22/06/2016James Leaver Current FED Tester Status.

Slides:



Advertisements
Similar presentations
6 Mar 2002Readout electronics1 Back to the drawing board Paul Dauncey Imperial College Outline: Real system New VFE chip A simple system Some questions.
Advertisements

Digital Filtering Performance in the ATLAS Level-1 Calorimeter Trigger David Hadley on behalf of the ATLAS Collaboration.
Adding electronic noise and pedestals to the CALICE simulation LCWS 19 – 23 rd April Catherine Fry (working with D Bowerman) Imperial College London.
First Results from Tracker 1  Cryostat Commissioning  AFE/VLSB Firmware and Readout  Cosmic Ray Setup  Tracker Readout  Software  Trigger Timing.
1 Scintillating Fibre Cosmic Ray Test Results Malcolm Ellis Imperial College London Monday 29 th March 2004.
29 June 2004Paul Dauncey1 ECAL Readout Tests Paul Dauncey For the CALICE-UK electronics group A. Baird, D. Bowerman, P. Dauncey, C. Fry, R. Halsall, M.
MICE Tracker Front End Progress Tracker Data Readout Basics Progress in Increasing Fraction of Muons Tracker Can Record Determination of Recordable Muons.
20 Feb 2002Readout electronics1 Status of the readout design Paul Dauncey Imperial College Outline: Basic concept Features of proposal VFE interface issues.
Comparing ZS to VR David Stuart, UC Santa Barbara June 19, 2007.
DSP online algorithms for the ATLAS TileCal Read Out Drivers Cristobal Cuenca Almenar IFIC (University of Valencia-CSIC)
Cluster Threshold Optimization from TIF data David Stuart, UC Santa Barbara July 26, 2007.
Status of Oxford Setup Matthew Chalk, Erik Devetak, Johan Fopma, Brian Hawes, Ben Jeffery, Nikhil Kundu, Andrei Nomerotski University of Oxford ( 18 August.
The first testing of the CERC and PCB Version II with cosmic rays Catherine Fry Imperial College London CALICE Meeting, CERN 28 th – 29 th June 2004 Prototype.
Veto Wall Test Hyupwoo Lee MINERvA/Jupiter Group Meeting July 18, 2007.
Octal ASD Certification Tests at Michigan J. Chapman, Tiesheng Dai, & Tuan Bui August 30, CERN.
Chemometrics Method comparison
Data and Computer Communications Chapter 8 – Multiplexing
GODIAN MABINDAH RUTHERFORD UNUSI RICHARD MWANGI.  Differential coding operates by making numbers small. This is a major goal in compression technology:
LECC2004: Performance of the CMS Silicon Tracker FED: Greg Iles13 September Performance of the CMS Silicon Tracker Front-End Driver 10th Workshop.
Performance test of STS demonstrators Anton Lymanets 15 th CBM collaboration meeting, April 12 th, 2010.
Tests with JT0623 & JT0947 at Indiana University Nagoya PMT database test results for JT0623 at 3220V: This tube has somewhat higher than usual gain. 5×10.
Prototype Test of SPring-8 FADC Module Da-Shung Su Wen-Chen Chang 02/07/2002.
Aras Papadelis, OTR meeting NIKHEF, June 17 th LHCb Outer Tracker Cross-talk measurements with Fe-55.
Status of the Beam Phase and Intensity Monitor for LHCb Richard Jacobsson Zbigniew Guzik Federico Alessio TFC Team: Motivation Aims Overview of the board.
“End station A setup” data analysis Josef Uher. Outline Introduction to setup and analysis Quartz bar start counter MA and MCP PMT in the prototype.
1 VeLo L1 Read Out Guido Haefeli VeLo Comprehensive Review 27/28 January 2003.
Commissioning and Operation of the CMS Tracker analogue optical link system at TIF with CMSSW: R.Bainbridge, A.Dos Santos Assis Jesus, K.A.Gill, V. Radicci.
Development of Multi-pixel photon counters(2) M.Taguchi, T.Nakaya, M.Yokoyama, S.Gomi(kyoto) T.Nakadaira, K.Yoshimura(KEK)
Features of the new Alibava firmware: 1. Universal for laboratory use (readout of stand-alone detector via USB interface) and for the telescope readout.
Ideas about Tests and Sequencing C.N.P.Gee Rutherford Appleton Laboratory 3rd March 2001.
FED RAL: Greg Iles5 March The 96 Channel FED Tester What needs to be tested ? Requirements for 96 channel tester ? Baseline design Functionality.
John Coughlan Tracker Week October FED Status Production Status Acceptance Testing.
