Fast Optical Readout of the Mu3e Pixel Detector Master Thesis Simon Corrodi March, 2014 Advisors: Dr. Niklaus Berger Department of Physics and Astronomy, Heidelberg University Prof. Dr. G¨unther Dissertori Department of Physics, ETH Zurich Zusammenfassung Das Mu3e Experiment sucht nach dem Lepton-Flavour-verletzenden Zerfall µ+ + − + 16 → e e e mit einer Sensitivit¨at von besser als 1 in 10 µ-Zerf¨allen. Um diese Sen- sitivit¨at zu erreichen, sind uber¨ eine Messzeit von ca. 1 Jahr 2 Milliarden Zerf¨alle pro Sekunde notwendig. Die Trajektorien der Zerfallsprodukte werden von Pixel-, szintillierenden Faser- und Kacheldetektoren gemessen und in Echtzeit in einer auf Grafikprozessoren basierenden Filterfarm komplett rekonstruiert.Derfur¨ die schnelle Auslese der Daten im Detektor vorhandene Platz ist stark limitiert. Das auf Kapton Flexprints, optischen Fasern und FPGAs basierende Auslesesys- tem verarbeitet 1 Tbit/s auf engstem Raum. In der vorliegenden Arbeit wurden optische Verbindungen in Kombination mit FPGA Baugruppen auf ihre Bandbreiten bei m¨oglichst kleinen Fehlerraten getestet. Bidirektionale Ubertragungen¨ mit 8 simultan genutzten Kan¨alen auf einer FPGA Tochterkarte mit SFP Steckern sind mit Fehlerraten unter < 10−16 (95 % C.L.) bei 6.4 Gbit/s realisiert worden. Optische Verbindungen im QSFP Standard k¨onnen mit einer Fehlerrate von (3.29 1.04) 10−16 bei 11.3 Gbit/s betrieben werden. ± · Die optischen Datenubertragungen¨ erfullen¨ die Anforderungen, die an das Mu3e Auslesesystems gestellt werden. Zus¨atzlich wurde gezeigt, dass Kapton Flexprints grunds¨atzlich mit einem neu angeschafften Laserplotter an der Universit¨at Heidelberg produziert werden k¨onnten. Abstract The Mu3e experiment searches for the lepton flavor violating decay µ+ e+e−e+ with a sensitivity better than 1 in 1016 µ-decays. To reach this sensitivity→ in a mea- surement period of approximately 1 year, 2 billion decays per seconds are required. The decay products’ trajectories are measured by pixel, scintillating fibers and tile detectors and fully reconstructed online by a filter farm based on graphics processing units. The available space inside the detector for the fast data readout is strongly limited. The readout system based on Kapton flexprints, optical fibers and FPGAs pro- cesses 1 Tbit/s in a very compact volume. In the presented work, optical links in combination with FPGA boards are tested with respect to their bandwidths at minimal bit error rates. Eight parallel duplex 6.4 Gbit/s links on one FPGA daughter board equipped with SFP plugs have been realized with bit error rates below < 10−16 (95 % C.L.). Optical links in QSFP standard have been operated at 11.3 Gbit/s with bit error rates of (3.29 1.04) 10−16. The optical data transmissions fulfill the requirements for the Mu3e± data acquisition· system. In addition, it has been proven that Kapton flexprints can be manufactured in principle with a new purchased laser cutting system at the University of Heidelberg. i Contents Contents ii I Introduction 1 1 Introduction 2 1.1 TheStandardModel........................... 2 1.1.1 Lepton Flavour Violating (Muon) Decays . 3 1.2 TheMu3eExperiment.......................... 5 1.3 Mu3e Readout Concept . 8 1.3.1 Pixel to Front-End Links . 9 1.3.2 Front-End FPGA . 10 1.3.3 Detector to Counting House Links . 11 1.3.4 Read-out FPGAs . 11 1.3.5 GPU Filter Farm . 12 II BasicsofDataTransmission 13 2 Physical Layer 14 2.1 Signal propagation . 14 2.1.1 Electrical Conductors . 14 2.1.2 Optical Wave Guides . 15 2.2 EncodingSchemes ............................ 15 2.2.1 Line Codes . 16 2.2.2 Running Disparity . 17 2.2.3 Scrambling . 19 2.2.4 Protocols . 19 2.3 Signal Quality Check . 23 2.3.1 Eye Diagrams . 23 2.3.2 Bathtub Diagrams . 23 2.3.3 Cyclic Redundancy Checks (CRC) . 25 3 Electronic Components 27 3.1 Logic Gates . 27 3.2 MemoryElements ............................ 27 3.2.1 Flip-Flops . 27 3.2.2 Random-Access Memory (RAM) . 28 ii 3.2.3 FirstInFirstOut(FIFO) . 28 3.2.4 Read-Only Memory ROM . 28 3.3 Phase Locked Loop (PLL) . 28 3.3.1 Clock Data Recovery (CDR) . 29 3.4 Linear Feedback Shift Register (LFSR) . 29 3.4.1 Pseudo Random Number Generators (PRN) . 29 3.4.2 Counter . 29 3.4.3 Other Uses . 30 3.5 GrayCounter............................... 30 4 Field Programmable Logic Gates (FPGA) 31 III Measurements 33 5 Optical Links 34 5.1 Soft- and Hardware . 34 5.1.1 Altera Stratix V Development Kit . 34 5.1.2 SantaLuz Mezzanine Board . 37 5.1.3 Plugs . 37 5.1.4 Cables . 39 5.2 Firmware ................................. 40 5.2.1 Data Transmission State Machine . 40 5.2.2 Bit Error Rate Tests (BERT) . 42 5.2.3 Altera Receiver Toolkit . 44 5.3 Measurements............................... 46 5.3.1 BER Upper Limit and Error Calculations . 46 5.3.2 Optical SFP Links . 48 5.3.3 Single Channel SFP . 