GLAST LAT Project March 24, 2003 GLAST Large Area Telescope: Gamma-ray Large Area Space Telescope Tracker Subsystem WBS 4.1.4 2A: Electronics Design and Status Robert Johnson Santa Cruz Institute for Particle Physics University of California at Santa Cruz Tracker Subsystem Manager johnson@scipp.ucsc.edu 2A Tracker Peer Review, WBS 4.1.4 1 GLAST LAT Project March 24, 2003 TKR Electronics Requirements • • 2A Details of the requirements are in released LAT documents: – LAT-SS-17 Level-3 Performance requirements – LAT-SS-134 Level-4 Mechanical & Thermal requirements – LAT-SS-152 Level-4 Electronics requirements The major challenges are – Low power: <168 W of conditioned power • Less than 0.29 W per MCM or 190 W/channel • The flight design achieves 0.25 W per MCM! – Low noise occupancy: (noise trigger rate <500 Hz) • The trigger requires occupancy less than 5/100,000 ch/trigger • Readout and onboard processing requires <1/10,000 ch/trigger • The beam-test/balloon flight tracker achieved much better… – Compact packaging: bring signals around the tray corner – Manufacturing and QC: 884,736 channels, >98% functional – Reliability: design, redundancy, testing Tracker Peer Review, WBS 4.1.4 2 GLAST LAT Project March 24, 2003 Some Detailed Requirements • Internal charge injection for calibration and test, with pulseheight control and arbitrary selection of channels. • Threshold uniformity: <15 mV rms across 64 channels. • Threshold control per GTFE chip by an internal DAC. • Dead time & Readout speed: less than 10% at 10 kHz cosmic rate. • TOT: measure up to 4 MIPs. • Non-destructive readback of configuration registers. • Error checking: event alignment; command & data parity. • Layer-OR jitter <250 ns for a charge deposit > 0.5 MIP. • Reliability: – Redundant readout paths. – Redundant power paths. – Protection against power shorts. • Radiation hardness: 4 kRad TID; >37 MeV/mg/cm2 SEL thresh • Passive cooling; temperature monitoring. 2A Tracker Peer Review, WBS 4.1.4 3 GLAST LAT Project March 24, 2003 Tracker Readout Architecture Emphasis on compactness, minimum of wiring, and redundancy: • • • • Serial, LVDS readout and control lines on flat flex-circuit cables. Either of the two communications cables can fail without affecting the other. Two readout and control paths for every 64-channel front-end chip. Any single chip can fail without preventing the readout of any other. 24 64-channel amplifier-discriminator chips for each detector layer • Trigger output = OR of all 1536 channels in a layer. GTRC Control signal flow GTRC Control signal flow GTRC • Read command moves data into 1 of 2 GTRC buffers. • Token moves data from GTRCs to TEM. GT FE GT FE GTRC Nine detector layers are read out on each side of each tower. Data flow to FPGA on DAQ TEM board. 2A GTRC GTRC 2 readout controller chips for each layer 9-99 8509A22 Tracker Peer Review, WBS 4.1.4 Data flow to FPGA on DAQ TEM board. 4 Control signal flow • Upon trigger (6-fold coincidence) data are latched into a 4event-deep buffer in each front-end chip. Data flow GLAST LAT Project March 24, 2003 A2 A1 A0 NSDATA_IN A3 TREQ_IN LEFT GTFE RD_IN NTREQ GTRC CTRLREG TACKB NSCMD SCMD_OUT A2 A1 A0 CLKB RESETB NSDATA_IN A3 NTOKEN_OUT NRESET TOKEN CLK NSDATA NTACK TREQ_IN LEFT GTFE RD_IN NTREQ GTRC CTRLREG TACKB NSCMD SCMD_OUT CLK NRESET NSDATA NTACK TOKEN • Block diagram of the ends of two readout layers and their connections to the TEM – Clock, Command, Trigger, and Reset are bussed to the GTRC chips – Token and Data daisy-chain up and down the 9 layers – Each layers sends its Layer-OR directly to the TEM – The TEM