FPGA firmware of DC5 FEE. Outline List of issue Data loss issue Command error issue (DCM to FEM) Command lost issue (PC with USB connection to GANDALF)
Tracker Week October CCLRC, Rutherford Appleton Laboratory, Oxon, UK Imperial College, London, UK Brunel University,
Trip-t testing progress report
Trigger Meeting: Greg Iles5 March The APV Emulator (APVE) Task 1. –The APV25 has a 10 event buffer in de-convolution mode. –Readout of an event =
Calibration of the gain and measurement of the noise for the apv25 electronics K. Gnanvo, N. Liyanage, C.Gu, K. Saenboonruang From INFN Italy: E. Cisbani,
09/01/2016James Leaver SLINK Current Progress. 09/01/2016James Leaver Hardware Setup Slink Receiver Generic PCI Card Slink Transmitter Transition Card.
LHCb VELO Upgrade Strip Chip Option: Data Processing Algorithms Giulio Forcolin, Abdul Afandi, Chris Parkes, Tomasz Szumlak* * AGH-Krakow Part I: LCMS.
BME 353 – BIOMEDICAL MEASUREMENTS AND INSTRUMENTATION MEASUREMENT PRINCIPLES.
FED 9U Analog Characterization Stefanos Dris CERN & Imperial College.
1 Level 1 Pre Processor and Interface L1PPI Guido Haefeli L1 Review 14. June 2002.
S.MonteilPRS timing1 November Clermont team - Calorimetry meeting Preshower timing / The cosmics and TED results / Collection of plots. 1. Overview.
Predicting the In-System Performance of the CMS Tracker Analog Readout Optical Links Stefanos Dris CERN & Imperial College, London.
CMS FED Testing Update M. Noy & J. Leaver Imperial College Silicon Group.
LECC2004: Performance of the CMS Silicon Tracker FED: Greg Iles13 September Performance of the CMS Silicon Tracker Front-End Driver 10th Workshop.
LKr readout and trigger R. Fantechi 3/2/2010. The CARE structure.
Tracker Week February presented by John Coughlan RAL FED Status FEDv2 Testing Pre-Series Manufacture Final Production.
General Tracker Meeting: Greg Iles4 December Status of the APV Emulator (APVE) First what whyhow –Reminder of what the APVE is, why we need it and.
S.MonteilPS COMMISSIONING1 MaPMT-VFE-FE ELECTRONICS COMMISSIONING AND MONITORING. OUTLINE 1)Ma-PMT TEST BENCHES MEASUREMENTS 2)VFE AND FE ELECTRONICS FEATURES.
Juan Valls - LECC03 Amsterdam 1 Recent System Test Results from the CMS TOB Detector  Introduction  ROD System Test Setup  ROD Electrical and Optical.
1M. Ellis - MICE Tracker PC - 1st October 2007 Station QA Analysis (G4MICE)  Looking at the same data as Hideyuki, but using G4MICE.  Have not yet had.
Rutherford Appleton Laboratory September 1999Fifth Workshop on Electronics for LHC Presented by S. Quinton.
FPGA based signal processing for the LHCb Vertex detector and Silicon Tracker Guido Haefeli EPFL, Lausanne Vertex 2005 November 7-11, 2005 Chuzenji Lake,
LECC2003: The 96 Chann FED Tester: Greg Iles30 September The 96 channel FED Tester Outline: (1) Background (2) Requirements of the FED Tester (3)
Development of Multi-pixel photon counters(2) M.Taguchi, T.Nakaya, M.Yokoyama, S.Gomi(kyoto) T.Nakadaira, K.Yoshimura(KEK) for KEKDTP photon sensor group.
3/06/06 CALOR 06Alexandre Zabi - Imperial College1 CMS ECAL Performance: Test Beam Results Alexandre Zabi on behalf of the CMS ECAL Group CMS ECAL.
20 April 2007MICE Tracker Phone Meeting1 Analysis of cosmic/self-triggerd data of station 5 Hideyuki Sakamoto MICE Tracker Phone Meeting 20 th April 2007.
SVD FADC Status Markus Friedl (HEPHY Vienna) Wetzlar SVD-PXD Meeting, 5 February 2013.
The LHCb Calorimeter Triggers LAL Orsay and INFN Bologna.
PreShower Characterisations
96-channel, 10-bit, 20 MSPS ADC board with Gb Ethernet optical output
Roberto Chierici - CERN
Vertex 2005 November 7-11, 2005 Chuzenji Lake, Nikko, Japan
Muon Recording Studies and Progress for the MICE Tracker
K. Sedlak, A. Stoykov, R. Scheuermann
NanoBPM Status and Multibunch Mark Slater, Cambridge University
BESIII EMC electronics
The CMS Tracking Readout and Front End Driver Testing
Shaped Digital Readout Noise in CAL
Presentation transcript:

