Genetic multiplexing, or how to read more with less electronics

1. Genetic multiplexing, or how to read more with less electronics Sébastien Procureur IRFU/SPhN

2. Content → Short introduction → Potential for particledetection & first idea (double sided) → Geneticmultiplexing → Resultswith a 50x50 cm² Micromegas prototype → Other types of multiplexing → Some applications → Conclusion and perspectives Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

3. Introduction to multiplexing Definition: an electronic process that allows more than one electrical signal to be sent using only one connection → first introduced in telegraphy in the lateXIXthcentury to speed up communications However, multiplexing is much older than telegraphy! → language: ~100,000 wordsexpressedwithonly 26 letters (or more…) and evenlesssounds → At the origin of poetry, anagrams, spoonerisms… Centrale nucléaire ⟷ le cancer et la ruine « Il parait que Pouchkine a toujours adoré les laitues » (J. Martin) → Very bad multiplexing, though: 265 ~ 12 millions ⟹ 10 letters would be enough CXLIICCCLXVIIIDCCXCV → numbers: infinitenumberswithonly 10 digits = 142,368,795 Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

4. Multiplexing and particle detectors Obvious interest: lower the number of electronic channels → easierintegration, cabling, cooling → cheaper (~1€/channel) → lowerconsumption An example: the Compass experiment @ Cern → 12 layers of Micromegas in the hottestregion → 1,000 strips per layer, total rate ~ 30 MHz ⟹ only ~ 20 channels (2%) with signal for a given event Risks of multiplexing: → degradation of S/B ⟹ canlower the detectionefficiency → ambiguities to solve (demultiplexing) Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

5. Digression: Micromegas principle Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

6. Multiplexing: first ideas → Initiated by the need to equip the CLAS12 cosmic bench with large reference detectors (tracking) → Use of Micromegas detectors (MPGDs) and the bulk manufacturing process (2006) → StéphanAune: 2 detectors on a single PCB (“double sided”) PCB • Top sidewith p1 large strips (a few cm) • Bottomsidewith p2thinstrips (~500 microns), • repeated p1 times → Can build a detector equivalent to p1xp2 strips, with p1+p2 channels → Optimum is p1=p2=p/2 ⟹ n=p²/4 Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

7. Double sided multiplexing → 6 such detectors werebuilt at the Saclay workshop, with an active area of 50x50 cm² • Top sidewith 32 large strips (~1.5 cm) • Bottomsidewith 32 thinstrips (488 microns), x32 1stsmall prototype Event display with 4 detectors 50 cm Thinstrips Large strips → Efficiency plateau reached for large stripsonly Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

8. Double sided multiplexing Allowed to characterizeseveral detectors for CLAS12, but several limitations/drawbacks → thinstripsidesdon’treach the efficiency plateau Thinstrip capacitance: 2 nF ⟹ 10% of the real charge iscollected Partiallycompensated by the 1 cm drift gap (3x more primaries) → large stripsidesreach the plateau… Large strip capacitance: 1 nF ⟹ 17% of the real charge iscollected Partiallycompensated by the 1 cm drift gap (3x more primaries) Partiallycompensated by the cluster size of 1 (2x more signal on the strip) → … but severalnoisy/deadstrips (3% loss per strip!) PCB PCB → Ambiguity of localization on the edge of the large strips → Unsolvableambiguity if more than 1 particle → Requires 2 workingbulks Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

9. Multiplexing & information Multiplexinginherently leads to a certain loss of information → in the previous pattern, the information on which group of thinstripssees the particleislost → thislost has to becompensated by an additional information, provided by another detector (large stripside) The best way to multiplex wouldbe to look for redundant information, and design a multiplexing pattern for which the lost information exactlycoincideswith the redundant one… → Is thereanyredundancy in the detector’s signal? Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

10. Genetic multiplexing Starting point: → in most cases, a signal isrecorded on at least 2 neighbouringstrips Wecanmake use of thisredundancy, and combine channelswithstrips in such a waythat 2 givenchannels are connected to neighbouring stripsonly once in the detector The sequence of channelsuniquely codes the position on the detector… 1 2 → blocks of thinstrips are no longer identical → the localization of the particledoesn’trequire large stripsanymore → the connection {channels}n ⟷ {strips}pcanberepresented by a p list of channelnumbers For n channels, there are a priori n(n-1)/2 unordered doublets combinations, and thus one canequip a detector with at most p = n(n-1)/2+1 strips Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

11. Genetic multiplexing Several possibilities to build the pattern, i.e. the sequence of p numbers: → generate the sequencerandomly: cannot build all the doublets channel # strip # → build the ith block from 1+k.i [n] (ideafrom Raphaël Dupré) build all the doublets if n prime → idem, but simply use the first availablechannel build almost all the doublets ∀ n Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