communicates only with the GTRC chips, always by serial LVDS. – The GTRC communicates with 24 GTFE chips on the MCM. NTOKEN_OUT Tracker Readout Architecture CLKB RESETO RESETB TEM 2A Tracker Peer Review, WBS 4.1.4 5 GLAST LAT Project March 24, 2003 MCM Readout Module Configuration • • • • • • • Detector 8-layer polyimide PWB Top edge thickened and machined to a 0.64 mm radius Bias circuit 1-layer flex circuit (“pitch adapter”) bonded over the radius Tray Structure Fully encapsulated wire bonds Conformal coating High-thermal 2 Omnetics nano connectors conductivity transfer adhesive Steel mounting screws + adhesive GTRC, 1 of 2 Machined corner radius with flex circuit bonded around the curve Readout IC TMCM, attached by screws GTFE, 1 of 24 359.0mm 24.58mm 18.0mm Connector, 1 of 2 Grounding Screws 3 Total 2A Mounting Screws, 1 of 8 Tracker Peer Review, WBS 4.1.4 6 GLAST LAT Project March 24, 2003 MCM PWB Layout Concept • Low-noise environment for the amplifier chips • 8 layers are used for good analog/digital separation plus low-impedance ground and power planes. • Analog and digital traces are well separated and never wrap around each other. • Power planes and split analog/digital GND planes • Complete planes are also used for the SSD bias and AVDD2low impedance path to the decoupling caps for the SSD signal return. 1. 2. 3. 4. 5. 6. 7. 8. Digital traces; SMT parts; ASICs Digital busses Split digital/analog power Split digital/analog ground Analog ground Analog traces from ASICs to the SMT parts AVDD2 (analog 1.5V) Detector bias Stackup and arrangement of conductors in the PWB R, C, etc. IC Digital Traces Digital Pwr/Gnd Analog Pwr Analog Gnd Analog Traces and Planes 2A Tracker Peer Review, WBS 4.1.4 7 GLAST LAT Project March 24, 2003 Pitch Adapter • 1-layer Kapton flex circuit • Ni + Au plating for wire bonding • Precision tooling holes (not shown) • Circuit & traces are trimmed to length after bonding to the PWB (see Presentation 6E) Bias HV ASIC Side SSD Side (228 m pitch) “ground” 2A Tracker Peer Review, WBS 4.1.4 8 GLAST LAT Project March 24, 2003 F.E. Readout Chip (GTFE) • • • Schematics-based design, using standard cells for logic. Standard-cell I/O pad ESD protection. Manual layout of the analog channel, I/O cells, memory, global routing. • • Automated place-and-route of the logic blocks. Design verification: Spice and gate level simulations; DRC; LVS; simulation of the final extracted netlist in Nanosim. Trigger and Data mask registers Standard-cell auto route Control logic, command decoders Standard-cell auto route Calibration mask and capacitors 4-deep event memory (addressed by TEM) Custom layout 2 custom DACs Cap 64 amplifier-discriminator channels. 2A I/O pads and protection structures Tracker Peer Review, WBS 4.1.4 9 GLAST LAT Project March 24, 2003 GTFE Block Diagram 64 • 64 amplifier-discriminator channels • 7-bit threshold DAC • Calibration mask register • 7-bit calibration DAC • Trigger mask register and trigger layer-OR • Data mask register • 4-deep event buffer • Pair of redundant command decoders • Pair of redundant trigger receivers • Leftward readout register • Rightward readout register • LVDS I/O cells ANALOG INPUTS FROM SILICON STRIPS CALIB. MASK & READOUT REGISTER CALIBRATE STROBE PREAMPLIFIERS CALIB DAC SHAPERS THRESH DAC DISCRIMINATORS M U X 64 7 7 CTRLREG (TRI STATE) TRIGGER MASK & READOUT REGISTER S ADDRESSED SEL DAC REGISTER EVENT_TRIG FAST-OR 64 TRIGGER FROM PREVIOUS GTFE 1 64 DATA MASK & READOUT REGISTER WRITE STROBE L1T READ CMD WRITE CONTROL 2 EVENT_DATA 64 64 MODE REG EVENT_OR 2 1 LEFT/RIGHT D C Q D C Q DEAF MODE