The Compact Muon Solenoid Experiment. Conference Report. Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland

Transcription

1 Available on CMS information server CMS CR -2017/438 The Compact Muon Solenoid Experiment Conference Report Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland 27 November 2017 (v4, 14 December 2017) The CMS Outer Tracker Upgrade for High Luminosity LHC Jelena Luetic for CMS Collaboration Abstract The era High Luminosity Large Hadron Collider will pose unprecedented challenges for detector design and operation. The planned luminosity upgraded machine is 5x10 34 cm 2 s 1, reaching an integrated luminosity more than 3000 fb 1 by end The CMS Tracker detector will have to be replaced in order to fully exploit delivered luminosity and cope with demanding operating conditions. The new detector will provide robust tracking as well as input for first level trigger. This report is focusing on replacement CMS Outer Tracker system, describing new layout and technological choices toger with some highlights research and development activities. Presented at VERTEX2017

2 The CMS Outer Tracker Upgrade for High Luminosity LHC Université Libre de Bruxelles (BE) The era High Luminosity Large Hadron Collider will pose unprecedented challenges for detector design and operation. The planned luminosity upgraded machine is 5x10 34 cm 2 s 1, reaching an integrated luminosity more than 3000 fb 1 by end The CMS Tracker detector will have to be replaced in order to fully exploit delivered luminosity and cope with demanding operating conditions. The new detector will provide robust tracking as well as input for first level trigger. This report is focusing on replacement CMS Outer Tracker system, describing new layout and technological choices toger with some highlights research and development activities. The 26th International Workshop on Vertex Detectors September, 2017 Las Caldas, Asturias, Spain Speaker. on behalf CMS collaboration c Copyright owned by author(s) under terms Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).

3 1. Introduction The high luminosity Large Hadron Collider (HL-LHC) will begin a new era in life CERN LHC. The accelerator upgrade will be installed during long shutdown period in 2024 and The main goal is to be able to deliver an increased instantaneous luminosity 5x10 34 cm 2 s 1, with a peak value reaching up to 7.5x10 34 cm 2 s 1. Such conditions pose unprecedented requirements on detectors as number pileup events increases to 140 events 1. Thus, good spatial and time resolution become increasingly important in order to distinguish between different primary events. During lifetime HL-LHC, an integrated luminosity fb 1 will be collected, thus allowing studies rare and suppressed events. Due to expected CMS Tracker detector deterioration in performance, it will have to be fully replaced before HL-LHC operation. The new detector will have improved radiation tolerance to withstand radiation levels up to n eq /cm 2. In order to distinguish between different primary interactions and maintain high tracking efficiency, an increase in detector granularity is planned. Additional improvement tracking efficiency arises from smaller material budget within Tracker volume. Selection good events at first trigger level will be enhanced by Tracker information, ensuring rejection tracks with low momentum. A complete overview current state plans and developments for CMS Tracker Phase-2 Upgrade can be found in [1]. 2. Outer Tracker geometry The new CMS Tracker detector comprises two major parts. The Inner Tracker (IT) placed at lower radii around collision point will be built using silicon pixel technology. The Outer Tracker (OT) located furr away from collision point, will be constructed with combination silicon pixel and silicon strip technology. The baseline design for Outer Tracker foresees two kinds modules, 2S modules with two closely spaced parallel strip sensors and PS modules that are constructed with one macro-pixel and one strip sensor. The Outer Tracker will consist six cylindrical layers in central barrel region z < 1200 mm and five double-discs on each side, in region 1200 < z < 2700 mm (Figure 1). Particles emerging from collision region ( z < 70 mm) will cross all six modules in pseudorapidity range up to η < 2.4, apart from a small region where barrel and endcap regions connect, thus ensuring hermeticity. Part barrel region is realized with modules tilted with respect to z axis, as it was shown in simulations that better performance can be achieved with reduced cost and amount material [1]. 3. Outer Tracker module design The upgrade Outer Tracker is closely connected to upgrade CMS Trigger system [2]. With increasing pile-up, trigger system will have to handle higher output rates interesting events and improve discriminatory power event selection. Incorporating more sources event information into trigger system and increasing time available to make a 1 In most challenging performance scenario, number pileup events could reach