22/06/2016James Leaver Current FED Tester Status

22/06/2016James Leaver Software Status FED Tester software is in a fairly refined state Universal Toolbox class enables full calibration, configuration and initialisation of system with a single function call Toolbox configures and provides access to: –FED Testers –FED –FED Kit –Frame data generator –FED event wrapper which transparently enables readout though Slink or VME Forms the core of several testing programs, which could easily be used at RAL Entire FED Tester Software package: 31,563 physical lines of code so far… TrimDAC Calibration FTE AOH Calibration FED Timing Calibration Calibrator Configurator Allow User To Make Changes Initialise Device Descriptions Initialiser Toolbox Provide Access To Devices Provide Access To Descriptions Initialise Devices

22/06/2016James Leaver TrimDAC & AOH Calibration TrimDAC calibration: –FED Tester output disabled –FED TrimDACs calibrated (Fed9UDevice method) –TrimDAC values increased to shift baseline below FED ADC range – ensures efficient use of ADC range when optical input from FED Testers is present AOH calibration: –FED Testers output alternate digital high / digital low signals to each FED channel –FED Tester AOH bias currents and gains are adjusted for each FED channel to optimise use of FED input ADC range

22/06/2016James Leaver TrimDAC & AOH Calibration: Results FED ADC range (digital low → digital high) post TrimDAC and AOH calibration:

22/06/2016James Leaver FED Timing Calibration FED Testers send tick marks to FED All FED coarse and fine delay settings are scanned through Appropriate FED delay settings for each FED channel are found, to ensure optimal data sampling points

22/06/2016James Leaver FED Timing Calibration: Results Positions of calibrated sampling points on a tick mark for FE Unit 0:

22/06/2016James Leaver FED Inter-Channel Crosstalk Wanted to measure the effects of crosstalk between FED channels Will show two sets of results here: –‘Worst Case’: 11 of 12 FE Unit channels carry a ‘noise’ signal, look at crosstalk on remaining channels –‘Nearest (& Next Nearest) Neighbour’: 1 of 12 FE Unit channels carry a ‘noise’ signal, look at crosstalk on neighbouring channels

22/06/2016James Leaver Crosstalk: Worst Case: Setup Sent typical frame to FE Unit channel 5 Sent similar frame to other FE Unit channels, but added a delay of 10 clock cycles –(Repeated for each FE Unit) Used FED Testers to simultaneously phase shift all frames relative to FED sampling point, from 0 to ~25 ns in ~100 ps steps –By capturing Scope Mode data from the FED at each phase step, can build up a high resolution image of signal seen by FED

22/06/2016James Leaver Crosstalk: Worst Case: Results FE Unit 0: A view of the overlapping frame digital headers - All channels apart from 5 see the blue input signal

22/06/2016James Leaver Crosstalk: Worst Case: Results(2) Level shift of ~4 ADC counts when blue frame goes low (power issue?) Spikes of ~10 ADC counts in amplitude

22/06/2016James Leaver Crosstalk: Nearest Neighbour: Setup Send typical empty frame to all FE Unit channels apart from 5 Send a frame to FE Unit channel 5 which has a 2-strip wide ‘pulse’ in the centre of the payload –(Repeated for each FE Unit) Use FED Testers to phase shift frames sent to channel 5 (keep constant phase for other channels) – build up high resolution signal images as before Repeat with 3 pulse heights: –Pulse 1: Pedestal to digital high –Pulse 2: Pedestal to (2/3) * (digital high) –Pulse 3: Pedestal to (1/3) * (digital high)

22/06/2016James Leaver Crosstalk: Nearest Neighbour: Results Pulse 1: Nearest Neighbour Pulse 1: Next Nearest Neighbour FE Unit 2: All channels apart from 5 see the blue input signal With maximum ‘noise’ pulse height: –Nearest neighbour crosstalk amplitude: ~3.5 ADC counts –Next nearest neighbour crosstalk amplitude: negligible

22/06/2016James Leaver Crosstalk: Nearest Neighbour: Results(2) Pulse 2: Nearest Neighbour Pulse 3: Nearest Neighbour

22/06/2016James Leaver FED Channel Noise Simple noise measurement: –Disabled FED Tester output and set appropriate FED TrimDAC / OptoRx values (constant across all channels) –Captured a Scope Mode event for each FED channel (Scope Length = 1020) –Found mean and standard deviation of signal at each channel:

22/06/2016James Leaver FED Hit Check Wanted to check that the numbers / locations of hits input to the FED match the numbers / locations of hits output by the FED in Zero Suppressed Mode.