W_ADDR 67 X 4 EVENT BUFFER 2 R_ADDR DATA HIT READ CONTROL 64 LOAD DATA READ SHIFT REGISTER DATA FROM PREVIOUS GTFE A DATA OUT OUT B S DATA HIT LEFT COMMAND SELECT CMDR CMDL CLKL LOAD/READ TRIGGER MASK CALIBRATE STROBE LOAD/READ DAC REGISTER LOAD/READ MODE REGISTER LOAD/READ CALIBRATE MASK LEFT COMMAND DECODER RIGHT COMMAND DECODER CLKR RESET 2A Tracker Peer Review, WBS 4.1.4 10 GLAST LAT Project March 24, 2003 Readout Controller Chip (GTRC) • All digital • Tanner standard-cells, except for • LVDS I/O cells. • SEU hardened configuration register. • RAM (64 hits, 2 buffers) • Design in VHDL; synthesis, auto place and route. • Verification: VHDL sim; DRC; LVS; Nanosim of extracted netlist. NTOKEN_OUT NSDATA_IN I/O EVENT MEMORY 0 EVENT READOUT CONTROL EVENT MEMORY 1 GTFE READOUT CONTROL RD_IN TOT NTREQ FASTOR I/O TREQ_IN I/O NTACK TRIGGER TACKB REGISTER READBACK CONTROL CTRLREG NSCMD SCMDOUT CLK CLKB CMD DECODER NRESET CONTROL REGISTER RESETB I/O NSDATA TOKEN 2A Tracker Peer Review, WBS 4.1.4 11 GLAST LAT Project March 24, 2003 Readout Cables • Standard-technology 4-layer Kapton flex circuits • 8 unique layouts (with identical schematics) • Require 36-inch panels for manufacture • Power traces/planes, LVDS signals, and thermistor loops • Procurement specification: LAT-PS-01132 Thermistor Location Solder pads for nanoconnector 4 Termination resistors Thru-holes for Micro-D connector EM cable manufactured by Parlex 2A Tracker Peer Review, WBS 4.1.4 12 GLAST LAT Project March 24, 2003 Electronics Cooling • Low IC power density (20 MHz clock): – GTFE chip: 7.8 mW for 0.33 cm2 0.024 W/ cm2 – GTRC chip: 32 mW for 0.12 cm2 0.26 W/cm2 • Total MCM power of 0.25 W is spread by the PWB over about 100 cm2 along the tray edge 2.4 mW/cm2. The 8-layer board has several full copper planes to spread the heat. • A 5-mil 3M high-thermal-conductivity transfer adhesive lies between the MCM and carbon-carbon tray closeout. • The carbon-carbon carries the heat to the sidewall through a 20 cm long boss and 10 fasteners into the tower sidewalls. • The 1.5 mm thick K13D/YS90 sidewalls carry the heat to the bottom of the tower. • Copper straps carry heat from the sidewalls into the Grid. • SSDs stay below 30°C operational. The ICs will be only a few degrees warmer. 2A Tracker Peer Review, WBS 4.1.4 13 GLAST LAT Project March 24, 2003 Example EM Tray • The bias circuit is a 2-layer Kapton flex circuit. • Top layer: 16 wire-bond pads (Ni-Au plating) + HV bias traces and bulls-eye pads for conductive adhesive. • Bottom layer: hatched “ground” plane. Encapsulated ASICs Bias Circuit Handle for assembly fixtures MCM Thermal Boss Connector Saver 2A Tracker Peer Review, WBS 4.1.4 14 GLAST LAT Project March 24, 2003 Grounding & Shielding LAT RF SHIELD SPACECRAFT DIST-GLT ASSY. 28 VOLTS MIL-1553 POWER ASSY. SIU FILTER BOX ENCLOSURE SSR FILTER FILTER T HSKP GRID COMMON EPU FILTER FILTER FILTER SPACECRAFT GROUND SUBSYSTEM COMMON LAT Overview TEM ASSY. TKR/CAL_FE FILTER FILTER ACD ASSY FILTER BONDING STRAP ( ONE PLACE ) 2A GRID OUTER SHIELD FILTER ACD_FE CABLE Tracker Peer Review, WBS 4.1.4 15 GLAST LAT Project March 24, 2003 Tracker Grounding & Shielding Aluminum covering on all 6 sides Conductive tape on joints Zillion screws tie the sidewalls into the trays Cu thermal straps provide the conduction path to the Grid 2A 8 cables ground the 19 trays & electronics to the TEM Tracker Peer Review, WBS 4.1.4 16 GLAST LAT Project March 24, 2003 Tray Grounding & Shielding • The bias circuit includes a hatched “gnd” plane (actually 1.5 V) that corresponds to the amplifier voltage reference. It separates