Peter LICHARD CERN (for NA62) 1 NA62 Straw detector read-out system NA62 layout Straw detector Requirements for straw detector electronics Readout electronics frontend backend integration in DAQ Plans

Peter LICHARD CERN (for NA62) 2  Straw detector is  Pion Spectrometer  Tracking  Momentum  2 chambers before magnet  2 chambers after magnet  Recently  VETO function  Eventually can contribute to trigger  Straw detector operates in vacuum (blue tube)  Low mass, no influence on observed particles

Peter LICHARD CERN (for NA62) 3 Straw detector  Straw construction  Tube 2.1 meter * 9.8mm  36  m thin PET foils PET = polyethylene terephthalate  coated on the inside with 0.05  m of Cu and 0.02  m of Au  The anode wire (Ø=30  m) is gold-plated tungsten  Straw function  Proportional (drift) chamber  Measure particle distance from wire by drift time  Operates in vacuum  1 bar working gas  Straw electrical properties  Propagation time full length ~7ns  Resistivity of anode 50 Ohm  Resistivity of cathode 70 Ohm  Metal few atomic layers, electrical properties determined by surface effects, no skin effect  Straw is very lossy transmission line  50 % of signal from far end  Impedance frequency dependent  1000 Ohms up to 1MHz, 330 Ohm at 20 MHz

Peter LICHARD CERN (for NA62) 4  Basic readout unit of 16 straws  Common electronics control, gas, LV, HV, signal return, etc  8 straws from 2 layers  4 layers of straws form a ‘view’  480 straws if all installed  30(15*2) groups of 16(8*2) straws  Hole in the middle for beam  View is a system readout unit  Data  Control  Services, DCS  Grounding, shielding, ‘earth’  4 views form a ’chamber’  X,Y,U,V  4*480 = 1920 straws  4 chambers  2 before magnet, 2 after  In total 1920*4 = 7680 straws (but hole for the beam) Straw detector

Peter LICHARD CERN (for NA62) 5 Falling edge has the same time for all straws on track. Rising edge gives the arrival time of the first cluster The closer is the track to the wall, the bigger is the signal (clusters closer) Don’t want to see clusters => shaping must be chosen in relation to gas properties Tracks from drift time measurement. Straw function

Peter LICHARD CERN (for NA62) 6

7 NA62 Straw tracker electronics Requirements on straw tracker from electronics point of view Good quality data (drift time) 130 um resolution for single straw Efficiency -> 100% This is required For all distances from the wire All along the length of straw Low noise Low crosstalk efficiency and resolution versus rate Ion tail cancellation Integration in global DAQ and TTC Reliable and stable operation with tools for monitoring Readout chain ‘classical’ for time measurement Frontend PA+SH+BLR+COMP CARIOCA chip as a baseline (LHCb) Other possibilities under study (including discrete electronics) Backend TDC + storage and event management Frontend control, online monitoring

Peter LICHARD CERN (for NA62) 8 NA62 straw tracker electronics  Required spatial resolution 130 um  Needed TDC resolution  ‘fast’ gas Ar/CO2 (70:30), muon test beam  On X is bin (not resolution)  Blue, unknown R-T dependence  Red, known R-T dependence

Tools needed in DAQ (on/off line)  Threshold scans  Threshold control must be integrated in DAQ  Need control for each straw  I2C interface per group of 16 straws  2 lines single, 3 lines differential  Test pulse scans  Test pulse control and timing in DAQ  Separate even and odd straws  Need to control also amplitude?  NOT calibration  Noise scans  Clock delay tuning for detector timing  Delay scans  Using test pulses with fine time tuning  From data  Distributions of  Rising edges  Falling edges  Pulse width  Hit maps  Noise maps  Online R-T dependence would be useful Peter LICHARD CERN (for NA62) 9