4 20 Chapter 2. Overview Phase-2 Tracker Upgrade r [mm] z [mm] Figure 2.3: Sketch one quarter tracker layout in r-z view. In Inner Tracker Figure 1: Tracker geometry green lines correspond to pixel modules made two readout chips and yellow lines to pixel modules with four readout chips. In Outer Tracker blue and red lines represent decision two types will lead modules to an improved describeddiscriminating in text. power. Outer Tracker modules will provide a rough information about track transverse momentum (p T ) to first level CMS Trigger system (L1 trigger), allowing for better transverse momentum resolution various objects at L1 performance expected to be achieved with upgraded tracker is discussed in Chapter 6, and mitigation pile-up effects. There is a plan to enhance first level trigger rate from present with details on L1 and fline tracking as well as primary vertex reconstruction presented in Section 100 khz 6.3. to 750 khz. The amount time available for data transfer and running L1 trigger algorithms corresponds to trigger latency and will increase to 12.5 µs from present value 3.2 µs. The front-end electronics and L1 trigger track reconstruction need to comply with 2.3seOverview new requirements. upgraded tracker concept Tracker input to L1 trigger 3.1 p T module concept The enhancement trigger performance involves both a higher output rate interesting The main idea behind Outer Tracker module design is implementing trigger at events and an improved discriminating power event selection, which is more challenging tracker level. Such a novel concept implies that modules have to provide data about event at in a high pileup environment. Improved discriminating power will be achieved by using more bunch crossing rate. The front-end electronics will be responsible for data volume reduction information in trigger decision, with a longer latency available for its processing. The use tracking by simple information and fast reconstruction, in L1 trigger roughly will determining improve incoming transverse particle momentum p T. The modules resolution are various n capable objects atrejecting L1 (e.g. all jets), will data from allowparticles exploitation below a certain information p T threshold. onthe track design isolation, andse willso-called contribute p T to modules mitigation exploits bending pileup. CMS particle plans to trajectory enhance inside first 3.8 level T magnetic trigger rate fromfield presently CMS 100detector khz toas 750 a function khz andto increase particle p T. The latency transverse frommomentum present is value determined 3.2 µs to 12.5 by making µs. The twront-end closely spaced electronics measurements and with L1 trigger two silicon track sensors reconstruction read out by need same to comply set withse ASICs. newthe requirements. ASICs will make correlations between hit pairs (later referred as "stubs"), The selecting necessity ones providing that are compatible tracking information with particles to above L1 trigger p T is threshold. a main driver Simulations for show design thatouter a p T cut Tracker, 2 GeV, including reducesits module data by aconcept. factor ten, The making use tracking stub information information transfer in at L1 trigger bunch implies crossing that frequency trackera has viable to option. send out The self-selected p T thresholdinformation is selected byat tuning every bunch readout crossing. chip Such setting functionality corresponding relies to upon search local window. data Moreover, reductionininorder to front-end ensure good electronics, track selection in order to limit performance volume and minimize data that contributions has to be sent fromout random at 40 MHz. coincidences, This ismodules achieved with with different modules thatspacings are capable between rejecting sensors signals will befrom built, particles with distances belowranginga certain between p T threshold, 1.6 mm and referred 4 mm. to as p T Stub modules coordinates [23]. are Tracks n transmitted from charged to particles track finding are bent system in that transverse combines plane data by from all3.8 T fieldmodules, and CMS sends magnet, it to with CMS L1 trigger bending system. angle Full depending event information on isp stored T in particle. pipeline The modules on are front-end composed readout ASIC, two single-sided waiting for closely-spaced L1 trigger to accept sensors signal read out whenbyit aiscommon transmittedset front-end to CMS ASICs data that acquisition correlate system. signals in two sensors and select hit pairs (referred to as stubs ) compatible with particles above chosen p T threshold (Fig. 2.4 (a)). A threshold around 2 GeV corresponds to a data volume reduction roughly one order magnitude, which is sufficient to enable transmission 2 stubs at 40 MHz, while all or signals are stored in front-end pipelines and read out when a trigger signal is received. The p T thresh