the SSDs from the structure and couples closely to the 120 V bias on the backs of the SSDs. • The MCMs have separate analog and digital ground planes with separate returns to the TEM, but they are coupled together on each MCM by an SMT jumper. • 3 long screws tie each MCM ground into the aluminum core Multi-Chip Module (MCM) C.C.F. Close-Out Aluminum HexCell C.F. Laminate Bias Plane Silicon Strip Detectors Mounting Screws Wire bonds & Right Angle Interconnect 2A Tracker Peer Review, WBS 4.1.4 17 GLAST LAT Project March 24, 2003 Tray Grounding & Shielding • The 1.5 V analog supply (AVDDA) feeds the source of the input FET and is thus the small-signal reference of the detector system. • It couples to the SSD bias via HV capacitors on the MCM and through the capacitance of the bias circuit. • This provides the small-signal current return path of the detector system. • All voltage supplies occupy planes on the MCM and are coupled to ground via numerous ceramic and tantalum capacitors on the MCM. 2A Tracker Peer Review, WBS 4.1.4 18 GLAST LAT Project March 24, 2003 EMI/EMC • The Tracker will be well shielded: – All transmitted signals are LVDS and digital (very low radiation and excellent noise rejection). In addition, power and ground reference planes are always directly under or over the signal pairs. – Aluminum foil (over carbon-fiber) covering all 6 tower module sides. – Conductive tape around the corners to connect the sides. • SSD strips are the sensitive nodes, but – They are well shielded from any radiation. – Only a very local reference is needed (the amplifiers are millimeters from the strips with well identified, short current return paths). – The local grounding around the SSDs is critical for noise performance. • EM emissions will be tested from the qual unit, but we expect it to satisfy requirements easily (433-RQMT-0005). Preliminary measurements on the BTEM showed no measurable emission, even with the aluminum shielding walls removed. 2A Tracker Peer Review, WBS 4.1.4 19 GLAST LAT Project March 24, 2003 EMI/EMC • Primary document 433-RQMT-005 • Radiated Emissions (RE101, RE102) – 20% of total emissions are allocated to subsystems outside LAT RF shield • ACD, Tracker, Heaters – 60% of total emissions are allocated to subsystems inside LAT shield • Radiated Susceptibility (RS101,RS103) – All subsystems must meet Section 5.3 of 433-RQMT-005 • Conducted Emissions (CE101, CECM) – Only the T&DF subsystem is affected and must meet requirements • Conducted Susceptibility (CS101, CS116) – Only the T&DF subsystem is affected and must meet requirements 2A Tracker Peer Review, WBS 4.1.4 20 GLAST LAT Project March 24, 2003 BTEM EMI Test No measurable EMI detected, clock on or off, even with the tower shield removed. Spectrum Analyzer BTEM with Shield removed Electric field antenna Magnetic field antenna 2A Tracker Peer Review, WBS 4.1.4 21 GLAST LAT Project March 24, 2003 Prototype Electronics Performance • See LAT-TD-1090. Reviewed at TKR ASIC review Dec 6, 2002. • Analog tests with “mini-MCM” plus full-length ladder (384 channels) – Gain and noise from charge-inject/threshold scans – Noise measurements from trigger-rate threshold scans – Noise occupancy from random triggers – Noise injection from digital readout – Gain versus number of channels pulsed – Pulse shapes and Time-Over-Threshold • Functional tests with full MCMs (no SSDs attached) – All digital functionality – Power consumption See Presentation 3B for – Voltage and timing margins more recent results on – DAC calibrations Engineering-Model trays. – Thermal cycling • Radiation Testing 2A Tracker Peer Review, WBS 4.1.4 22 GLAST LAT Project March 24, 2003 Threshold Dispersion The RMS dispersion, with or without SSDs connected, is below 8 mV, better than the requirements. 