12. Genetic multiplexing A signal canbe deposited on more than 2 strips… so the repetition of k-uplets (k>2) shouldbechecked → A priori no problem, as there are much more k-upletsthan doublets… → But the repetition of small k-upletsdoesappear in this construction: i=1 i=2 i=3 Repetition of the triplet 9-1-5 i=4 i=5 → Can beimproved by reordering the blocks: i=1 i=3 Repetition of the quadruplet 3-6-9-1 i=2 i=4 i=5 Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

13. Double sided vs genetic multiplexing → Patent n° 12 62815 (S. Procureur, R. Dupré, S. Aune): « Circuit de connexion multiplexé et dispositif de connexion permettant notamment de réaliser un multiplexage » Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

14. Prototype 50x50 cm² active area, readwith n = 61 channels (highest prime numberbelow 64…) - 488 micron pitch - could have equiped up to 61x60/2+1=1831 strips (~90 cm) - p= 1024 strips → Smallest k-upletrepeated: k=15 Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

15. Prototype 50x50 cm² active area, readwith n = 61 channels (highest prime numberbelow 64…) - 488 micron pitch - could have equiped up to 61x60/2+1=1831 strips (~90 cm) - p= 1024 strips connector mesh defect 50 cm Strip capacitance: 1.3 nF (compared to 2 nF for thinstripfrom double sided) Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

16. Results with cosmics Prototype tested in the CLAS12 cosmicbench (60x60 cm² couple of scintillators) • 1 cm drift gap • 128 micron amplification gap • Gas: Ar+5%isobutane • Edrift=300 V/cm • 1.5 m cable (70 pF/m) • T2K electronics (AFTER) • Shaping time: 200 ns • Samplingfrequency 60 ns • Offline common mode subtraction event display amplitude [ADC] → Demultiplexingneeded! time sample → All strips OK → position distribution as expected → mean noise on strips: 3,950 electrons Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

17. Results with cosmics Prototype tested in the CLAS12 cosmicbench (60x60 cm² couple of scintillators) usual Landau distribution → effective gains up to 3,000 in spite of large capacitances → < 2% of eventswith a cluster size of 1 (cannotbelocalized), as expected Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

18. Genetic multiplexing and flux 1st simulation of the ambiguityprobability at ≠ flux and for ≠ degrees of multiplexing • Samegeometry (gaps, size, pitch) • Primaryelectrons on Poisson distribution • Transverse diffusion • Time window: 100 ns • Assume independentparticles → can stand up to 0.5 kHz/cm² in this configuration (1 MHz in total) → at 10 kHz/cm², the electronicscanbe reduced by a factor of 2 (1% ambiguities) Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

19. Other types of multiplexing • Geneticmultiplexingcanalsobeusedwith drift chambers 1 3 5 7 3 7 4 1 7 6 5 2 4 6 1 5 2 6 4 3 2 1 • Multiplexingbetween detectors (genetic or not: canalternatively assume a straight track) → alsoworkswith pixels or pads • Standard geneticmultiplexing in a single detector with pixels or pads is more difficult → toomanyneighbours 1 unique channel per block dressedwith multiplexedones → alternative patterns are possible, if the signal isspread on several pixels Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

20. Some applications → Homeland security: scan large volumes requires large detectors with high resolution recentstudies on scans withcosmic muons → high resolution small size K. Gnanvoet al., GEM detectors Simulation with large Micromegas S. Pesenteet al., drift chambers (8 minutes!) → large area, poorresolution Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

21. Some applications - Muon tomography for volcanology: requires large detectors withlowconsumption first resultswith 80x80 cm² scintillators (~1 cm resolution) N. Lesparreet al. → ~ 50 W for the whole installation, hostile environment - Dosimetry need light, portative setup → lowconsumption TPC with 250k pixels, 10k channels - Applications in particlephysics → sLHCproject: > 1,000 m² of MPGDs, millions of electronicchannels (~1€/channel) → ILC TPC ( ~ 1 million pads) Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

22. Conclusion & perspectives Modern particlephysics and many more applications require: → large area setups → high spatial resolution → integration, lowconsumption → tight budget! Multiplexingbecomes more and more feasiblethanks to advances in instrumentation & electronics → concept of geneticmultiplexingvalidatedwith a Micromegas → almost no multiplexing at a spectrometerlevel⟹ a lot to bedone Optimizationneeded for a given flux/configuration → if n channelssuffice for 99% of the interestingevents, isit relevant to have 2n, 3n channels more for the remaining 1%? Next steps for geneticmultiplexing: - Flux studies to validate the preliminary simulations - Resistive, multiplexed Micromegas to furtherincrease S/B (thissummer?) → Goal: 1m² detector with 100 micron 2D-resolution and < 200 channels Genetic Multiplexing Saclay, 14/05/2013 S.Procureur

23. Many people participated in the preparation/integration of the prototype… • Marc Anfreville & Robert Durand (bulk process) • Arnaud Giganon (electric tests, integration) • David Attié • Stéphan Aune • Rémi Granelli • Gabriel Charles • IrakliMandjavidze • Caroline Lahonde-Hamdoun • … Genetic Multiplexing Saclay, 14/05/2013 S.Procureur