50 5.3.4 Multi-Channel SFP . 55 5.3.5 Optical QSFP Links . 60 5.4 Discussion................................. 61 5.4.1 Summary . 62 5.4.2 Crucial Points . 63 6 Readout Chain Firmware Components 65 6.1 Front-EndFPGA............................. 65 6.1.1 Hit Data Structure . 65 6.1.2 Concept I . 66 6.1.3 Concept II . 66 6.1.4 Comparison . 66 6.2 Coordinate Transformation on FPGAs . 66 6.2.1 Coordinate Systems . 67 6.2.2 The Transformation . 68 6.2.3 The Implementation . 68 6.2.4 Performance . 70 6.2.5 Conclusion . 70 iii 7 LVDS on Kapton FlexPrints 71 7.1 Kapton . 71 7.2 Low-Voltage Differential Signaling (LVDS) . 71 7.3 Laser Platform . 72 7.4 Proof of Concept . 73 7.5 FutureWork ............................... 75 IV Outlook 76 8 Outlook 77 8.1 Readout Chain in General . 77 8.2 DataStructure .............................. 77 8.2.1 Starting Point . 77 8.2.2 Error Detection . 78 8.2.3 Proposed Format . 78 8.3 Phase Ia Readout Chain . 78 A Appendix 80 A.1 StratixVTransceivers .......................... 80 A.1.1 Physical Media Attachment (PMA) . 80 A.1.2 Physical Coding Sublayer (PCS) . 82 A.2 Quartus II and ModelSim . 85 A.2.1 ModelSim............................. 86 A.3 Multi-ChannelResults .......................... 88 A.4 MuPix4 Emulator . 88 A.5 SantaLuz Crosstalk Measurments . 89 A.6 MuPixAddressScheme ......................... 90 List of Figures 92 List of Tables 94 Bibliography 95 Acknowledgements 99 iv Part I Introduction 1 Chapter 1 Introduction The Standard Model (SM) of particle physics describes the constituents of matter as well as their interactions. It is described in more detail in a first section, followed by the observation of lepton flavor violation through neutrino oscillations and its consequences for the theory. These motivate the search of lepton flavor violating processes in charged leptons as described in another section. The Mu3e experiment looks for the charged lepton flavour violating decay µ eee. In a second chapter, the design of this experiment is discussed. Particularly,→ the experiment’s readout chain, the main scope of this thesis, is presented in detail. 1.1 The Standard Model The Standard Model (SM) of particle physics is a quantum field theory which de- scribes the fundamental constituents and interactions of matter. As shown in figure 1.1, matter consists of six quarks and six leptons, and their anti-particles, which are arranged in three generations. The interactions between quarks and leptons are mediated by four types of gauge bosons. The first generation consists of up (u) and down (d) quarks with electrical charges of +2/3 and 1/3 respectively , the negatively charged electron (e−) and the neutral − neutrino (νe). The lepton family number Le is characteristic for the leptons of this family. The second and third generation consist both in each case of two quarks with the same charge as the first generation - these are charm (c) and strange (s) in the second and top (t) and bottom (b) in the third generation. Their associated leptons, again with the same electrical charge as the ones in the first generation, are muons − (µ ) and the neutrino (νµ), tau (τ) and the neutrino (ντ ). Their characteristic lepton family numbers are Lµ and Lτ . In the SM neutrinos are massless and lepton flavour is a conserved quantity. Quarks and leptons are spin 1/2 particles whose interaction is mediated by spin 1 particles, the gauge bosons. The eight gluons mediate the strong interaction, photons (γ) the electromagnetic interaction and Z, W + and W − bosons the weak force. The model has demonstrated huge and continued successes, particularly the re- cent discovery of the long predicted higgs boson in 2012 [1] at the LHC. Gravitation is not included in the standard model [2, 3]. 2 Figure 1.1: Standard Model Particles [4, modified ]. Lepton Flavour Violation Different experiments have observed mixing of neutrino flavours. Super-Kamiokande and others have observed [5] mixing in atmospheric and solar neutrinos, SNO [6] in solar neutrinos and KamLAND [7] in reactor neutrinos. The mixing angles in the Pontecorvo Maki Nakagawa Sakata (PMNS) matrix, the matrix which describes the neutrino mixing, are close to maximal [8]. Neutrino oscillation is only possible if neutrinos have a non-vanishing mass, which is not foreseen in the SM. An extension of the Minimal Standard Model by heavy right-handed neutrinos, called νSM, is required to incorporate neutrino masses con- sistent with oscillation experiments. The reason why the neutrino masses are signif- icantly smaller than other particle’s masses remains a puzzle [9]. Even though the PMNS matrix appears also in charged lepton currents, lepton flavour violation has never been observed in charged leptons. These flavour-changing neutral currents are suppressed by a mechanism described by Glashow, Iliopoulos and Maiani in 1970 [10].