100 80 Hit Efficiency The dispersion is independent of the number of channels simultaneously pulsed. G Chip, with 32 Channels Pulsed Simultaneously 60 No load on inputs. 40 20 0 0 10 20 30 40 50 60 Threshold DAC Setting 2A Tracker Peer Review, WBS 4.1.4 23 GLAST LAT Project March 24, 2003 Trigger Rates Trigger Rate per Chip (64 Channels) 106 Rate (Hz) 10 5 104 1.3 fC=1/4 MIP 103 102 Three ladders G0 connected G1 to mini-MCMs G2 were tested. F3 These are the F21 results for F22 Ladder-0 Long TOT pulses; look like cosmic rays 101 100 0 1 2 3 Threshold/Gain (fC) 2A Tracker Peer Review, WBS 4.1.4 24 GLAST LAT Project March 24, 2003 Noise Occupancy Our first direct measurement of noise occupancy with the new electronics system. Ladder 0, 80V bias, 1.3 fC threshold, 1.5 million triggers. Noise Hits 6 4 2 0 0 100 200 300 Channel Number 2A Tracker Peer Review, WBS 4.1.4 25 GLAST LAT Project March 24, 2003 MCM Power Consumption AVDDA 1.5 V AVDDB 2.5 V No Clock 77.4 mW 39.3 mW 20 MHz 77.4 mW 39.3 mW DVDD 2.5 V address 5 82.8 mW 134.3 mW MCM Power (W) 2A DVDD 2.5 V address 0 Allocation MCM Address > 0 0.251 MCM Address = 0 0.255 Tower 9.05 10.5 W 16 Towers 145 168 W Tracker Peer Review, WBS 4.1.4 138.5 mW 26 GLAST LAT Project March 24, 2003 ASIC Review Action Items • • • • • • • • • • 2A Test full trays Test noise 1 channel at a time Noise occupancy vs threshold (deviation from gaussian noise) Noise versus threshold Efficiency vs threshold Test FIB ICs with ladder Test at high rates Wafer probing system (see Presentation 5D) Approved parts (see Presentation 5B) Procurements (see Presentation 5A) Tracker Peer Review, WBS 4.1.4 27 GLAST LAT Project March 24, 2003 Actions from ASIC Review • Measurements with complete trays (EM mini-tower) – Four trays, with a total of 6 SSD planes have been tested. – Initially there were noise problems that were solved by fixing the grounding • Grounding of the Al core via long screws in the MCM (always has been a Level-IV specification). • Grounding of the metallic tray service/storage box. – Now the noise and gain are consistent with what has been seen on the mini-MCMs connected to single ladders. Aluminum handle (part of service box), also grounded to the core. Long grounding screw in platedthrough hole. (Conformal coat had to be scraped off.) 2A Tracker Peer Review, WBS 4.1.4 28 GLAST LAT Project March 24, 2003 Actions from the ASIC Review • Example Layer-OR counting rate from 1 GTFE chip on a mini-tower tray (OR of 64 channels). Threshold scan FE 21 100000 10000 Counting rate (Hz) 1000 100 10 1 0,1 0,01 0 10 20 30 40 50 60 Threshold DAC 2A Tracker Peer Review, WBS 4.1.4 29 GLAST LAT Project March 24, 2003 Actions from ASIC Review • Make threshold scans on G and F version chips pulsing only one channel per chip at a time. – Pulsing only one channel, the noise sigma was about the same for G and F versions. – This confirmed the suspicion that the higher noise sigma seen in the F chip was related to crosstalk effects when multiple channels were pulsed. • Measure noise occupancy versus threshold and determine at which threshold the occupancy deviates from gaussian noise. – The resulting plots give noise sigmas consistent with expectations. – Deviations from the exponential curve at low threshold are due to saturation from the time-over-threshold. We cannot see any evidence of spurious pickup. 