5 (a) PS Module (b) 2S Module Figure 2: Two varieties CMS Outer tracker p T modules 3.2 PS and 2S modules Two versions p T modules are envisioned for Outer Tracker upgrade. Modules with two parallel strip sensors (2S modules) will be placed in outer layers tracker volume, while in inner layers, re will be modules with one strip and one macro-pixel sensor (PS modules). Both variants can be seen in Figure 2. Sensors used for module construction will be n-on-p type with an active thickness around 200 µm. The surface 2S module sensors is 10x10 cm 2, with strip length 5 cm and pitch 90 µm. The design strip sensor PS module is similar to one 2S with an area 5x10 cm 2 and strip length 2.35 cm. The PS strip sensor pitch is 100 µm. The macro-pixel sensor has 1467 µm long pixels with pitch 100 µm [3]. The spacing between sensors is achieved by gluing m onto aluminium/carbon fibre spacers (Al-CF) that provide mechanical stability and efficient heat removal. Strip sensors will be wire bonded to front-end hybrids for read-out. The hybrid consists a single Kapton flex circuit board that is folded and laminated onto stiffeners same thickness as Al-CF spacers between sensors. 2S hybrids carry eight CMS Binary Chips (CBCs) [4] connected to top and bottom sensor and a Concentrator Integrated Circuit (CIC) [5], which buffers, aggregates and sparsifies data from CBC chips sending it furr down to readout link. CIC is also responsible for delivering clock, trigger and control signals to CBC chips. The PS front-end hybrid reads out strip sensors with Short-Strip ASICs (SSAs) [6] and CIC chip. The pixel sensor is bump-bonded to Macro-pixel ASIC (MPA), responsible for reading out pixel data and correlating it to data from SSAs [7]. All chips have a binary readout. Electronics for powering front-end chips and optical data readout are placed on service hybrids. 2S modules contain one service hybrid on one end module. In PS module, services are split in two hybrids, on opposite ends module, with one side housing powering functionalities while or performs control and readout operations. Electronics components associated with service hybrid are part common CERN developments for multiple upgrade projects. These include following: DC-DC converter - generates voltages for front-end electronics and optoelectronics, 3

6 HV connector - provides bias voltage for sensors, LpGBT chip - serializes/deserializes data to/from back-end, controls front-end ASICs and distributes clock and trigger, VTRx+ optoelectronic transceiver - converts data signals to optical/electric. Service hybrids are connected to back-end electronics transferring stub data to dedicated track trigger boards and L1 triggered data to data acquisition system. 3.3 Track trigger The design Outer Tracker is largely driven by need to process large amount L1 trigger data. The main goal system is to reduce amount tracks with low transverse momentum (p T < 2 GeV), which is done by correlating two closely spaced measurements taken with modules described earlier in this section. This procedure allows formation and selection stubs consistent with chosen transverse momentum cut with high efficiency. Good performance throughout detector is ensured by a programmable correlation window and fset between two sensors. Data reduction in scale 10 to 100 times is expected, depending on track location. In CMS experiment three different approaches to L1 tracking have been investigated: AM FPGA: based on FPGA and specially designed associative memory ASICs, which rely on pattern recognition previously loaded patterns inside ASIC, with final selection done by FPGA, Hough transform: two step approach with coarse stub selection done using Hough transform [8] and final track selection with Kalman filter [9], Tracklet: based on road-search algorithm, where stubs in neighbouring layers form a seed, which is used in a linearised χ 2 fit to obtain final track parameters. For each se approaches, a dedicated demonstration system has been constructed. Demonstrators have shown that all three approaches would be able to achieve necessary performance in terms delivering tracks within imposed timing constraints using simulated data. Future developments are aimed at combining three approaches for building a final system. 4. Prototyping and testing Several prototype modules have been built and ir performance and stability have been tested. Measurements have been performed to ensure a good performance in conditions as close as possible to real operating conditions and to estimate possible effects external conditions such as humidity and temperature. In order to evaluate module response and irradiation effects, several beam tests have been conducted at CERN, DESY and FNAL. 4

