1 / 38

SLHC Optoelectronics

SLHC Optoelectronics. Readout architectures Technologies for TX High speed multiplexing Packaging Radiation hardness and reliability testing. Architecture (1) Low speed links. Like current SCT: 2 data links/module (redundancy ). Architecture (2) High Speed Links.

locke
Download Presentation

SLHC Optoelectronics

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SLHC Optoelectronics • Readout architectures • Technologies for TX • High speed multiplexing • Packaging • Radiation hardness and reliability testing ATLAS Tracker Upgrade Liverpool December '06

  2. Architecture (1) Low speed links • Like current SCT: 2 data links/module (redundancy) ATLAS Tracker Upgrade Liverpool December '06

  3. Architecture (2) High Speed Links • High speed MUX at end of supermodule. Modules Data Concentrator 1 Data Concentrator 8 MUX LD ATLAS Tracker Upgrade Liverpool December '06

  4. Architecture for Pixels • MCC reads out n pixel chips. For SLHC, pixel chip size similar  occupancies x10  data transmission rates x10. • Pixels need 1-2 Gbits/s. • Question: can we have a common architecture for pixels and strips? ATLAS Tracker Upgrade Liverpool December '06

  5. Cost Estimates (1) • Assume Strawman layout for strips • 21824 modules barrel 11968 modules disks • Low speed links: scale based on actual costs of SCT links + input from S-C Lee on commercial costs for opto-packages. • Total for strips = 32.7 MCHF (components only). ATLAS Tracker Upgrade Liverpool December '06

  6. Costs (2) • High speed links • Multiplex 30 modules  one fibre. Data rate ~3 GBits/s. • Scale costs from actual costs of LHCb GOL/VCSEL readout at 1.6 GBits/s. Larger uncertainties here. No redundancy in this calculation. • Estimated total cost for strips ~ 2 MCHF. • This estimate is approximate but conclusion  High speed links costs << low speed links. Francois Vasey “you have to fill the bandwidth of the fibre to be cost effective”. • Note a mixed solution with low speed fibres along supermodule and multiplexing to high speed links at the end of a stave would also be very expensive (only ~ 30% of the cost is for fibre). ATLAS Tracker Upgrade Liverpool December '06

  7. Advantages Low Speed Links • Grounding: separate ground for each module but • have to join grounds for serial powering • ATLAS SCT had common grounds for larger number of modules (40 or 52) in End caps with no significant change in noise (TTC Redundancy interlinks s/c DGND). • For barrel SCT had 100 W redundancy links between DGND for neighboring modules in loop of 12 modules. No increase in noise seen. ATLAS Tracker Upgrade Liverpool December '06

  8. Disadvantages Low Speed Links • Cost much higher! • Packaging more difficult (space constraints more severe) • Single source for rad-hard SIMM fibre; might not be available for SLHC. • Fragile fibres on detector are not good (had many fibre breakages, some not repairable). • After mounting on detector, failures could not be replaced ( failures 4 or 5 years before start of operation were not recoverable). • Mounting is very difficult, time consuming  labour costs high. • Needs new chip set and we have nobody to work on it • Nobody interested in links to work on this option. ATLAS Tracker Upgrade Liverpool December '06

  9. Low vs High Speed Links • Low speed versus high speed affects everything in system: • Fibres, transmitters, packaging  affects supermodule engineering. • We won’t make much progress until we make a decision. • My suggestion: • Adopt high speed links as baseline. • Low speed links as fall back if we can’t solve noise problems for modules. ATLAS Tracker Upgrade Liverpool December '06

  10. Technologies • VCSELs @ 850 nm (ATLAS) + SIMM/GRIN fibre. • EELs @ 1310 nm (CMS) + SM fibre. • VCSELs @ 1310 nm (new) + SM fibre. ATLAS Tracker Upgrade Liverpool December '06

  11. VCSELs @ 850 nm (1) • Advantages: • Very radiation hard, very small threshold shifts at SLHC fluences ( next transparencies) • Easy to couple into MM fibres. • Disadvantages • Needs custom packaging a la SCT/Pixel. • Bandwidth of SIMM fibre too low and GRIN fibre not radiation hard enough  mixed SIMM/GRIN fibre. • Needs custom fibres  concerns on cost and availability. ATLAS Tracker Upgrade Liverpool December '06

  12. Radiation Hardness VCSELs • See talk by Issever at LECC 2006 http://indico.cern.ch/contributionDisplay.py?contribId=47&sessionId=12&confId=574 • No significant change in slope efficiencies up to SLHC fluence • Small threshold shifts  transparency • But some channels became sick after annealing @ 15 mA. Not understood  needs further studies… ATLAS Tracker Upgrade Liverpool December '06

