motherboard – side 2 Relays for switching input to ADCs on mcontroller LECC 2002, Colmar 8-13 September David Barney, CERN Preshower Electronics Chain LECC 2002, Colmar 8-13 September David Barney, CERN PACE-2 Signal shape etc. 32 channels 160 columns Sample 1 2 3 Time latency read pointer write pointer 3 samples per trigger (LV1) are stored in the memory for each of the 32 input channels. These data are multiplexed to an ADC at 20 MHz LECC 2002, Colmar 8-13 September David Barney, CERN