Joint Atlas CMS SLHC Optoelectronics Working Group A

Joint Atlas CMS SLHC Optoelectronics Working Group A: Lessons Learned n n n K. K Gan F. Vasey T. Weinberg n Cost n n n Quality n n francois. vasey@cern. ch Today Long term monitoring Technology n 19 -Mar-07 Material Personnel Components Architectures 0

Cost in CHF, material ATLAS PIX Quantity ATLAS SCT CMS TK 4144 12264 42800 1’ 527 k 3’ 486 k 12’ 597 k Development Cost 455 k 250 k 2’ 000 k Production Cost per link Development Cost per link 369 284 294 110 20 47 Production Cost 19 -Mar-07 francois. vasey@cern. ch 1

Cost in CHF, material (2) 19 -Mar-07 francois. vasey@cern. ch 2

Cost in CHF, material (3) 19 -Mar-07 francois. vasey@cern. ch 3

Cost: lessons learned n All links ended up being expensive n n n Overproduction (included in cost) n n n Ø ~ 300 CHF/link, production cost 25%-30% for cables and connectors (more for Atlas Pixels) 70%-75% for actives + hybrid/flex Plus 10% - 30% development cost Covers assembly yield, test systems and prototypes, QA testing, installation yield, unforeseen problems Up to 130% - 140% * needed quantity for CMS TK Spares down to % level Maximize bandwidth usage, limit number of variants, foresee overproduction 19 -Mar-07 francois. vasey@cern. ch 4

Cost in MY, personnel (large uncertainty) ATLAS PIX Quantity ATLAS SCT CMS TK 4144 12264 42800 R&D 12 17 50 Production 11 23 55 Installation 5 Links/MY 19 -Mar-07 148 francois. vasey@cern. ch 10 307 372 5

Cost in MY, personnel (2) 19 -Mar-07 francois. vasey@cern. ch 6

Cost: lessons learned (2) n Over 180 MY of effort n n Effort per link seems to stabilize above 10 k pcs n Ø R&D: Prod: Install 45%: 10% 11 years Not included in ATLAS SCT and CMS TK project cost Very few points, large uncertainty Share resources for systems < 10 k links Ø 10 k links ~ 30 MY 19 -Mar-07 francois. vasey@cern. ch 7

Quality n Over 180 MY of effort n n Effort per link seems to stabilize above 10 k pcs n Ø R&D: Prod: Install 45%: 10% 11 years Not included in ATLAS SCT and CMS TK project cost Very few points, large uncertainty Share resources for systems < 10 k links Ø 10 k links ~ 30 MY 19 -Mar-07 francois. vasey@cern. ch 8

Quality (19 March 2007) ATLAS SCT Data Quantity Dead ATLAS SCT TTC 4144 0. 2% 0. 04% ESD mishandling Broken fibers 0. 4% ? % 0. 3% Slow Turn On Light coupling into MM fiber Excess loss Broken ribbons packaging mishandling system 19 -Mar-07 42800 0. 5% packaging Problematic CMS TK francois. vasey@cern. ch mishandling 9

Lessons learned by ATLAS: Quality n System n n n Avoid use of single fibres on detector Always use balanced codes Separate front-end and opto-modules (i. e. with TP) Test system with all final production components before launching production QA n n Ensure QA is for identical conditions to final system More longer term testing Better traceability of production lots Respect ESD precautions 19 -Mar-07 francois. vasey@cern. ch 10

Lessons learned by CMS: Quality n System n n n Avoid fiber pigtails. Do not allow excessive fibre-slack without corresponding management scheme. Use ruggedized ribbon/fiber only. Avoid simplified and/or compact connectors which are fragile and difficult to dismount and clean. QA n Develop and distribute fibre-test tools which allow on-line channel quality testing, providing immediate feedback during construction. 19 -Mar-07 francois. vasey@cern. ch 11

Quality: long term monitoring n ATLAS n n n BER scans vs discriminating threshold in DRx 12 Measure supply current to front end PIN diode CMS n n n 19 -Mar-07 francois. vasey@cern. ch Automatic monitoring of analog link operating point and gain embedded in CMSSW framework Lost tokens flagged in digital control link Spare fiber channels looped back to measure fiber darkening 12

Conclusion n Excellent progress n n n Technical Social Technology lessons still to be written Full document ready for Prague? Results are immediately useful ATLAS and CMS not vastly different, with complementary experience. 19 -Mar-07 francois. vasey@cern. ch 13