2A Tracker Peer Review, WBS 4.1.4 30 GLAST LAT Project March 24, 2003 Example Noise Rate Plot OR of 64 Channels. With a gain of 75 mV/fC the fitted ENC is 1710 electrons. GTFE3 18 16 14 ln(rate) 12 10 ln(rate) 8 Fit 6 4 2 0 -2 0 1 2 3 4 5 Thr**2 (fC**2) 2A Tracker Peer Review, WBS 4.1.4 31 GLAST LAT Project March 24, 2003 Single-Channel Noise Rates • Only one channel at a time is enabled in the trigger mask, and the Layer-OR rate is measured by a frequency counter at each threshold setting. • Two channels from the first chip are shown below. • Fitted noise ENC: 1580 electrons and 1542 electrons respectively. • No evidence of excess noise down to as low as 0.5 fC threshold. Channel 31, Chip 0, Ladder 11 14 14 12 12 10 10 8 8 ln(rate) ln(rate) Channel 0, Chip 0, Ladder 11 6 4 2 4 2 0 -2 6 0 0 0.5 1 1.5 2 2.5 3 -2 0 0.5 Thr**2 (fC**2) 2A 1 1.5 2 2.5 Thr**2 (fC**2) Tracker Peer Review, WBS 4.1.4 32 3 GLAST LAT Project March 24, 2003 Actions from ASIC Review Averaged Chip Noise vs Threshold Setting 2000 1500 Chip Chip Chip Chip 1000 0 1 2 3 Error bars shown for Chip 0 are the RMS spreads of the 64 channels Chips 0, 2, and 3 are FIB’ed ICs 500 10 14 18 22 26 30 Threshold 2A Tracker Peer Review, WBS 4.1.4 33 GLAST LAT Project March 24, 2003 Actions from ASIC Review • Study the how the efficiency varies with threshold. – This was studied by the Bari group by Monte Carlo. – See LAT-TD-1128. The simulation is for single muons. 2A Threshold (MIP) 0 degrees 40 degrees 0.25 99.9 99.9 0.30 98.5 96.0 0.35 98.2 94.0 0.40 97.7 92.4 0.45 96.3 90.8 0.50 92.0 89.5 Tracker Peer Review, WBS 4.1.4 34 GLAST LAT Project March 24, 2003 Actions from the ASIC Review • Test the system at high rates – This was done with a ladder, mini-MCM, and SLAC EGSE. – Self trigger, using the Layer-OR, and lower the thresholds on a subset of channels to achieve high trigger rates. – No new problems were found. – Below is a source profile at low and high trigger rates. 4 3.5 Source(100 Hz) + noise (1 kHz) Source (100 Hz) Counting rate (Hz) 3 2.5 2 1.5 1 0.5 0 64 128 Strip number 2A Tracker Peer Review, WBS 4.1.4 35 GLAST LAT Project March 24, 2003 Prelim. Results on the Flight ASICs • GTFE V-G3: 1 wafer was diced without prior probe testing: – 14 ICs were mounted on mini-MCMs and 4 on a full-size MCM (also populated with 20 older-version GTFE chips). – All 18 randomly selected chips worked 100%. – No evidence of the comparator instability that plagued the previous version (all 18 chips show identical behavior, with stable Layer-OR outputs even at the minimum threshold setting). – The timing margin on the register read-back was corrected: • All 4 chips on the full-size MCM load and read correctly at VDD=2.5V up to 28 MHz (old versions fail at 23 MHz). • All 4 chips also load and read correctly at VDD=2.25V and 20 MHz. – One mini-MCM was connected to a full-size ladder. Noise performance is similar to the previous versions. • GTRC V-6: 1 wafer was diced without prior probe testing – Tested with probe card & test suite, as well as on mini-MCM and full MCM – All functionality is correct. – Timing margin improved: data readout works up to 30 MHz at 2.5V 2A Tracker Peer Review, WBS 4.1.4 36 GLAST LAT Project March 24, 2003 GTFE Version-G3 Noise Fits • Example noise rate vs. threshold for a channel connected to an SSD strip. Channel 12 14 12 • Distribution of fitted noise values, for a fitted average gain of 74 mV/fC. 1638 electrons 8 6 4 GTFE G3 2 0 0 0.5 1 1.5 25 2.5 2 Thr**2 (fC**2) 3 20 Avg.=1634 e 15 10 5 50 19 00 18 50 16 00 15 50 13 00 12 0 50 10 90 0 75 0 60 0 45 0 30 15 0 0 0 -2 Frequency ln(rate) 10 Noise 2A Tracker Peer Review, WBS 4.1.4 37