  13. Results – Threshold Shift (After-Before) 10 days @ 10mA annealed and 5 days @ 15mA annealed ΔThreshold [mA] LHC, annealed @ 20mA, proton implant ATLAS Tracker Upgrade Liverpool December '06

  14. VCSELs @ 850 nm (2) • Fibre Bandwidth • Very radiation hard SIMM fibre has low bandwidth > ~50 MHz km • Radiation tolerant GRIN fibre has higher bandwidth 1121 MHz km. • Splice 8m of SIMM fibre to ~ 80m GRIN fibre (as done for current Pixel readout). • Could operate up to ~ 5 Gbits/s. • Demonstration of Bandwidth • Tests by KK Gan (OSU) See talk at LECC 2006 http://indico.cern.ch/contributionDisplay.py?contribId=48&sessionId=12&confId=574 • Scope photos of eye diagrams at 2 Gbits/s  transparency ATLAS Tracker Upgrade Liverpool December '06

  15. ATLAS Tracker Upgrade Liverpool December '06

  16. EELs • Advantages • Couple to SM fibre at 1310 nm  choice of commercial fibres that are sufficiently radiation hard and very high bandwidth. • Can survive SLHC fluences. • Disadvantages • Higher thresholds than VCSELs and larger threshold shifts with radiation. • More difficult to couple to fibre but can be done by telecoms companies (as for current CMS). ATLAS Tracker Upgrade Liverpool December '06

  17. Radiation damage EELs 300 MeV/c p • Threshold shift for SLHC worst case 2 1015p cm2 is ~105 mA. • 70% of damage will be annealed during operation • Threshold before irradiation ~ 4 mA.  after full SLHC fluence threshold ~ 36 mA (high!). K. Gill, SPIE 2002. http://cms-tk-opto.web.cern.ch/cms%2Dtk%2Dopto/tk/publications/wdocs/kg_spie2002.pdf ATLAS Tracker Upgrade Liverpool December '06

  18. VCSELs @ 1310 • Advantages • Best of both worlds. High bandwidth, availability of several sources of commercial radiation hard fibre. VCSELs @ 1310 nm expected to be very radiation hard. • Disadvantages • New technology  concern about availability. • Need to verify radiation hardness and reliability. Major effort in conjunction with CMS. ATLAS Tracker Upgrade Liverpool December '06

  19. High Speed Multiplexing • Current technology: GOL on 0.25 mm CMOS. Operates at 1.6 GBits/s . Radiation hard. • Should be possible to go faster with 0.13 or 0.09 mm. • CERN project aims to develop MUX ASIC. • SMU also developing MUX as part of LOC. ATLAS Tracker Upgrade Liverpool December '06

  20. CERN VBDL Proposal • Versatile bi-directional links for ATLAS & CMS • Use for data, TTC and experimental control. • Following transparencies from P. Morerira, LECC 2006 talk http://indico.cern.ch/contributionDisplay.py?contribId=128&sessionId=22&confId=574 ATLAS Tracker Upgrade Liverpool December '06

  21. Transceiver Module MCM containing the ASIC, optoelectronic components and optical and electrical connectors. Common definition O/E • DAQ • Timing • Trigger • Experiment Control GBT E/O Configurable to multiple optical networks (user driven) Multi-protocol ASIC Qualified optoelectronic components (COTS)

  22. Limiting Amplifier • Specifications: • Data rate: 3.60 Gbit/s • Gain: > 55 dB • Bandwidth > 2.52 GHz • Equivalent input noise: < 1 mV • Minimum input signal (differential): 10 mV • Maximum input signal (differential): 600 mV

  23. Limiting Amplifier Gain Cell Gain Cell Gain Cell Bias Gain Cell OffsetCancellation Output Buffer Size: 194 mm × 194 mm

  24. Limiting Amplifier 3.35 Gbit/s 1 Gbit/s

  25. CERN VBDL Summary (1) • Versatile Link solution for: • Timing Trigger Links; • Data Acquisition Links; • Experiment Control Links. • The system allows flexible link topologies: • Bi-directional • Uni-directional • Point-to-Point • Point-to-Multipoint

  26. CERN VBDL Summary (2) • Specifications and Interfaces are still evolving for which we need the feedback of the potential users • Some universal building blocks have already been prototyped: • Laser driver • Encoder/decoder: Line code and FEC • Limiting amplifier • The Versatile Bi-Directional Link project has been proposed by the Microelectronics group as a CERN common development.