7 4.1 CBC2 mini module The performance CBC chip has been tested by building two CBC2 mini-modules. Such a module comprises a hybrid with two early prototype versions CBC chip with most stub finding logic implemented [12]. The hybrid is wire-bonded to two silicon strip sensors with 254 strips 5 cm length. Several mini-modules were built and ir performance tested in beam tests as well as in various system tests, assessing stability long term operation in nominal conditions. A separate long-term operational stability test was performed with one CBC2 minimodule. The module was operated at -30 C for a week in order to measure effects long term cold operation on noise levels. The noise was measured as a standard deviation S-curve, showing that noise level as a function time was stable around what corresponds roughly to 1000 electrons, which is well below signal level for a MIP particle (Figure 3). The fluctuations in measurement correspond to small changes in environmental conditions Beam test A 2S mini-module beam test was carried out at CERN H6B beam line using 120 GeV pions. Reference tracks were reconstructed with AIDA telescope [10] and matched to hits in trigger plane using ATLAS FE-I4 chip [11]. A comparison between stub efficiencies irradiated and nonirradiated mini-modules was performed with results shown in Figure 4. The irradiated mini-module was exposed to a fluence 6x10 14 n eq /cm 2 corresponding to twice maximal fluence expected in CMS Outer Tracker. The stub reconstruction efficiency is defined as fraction events with one reconstructed track that contains at least one reconstructed stub. The matching between track and stub is performed within 4σ spatial resolution sensor. For non-irradiated module, sharp rise in efficiency and plateau at 99% prove that 56 this module can efficiently select tracks above acertain threshold. For Chapter irradiated 3. The Outer module, Tracker plateau is reached at 95% efficiency. The p T resolution in both cases is around 5% Temperature = -30 C ] Average noise [V cth CBC0 0.5 CBC Time [h] Figure 3.20: Noise a 2S mini-module operated at 30 C for approximately one week. The Figure 3: Mini-module long term cold operation. noise is measured as standard deviation derivative S-curve both cases. The noise was converted to electrons by using internal test pulse mechanism CBC2. A comparison noise per chip is 5 shown in Fig. 3.19, right. The noise all chips is below specification 1000 electrons. No significant difference in noise is observed for most chips; for two chips that are closest to shielding service hybrid noise is increased by about 10% or less. While this will be subject furr optimizations, study indicates that operation DC-DC converters in close vicinity to sensors and readout hybrids is feasible. Information on system tests with a mini-module and service hybrids can be

8 3.3. The p T modules Figure 3.21: FigureStub 4: Mini-module reconstruction stub reconstruction efficiency efficiency for a as non-irradiated a function particle (red) incident andangle. irradiated The (blue) 2S mini-module. red curve shows Themeasurements mini-module for a non-irradiated was irradiated mini-module, to a fluence while blue 6curve 10shows 14 n eq measurements refers to done forthreshold an irradiated mini-module. setting, while d is sensor spacing. The thresholds used /cm 2. The variable V cth correspond to about 4900 and 3500 electrons for unirradiated and irradiated module, respectively. Radii 69 cm and 60 cm were used for calculation p T from tilt angle non-irradiated 4.2 CBC2 full module and irradiated module, respectively (Section ). The different radii compensate for fact that modules had different sensor spacing but were operated with same stub acceptance window. This version prototype module was built with two full 2S sensors wire-bonded to two Kapton flex hybrids including eight CBC2 chips each. In absence concentrator chip and service hybrid, all differential lines from chips were read out directly from connector and H6B beamdedicated line using low and120 highgev voltage pions. connections Tracks wereare made. reconstructed As a part using module data characterization, from AIDA telescope [38] and noise matched level was measured to hits at in rooma temperature trigger plane, for twowhich differentutilizes modules. Studies ATLAS based onfe-i4 chip [39]. Events with internal one calibration track matched pulse toreadout trigger CBC2 chip plane showare thatused. measured The stub noisereconstruction corresponds to efficiency is definedaround as 1000 fraction electrons and se is relatively events uniform with across at least all one testedtrack-matched hybrids. Anor test stub. wastracks performed on a full module powered with and without prototype service hybrid. This prototype and stubs have to match within 4s spatial resolution device under test. For non-irradiated hybrid contains a DC-DC converter to supply necessary voltages to chips. The goal test module measured turn-on threshold, defined as p T for which stub efficiency reaches was to see if close proximity a DC-DC converter 50%, is 1.88 GeV with a p T resolution 5% 5 would introduce additional noise into, to be compared to an expected turn-on threshold system. The results can be seen in Figure 5. The noise levels are comparable across module value 2with GeV. onlythe two chips sharpclosest turn-on to and converter having plateau significantly value different 99% efficiency noise levels. demonstrate These are, that this module type however, canstill select well below efficiently signal stubs levels. above chosen threshold and thus meets specifications. For irradiated mini-module a plateau efficiency above 95% is observed under beam test conditions Testing and fullfor CBC2 a fluence module inabout test beam twicewas carried expected out at one. samethe H6B ptest T resolution beam facility is preserved at 5%. More at CERN details used and also for results testing from CBC2 thismini-module. beam testthe aremodule provided was mounted in Section a translator stage and scanned with beam to measure stub reconstruction efficiency per strip. The aver- The test age beam efficiency facility was found at CERN to be around was97% alsoand utilized is shown in tofigure characterize 6. The efficiency performance is lower than one three 2Sone full-size measuredmodules for mini-modules available and it arises at from time contamination writing. with For this events measurement where sensors were synchronization maintainedbetween at 250 Vtelescope and andv cth module valuewas lost. set to 115. The stub reconstruction efficiency per strip was n measured as number events with a stub matched to a telescope track reconstructed within 208 µm. Events were 6 required to have optimal module-to-telescope synchronization and TDC (Time to Digital Converter) phase. Figure 3.22 shows stub efficiency 6 across module. The region between strips 185 and 239 has no data because it was not scanned by moving beam and measurements strips located at module ends suffer from large statistical uncertainty due to limited amount data collected during