NA62 straw tracker readout  Current choice for NA62 DAQ does not provide control flow to the detector  Only TEL1 (readout) from LHCB was selected but not the module for detector control  There are 3 options considered for straw tracker DAQ electronics (original consideration) 1. Use TELL1 boards + TDC mezzanine Most probably this option requires design and production of extra board for detector control and power distribution. It is assumed here, that the monitoring of the detector is done in DAQ. If it is not the case, the extra board could in small incremental cost include this feature. Test beam and lab small DAQ would try to adapt and profit from the NA62 DAQ development. Must prove connectivity with bulky cables. 2. Use TELL1 + custom mezzanine TDC would be designed with resources inside mezzanine FPGA. Detector control (and power supply, if possible) would be integrated as well. This option still requires bulky cables entering mezzanine, this must be proved possible. Otherwise we would need to reduce the number of straws per TELL1. 3. Design custom board. (baseline) Common NA62 functionalities would be copied (if possible) from current TELL1 design. This involves both TTC and data storage and manipulation. No mezzanine needed as everything is integrated in custom logic.  After considering all practical implications we have chosen 4-th option. We use TEL62 for DAQ and use Straw Readout Board (SRB) for control and monitoring Peter LICHARD CERN (for NA62) 10

Peter LICHARD CERN (for NA62) 11 NA62 straw tracker FrontEnd electronics Front end PCB  Gas tight PCB using blind vias (bottom faces working gas)  thick PCB to ensure mechanical stability  High voltage connection integrated on the board + HV filter  gas feedthrough  numerous configurable GND connections to try different grounding scheme with detector

Peter LICHARD CERN (for NA62) 12 NA62 straw tracker FrontEnd electronics  Front end board functionality  Two CARIOCA chips (16 straws) for signal processing  12-bit DAC with I2C control for threshold setting  LVDS buffers for outputs (to avoid current imbalances to feed-back to frontend)  Even and odd test pulse buffers  Power ~8V/0.12A  HV filter and connector Front end connectivity  Data to DAQ on 16 differential (LVDS) lines for group of 16 straws  Control from DAQ on 5 differential lines (I2C, TP)  Power 8V  DCS – temperature sensors and power sensing lines  changing connectors for HONDA VHDCI, 68 pins

Peter LICHARD CERN (for NA62) 13 NA62 straw tracker electronics  TDC at backend Transmission for required length >= 10meters would mean to include active cable equalizer for every line 10 meters AWG28 TWP can cause few ns error TDC must be put directly on the detector Problems Cables heavy and rigid, not easy to handle Always perpendicularly connected otherwise EMC compromised Crosstalk

FrontEnd electronics with RJ45 connector Peter LICHARD CERN (for NA62) 14 Use smaller cable to circumvent mechanical problems standard ethernet cable Serial I/O interfaces FPGA contains TDC, derandomizer, serial links, tools for scans and control Using high speed serial links can also help with crosstalk as CARIOCA has a bandwidth of ~20 MHz Data flow from 1 FE board 30 bits BX + 3(4) bits fine time + r/f edge + 4 bits straw ID + 1 bit control = 40 bits 8B10B coding ~800 MBits/s possible in CYCLONEIII, aim for ~160 MBits/s Category 5e ethernet cable (shielded), 5 meters Average rate 33kHz Need 42 MBits/s 1 link 160 Mbits/s sufficient Max rate 500kHz on some straws Need 320 Mbits/s 2 links 160 MBits/s (using spare output), can also increase the speed if needed Derandomizer on TDC chip to prevent data loss

FrontEnd electronics with RJ45 connector Peter LICHARD CERN (for NA62) 15  Tested now with ‘module-0’  Serial links 400 Mbits/s  1 control  1 control read-back  1 data  1 clock  The first impression from noise scans on- and off-detector seems to be very good  16 TDC channels with 0.78ns resolution in low cost CYCLONEIII (20Euro)  ‘0’ integral non-linearity  70 ps differential non-linearity caused by signal routing  1 pulse (rising+falling edge) in 12ns  Minimum CARIOCA pulse >> 12ns  Full qualification needed before going for full production