  27. LOC • LOC being developed by SMU for LAr @ SLHC • Integrate electronics and laser/PIN on chip using SOS technology  see talk by Jingbo Ye at this workshop. ATLAS Tracker Upgrade Liverpool December '06

  28. Link-on-Chip Architecture Flip-chip bonding PLL and clock generator REFclock • Improve performance • No off-chip high speed lines • Flip-chip bonding reduces capacitance and inductance • Reduce power consumption • No 50-Ohm transmission lines between chips • Designed and Implemented in Silicon-on-Sapphire technology • Targeting speed:>2.5Gbps encoder Laser Laser Driver serializer Parallel Data TX transmitter Module Optical data Receiver Module Parallel Data Flip-chip bonding Photonic Decoder De- serializer TIA/LA PIN REFclock Clock/Data recovery

  29. Flipped OE devices on SoS substrate flip chip attachment UTSi integrated photo detector UTSi integrated circuitry receiver circuitry VCSEL driver circuitry quad PIN array quad VCSEL array active CMOS layer 200 um transparent sapphire substrate (UTSi) MMF ribbon fiber • Flip-chip bonding of OE devices to CMOS on sapphire • No wire-bonds – package performance scales to higher data rates • Rugged and compact package ATLAS Tracker Upgrade Liverpool December '06

  30. Transceiver IC with OE Devices and Link Performance ATLAS Tracker Upgrade Liverpool December '06 Transceiver link eye at 3.2Gbps at 2.0Gbps

  31. Packaging • Non trivial because must be radiation hard, non-magnetic, low mass, low Z material and fit in available space • Full custom packaging. Eg SCT opto-package or Pixel MT coupled arrays • Find Telecoms company package that is compatible with our requirements (CMS) ATLAS Tracker Upgrade Liverpool December '06

  32. Custom Packaging - SCT ATLAS Tracker Upgrade Liverpool December '06

  33. Used for Pixel on-detector and SCT/Pixel off-detector BOC 12 way array with MT guide pins for coupling to 12 way ribbon fibre ATLAS Tracker Upgrade Liverpool December '06

  34. Telecoms Packaging • CMS Analogue Opto Hybrid (AOH) • 3 channels laser drivers, lasers and fibres ATLAS Tracker Upgrade Liverpool December '06

  35. CMS Laser Sub-Mount • Compact: 4.5 * 4 * 1.3 mm3 • Fibre ends gold plated • Active alignment: resistor pads used to solder fibre in location. Glue only for strain relief of fibre • Radiation hard by design Removed for CMS (save $)

  36. Pixel Links • Consider option to keep similar architecture • Number of links would be similar to current Pixel system and increase in luminosity  10 times higher data rate Need ~ 1 Gbit/s • Develop similar architecture chips to current DORIC and VDC in 0.13 mm • See talk by K.K. Gan at this workshop ATLAS Tracker Upgrade Liverpool December '06

  37. Common ATLAS/CMS WGs • Theme a- Lessons learned and to be learned. • Collect info on successes and mistakes of the groups involved in the present detectors. Follow up on the technology choices made over 10 years ago. Produce a transparent account of the costs incurred. Create a repository for all publications. Monitor and follow up the performance and ageing of the installed links. • Theme b- Radiation hardness and reliability of optoelectronic components. • Establish common procedures and common ways to represent the irradiation data, share facilities and coordinate irradiation runs, avoid redundant tests and share results. • Theme c- Common optical link reference test bench. • Define a reference test system for multi-gigabit/s optical links. Define test procedures and evaluation criteria. Specify the interface to the links to be tested. Develop hard, software and FPGA-IP blocks. Purchase test equipment and build reference test bench. Test proposed SLHC links on common reference bench and evaluate with common criteria. ATLAS Tracker Upgrade Liverpool December '06

  38. Outlook • Much more work to do on radiation hardness: • Understand 850 nm VCSEL performance better • Compare damage factors in p/p/n tests. • Start testing 1310 nm VCSELs. • Continue fibre testing to SLHC doses • Test Si and InGaAsP p-i-n diodes to SLHC fluences • Packaging • Custom versus modified COTs • System issues • Need decision on low speed vs high speed links • If we adopt high speed links many detailed questions: • How many modules/link? • Do we want intermediate electrical multiplexing? • How do we introduce some level of redundancy? ATLAS Tracker Upgrade Liverpool December '06

More Related