9 Figure 3.18: Noise per CBC2 two 2S full-size modules operated at room temperature. The noise is taken as width a cumulative distribution function normal distribution fitted to S-curve. The uncertainties are given by RMS measurement The Outer Tracker Upgrade for High Luminositymeets LHC specifications. More details Jelena Luetic 1488CMS to approximately 1000 electrons and refore about 1489 measurement method as well as illustrations measured S-curves and noise per chanthe pt modules nel are provided in Section Figure 3.18: Noise per CBC2 two 2S full-size modules operated at room temperature. The a 2S module (a) Noise measurements for twofull (b) Noisewith measurement with prototype housfigure 3.19: Left: photo modules system test setup and a service servicehybrid hybrid arnoise is taken as width a cumulative distribution function normal ing distribution a DC-DC ranged as in final module. Right: comparison converter noise per chip as measured with direct fitted to S-curve. 58 The uncertainties are given by RMS measurement. Chapter 3. The Outer Tracker powering (red) and with powering through service hybrid (green). Error bars correspond Fullper 2S chip. module system tests to standard deviationfigure 5: noise to approximately 1000 electrons and refore meets specifications. More details about measurement method as well as illustrations measured S-curves and noise per chan1491 In most complete system test performed so far, a full-size nel are provided in Section S module was combined with a prototype 1 service hybrid (Fig. 3.19, left). The service hybrid was geometrically arranged as it will be in final module. The module s noise was extracted from S-curves and a comparison is0.98 made between noise obtained when module receives its low voltages via service hybrid and consequently via two DC-DC converters, and direct powering from a labo0.96 ratory power supply. The provision bias voltage was via a dedicated circuit, identical for Stub efficiency Average efficiency = ± Figure 3.19: Left: photo system test setup with a 2S module and a service hybrid ar0.86 ranged as in final module. Right: comparison noise per chip as measured with direct powering (red) and with powering through service hybrid (green). Error bars correspond to standard deviation noise per chip Strip number In most complete system test performed so far, a full-size 2S module was combined with a Figure 3.22: Stub reconstruction efficiency a full-size 2S module measured at CERN beam prototype service hybrid (Fig.6: 3.19, left).2s Themodule service hybrid was geometrically arranged Figure Full stub reconstruction efficiency measured with test beam test facility. The region corresponding to strips was not scanned by beam. as it will be in final module. The module s noise was extracted from S-curves and a comparison is made between noise obtained when module receives its low voltages via service hybrid and consequently via two DC-DC converters, and direct powering from a laboratory power supply. The provision bias voltage was via a dedicated circuit, identical for 4.3 MaPSA-light asymmetric errors on each measurement are taken into account, is ± The efficiency is approximately thanwith that48measured in 2S mini-module. This is due A prototype version 2% an lower MPA chip channels (MPA-light) was used to construct an to a different configuration module, modulescalled and macro-pixel a remainingsub-assembly potential contamination assembly operational six chips bump bonded to a bare (MaPSA). events where module was not synchronized with telescope. The average value The assembly was also used for optimization bump bonding process. In order to asses efficiency per chip is presented in Section stub-finding capabilities MPA chip, two MaPSAs were stacked toger with a few milthe MaPSA-light was characterized using FNAL test beammode, facility providing limetre gap, creatingassembly a PS micro-module. One assembly was set to strip emulation in which 120 Thecan response MaPSA-light assembly was determined by measuring GeV MPA protons. readout chip output strip-detector like signals, while or was responsible for hit reconstruction efficiency as a function phase difference particle arrival stub finding logic. time and MPA clock. The particle arrival time is estimated by using Fermilab main injector clock, which introduces a jitter due to difference in frequencies beam MaPSA-light was testedefficiency in detail atspectrum a FNAL test facilityand providing 120 GeV clockthe and MPA clock.assembly The measured is beam thus fitted unfolded, yielding protons. The hit reconstruction efficiency as a function phase difference particle ar jitter-corrected efficiency distribution. The result is shown in Fig The hit efficiency is close to 100% and dependence efficiency on arrival time is consistent with specifications Mechanical structures The TB2S mechanics reuses proven solutions from current tracker, adapted to new lay- The Tracker Barrel with 2S modules (TB2S)