Peter LICHARD CERN (for NA62) 16 NA62 straw tracker BackEnd electronics  Control  SOB (start of burst) for counters reset and enabling TDC measurement  EOB (end of burst) for disabling TDC  Fine clock delay for detector TOF tuning  Threshold setting for each straw  Even/odd test pulses  Tools for scans and monitoring  Event management and online monitoring  ‘triggerless’ system for straw, everything buffered till higher level trigger decision  Data rate  Max rate per straw 0.5MHz  Mean rate 33kHz  Need to store 32 bit clock ID from SPS START_OF_BURST + 8 bit ‘fine’ time + rising/falling edge + straw ID 9 bits  33kHz * ~50 bits = 1.65Mbit/straw; 740Mbits/view  Desired functionality split to 2 boards  Straw Readout Board (SRB)  Receives precise clock and TTC signals from SPS (SOB, EOB)  Frontend control  Online monitoring  Serving 15 covers; ½ a view  Sends data to TELL62 over 1 optical fiber (~400 Mbits/s)  Can generate flow-control signals (choke, error)  TELL62  Developed for LHCB, adapted for NA62 (TEL62)  Event building  Event management  Receives full NA62 TTC control

Peter LICHARD CERN (for NA62) 17 NA62 straw tracker readout board (SRB)  TTC interface for precise clock and commands  Flow control (choke, error) if needed  Serving 15 FE boards  9U VME board with control and monitoring through VME bus  Event pre-processing  Data to TEL62 sent via optical link  Eventual trigger L0 sent via optical link  2 needed per view  8 per station  If data flow from straws close to the beam is too high we can increase number of SRBs to 12  32 (48) needed for detector  Scaled-down version in 6U VME now in use FPGA VME TRIG+ DATA TTC

TEL62 for straw readout  Mezzanine for TEL62  Need to design, produce and test a mezzanine with 12 high speed serial links (12 SRBs sending data to TEL62)  Commercial solution with standard chips tested  2.4Gbit link using ethernet cable tested  1.2Gbit link needed now  Necessary to use shielded cable  10 meters with better then 10**13 BERR  Firmware  Mezzanine data management and handling  TEL62 FPGA  Event building and management  Not easy with bad trailing edges  Include same hits in different events?  Flow control  Need 5  4 pieces, 1 for each chamber  1 spare Peter LICHARD CERN (for NA62) 18

Peter LICHARD CERN (for NA62) 19 NA62 straw tracker data flow 1 chamber contains 2 modules 1 module contains 2 views 1 view is a smallest readout system 30 FE boards for 1 view (2 groups of 15 boards) 2 Straw Readout Boards 2 fibers TEL62 Serving 2 chambers (2x4 views)

Physical layout  Racks  Each chamber has got 1 rack  Should be less then 10 meters from the chamber due to cables  1 VME 9U crate for SRBs  2 MPOD crates  1 needed at the moment  Spare place for 1 more depending on frontend electronics  1 TEL62 crate housing 1 TEL62  If possible, try to combine at least 2 TEL62 in 1 crate  Do we need a special module to combine 13 flow-control signals for each chamber?  Services  There is no manpower for DCS  All designed to run without Peter LICHARD CERN (for NA62) 20

TTC distribution  2 sets of LTU+TTCex ordered  1 set used in experiment  1 set in lab and spare  Each of 4 chambers has got a TTC link with 1:16 splitter for 12 SRBs and 1 TEL62  TTCex equipped with 4 lasers  Either each laser drives directly chamber link  Or we use optical 1:4 splitter and keep 3 lasers as spare (failure rate of lasers not known) Peter LICHARD CERN (for NA62) 21

Peter LICHARD CERN (for NA62) 22 Plans  Finish characterization and testing of the frontend  Validation on frontend in February  Re-design (some modification to existing cover design) and pre-production -> April  Full production start in June  Finish characterization and testing of the backend  Validation coupled to frontend  Design of full SRB till April  Prototypes of full SRB in June  Firmware  Simplified version June  Full functionality September (?, no chance if we participate to dry run)  TEL62  Mezzanine design starts after validation of FE and BE  Probably till April  Prototype in June  Firmware September  Test beam  2 weeks in June  Dry run ???  Technical run  Partially equipped module-0. Final components as much as possible.  If mezzanine and TEL62 firmware not ready, readout through VME