10 Jelena Luetic Efficiency rival time and MPA clock was measured to asses response MaPSA-light assembly. Due to different frequencies beam clock and MPA clock jitter was introduced into measurement particle arrival time. The jitter-corrected efficiency distribution is obtained by fitting and unfolding efficiency spectrum. The result is shown in Figure 7. The hit efficiency 3.4. Mechanical structures 59 is close to 100% and dependence efficiency on arrival time is consistent with specifications Fitted Fitted, jitter corrected Data t (ns) Figure 3.23: Hit efficiency asfigure a function between particle arrival 7: MaPSA testphase beam difference results time and 40 MHz clock edge in MPA. The measured efficiency spectrum is shown by black dots, red curve represents a fit to measurement, and black curve shows jitter corrected efficiency distribution. The threshold had been set to about 3000 electrons. 5. Conclusions The upcoming HL-LHC upgrade poses numerous challenges for experiments for maintaining an excellent physics performance. In order to operate in high luminosity conditions, CMS Outer Tracker will have to undergo a full replacement. New modules that are able to perform filtering based on transverse momentum incoming particle and provide information to L1 triggering system are currently being designed. A number prototypes has already been built and y are showing promising results even after being exposed to high radiation fluences. References [1] K. Klein, The Phase-2 Upgrade CMS Tracker, CERN-LHCC , CMS-TDR-014, CERN, June Figure 3.24:Geneva, Drawings a TB2S ladder with its 12 modules (left) and support wheel (right). [2] CMS Collaboration, Technical Proposal for Phase-II Upgrade CMS Detector, CERN-LHCC , LHCC-P-008, CMS-TDR (2015) length for and back, forming a U-shaped circuit. [3] full A. Dierlamm ladder CMS Collaboration, Characterisation silicon sensor materials and designs for CMSstructure Tracker Upgrade, PoS (Vertex2012) 016 (2013).metallic jig. The components (carbon fibre The ladder is assembled in a high-precision C-priles, cross pieces, inserts, and cooling pipe) positioned on jig and a low [4] D. Braga, Development Readout Electronics forare High Luminosity Upgrade CMSviscosity epoxy adhesive is applied to Imperial joints. College The adhesive is cured at room temperature to avoid Outer Strip Tracker, PhD sis, London, January deformations caused by difference in rmal expansion metallic jig and carbon fibre components Low and high voltage wires as well as optical 8 fibres are routed along modules and inside C-priles to one end each ladder. There will be in total 36 wires per ladder, three for each module (LV, HV, common ground). At end ladder those wires are connected to one multi-core cable. The 24 optical fibres (two per module) are connected at ladder end to 24-fibre bundles Small-size all-metal fittings are used to connect each ladder cooling pipe to supply and

11 [5] D. Braga et al., I/O data formats for Concentrator Integrated Circuit, Shared%20Documents/Forms/AllItems.aspx. [6] A. Caratelli et al., Short Strip ASIC Specifications Document, September [7] D. Ceresa et al., Macro Pixel ASIC (MPA): readout ASIC for pixel-strip (PS) module CMS outer tracker at HL-LHC, 2014 JINST 9 C11012, doi: / /9/11/c [8] P. V. C. Hough, Method and means for recognizing complex patterns, US Patent 3,069,654, December [9] R.Fruhwirth, Application Kalman filtering to track and vertex fitting, Nucl. Instrum. Meth. A262 (1987) 444, doi: / (87) [10] I. Rubinskiy, An EUDET/AIDA Pixel Beam Telescope for Detector Development, Physics Procedia 37 (2012) 923, doi: /j.phpro [11] M. Barbero, The FE-I4 Pixel Readout Chip and IBL Module, Technical Report, ATL-UPGRADE-PROC , CERN, Geneva, [12] G. Hall et al., CBC2: A CMS microstrip readout ASIC with logic for track-trigger modules at HL-LHC, Nuclear Instruments and Methods in Physics Research Section A, Volume 765, 2014, Pages , doi: /j.nima