22/06/2016James Leaver FED Hit Check: Setup Ran with 100 kHz random triggers, FED in Zero Suppressed Mode Used randomly generated events with simulated CMS cluster distribution (2% Tracker occupancy) Set FED strip high & low thresholds to 50 Set the hit height in our randomly generated frames to vary (almost) between pedestal and digital high –i.e. from just below the FED strip threshold to near the top of the FED ADC range Generated 1024 events (FED Tester capacity), read them out through the FED, generated another 1024 events, repeated 100 times Compared every input hit ADC value with every output hit ADC value; expect a linear scatter graph of non-zero values if FED detects every input hit correctly

22/06/2016James Leaver FED Hit Check: Results FED Channel 0: Graph contains data from ~5 x 10 5 hits

22/06/2016James Leaver FED Hit Check: Results(2) Close-up of linear region

22/06/2016James Leaver FED Hit Check: Results(3) Extreme close-up of linear region

22/06/2016James Leaver FED Hit Check: Results(4) A close-up of the region around the FED strip threshold: The FED appears to detect a small fraction of hits that occur below the strip threshold!

22/06/2016James Leaver FED Hit Check: Results(5) A scatter plot of the same data shows the effect more clearly: ‘Hits’ below FED threshold ‘Hits’ detected when input signal is zero

22/06/2016James Leaver FED Hit Check: Unexpected Non-Zero ADC Values Cluster 1 Event Data Cluster 2 Noise Why does the FED return non-zero ADC values for strips that do not contain valid hits? → Due to data packaging format: i.e. 1 Cluster FED Strip Threshold 1 Value Below Threshold 2 clusters separated by 1 strip: Transmitted as 1 cluster Read out noise level at the strip in- between, instead of zero Non-zero ADC value = ‘false hit’ OR 1 cluster with 1 strip below threshold: Transmitted as 1 cluster Read out below-threshold hit level, instead of zero Non-zero ADC value = ‘invalid hit’

22/06/2016James Leaver FED Hit Check: Unexpected Non-Zero ADC Values(2) The non-zero ADC values transmitted due to data formatting might be intentional… –Or perhaps the Fed9UEvent class could filter them out with a threshold cut… To work around the issue, the FED Hit Check was repeated using randomly generated frames containing only single strip clusters, separated by more than 1 strip –Reduced Tracker occupancy to 1% to account for increased data volume

22/06/2016James Leaver FED Hit Check: Single Hit Results All invalid ‘hits’ have vanished Result: FED is correctly identifying all input hits

22/06/2016James Leaver FED Hit Check: Single Hit Results(2) Results from all 96 FED channels:

22/06/2016James Leaver FED Efficiency Test Wanted to test that FED operates correctly with: –100 kHz random triggers –High Tracker occupancy –Slink readout –FED throttling system in place Also wanted to test FED efficiency (fraction of events lost vs. Tracker occupancy) predictions made by Emlyn –However, the assumptions used in earlier predictions have changed: Emlyn used 2 bytes per hit (strip position, ADC value) Now have cluster finding (cluster position, cluster width, ADC values) Emlyn used back-to-back frames –Could reproduce Emlyn’s conditions, with fixed 2 strip wide clusters and back-to-back frames… –… But our goal is to replicate CMS, so more realistic conditions were used (FED efficiency prediction should still be reasonably accurate)

22/06/2016James Leaver FED Efficiency: Buffering in ZS Mode Large buffers: 80 MB/s FE-BE link dominates Occupancy reaches ~9% before events are lost 140 kHz (b2b 140 kHz (b2b frames) Large buffers: slink dominates Events lost when: –Occupancy ~1.4% at 100 MB/s –Occupancy ~2.8% at 200 MB/s Emlyn’s predictions: → Slink data rate determines FED efficiency

22/06/2016James Leaver FED Efficiency: Experimental Setup Random 100 kHz triggers FED in Zero Suppressed mode Readout rate through Slink limited to maximum of 200 MB/s Sent randomly generated frames with simulated CMS cluster distribution Increased simulated Tracker occupancy of generated frames and recorded fraction of events vetoed by FED

22/06/2016James Leaver FED Efficiency: Cluster Distribution Mean = 2.96 strips per cluster Average size of hit = 1.91 bytes (Emlyn’s average hit size = 2 bytes) Mean = 2.96 strips per cluster Average size of hit = 1.91 bytes (Emlyn’s average hit size = 2 bytes) Simulated cluster distribution was generated from the plots in Figure 4 of Ian Tomalin’s CMS-IN 2005/025 Note: Note: Would clusters of this size really exist?

22/06/2016James Leaver FED Efficiency: Results Events lost when occupancy exceeds ~2.8% with 200 MB/s readout rate

22/06/2016James Leaver FED Efficiency: CRC Errors FED Efficiency Test yields unexplained distribution of CRC errors at high Tracker occupancies…

22/06/2016James Leaver FED Efficiency: Theoretical Vs Measured Data Rates Theoretical data rate from CMS Note 2002/047: –TIB1 ‘Full’ FED (180 of 192 APVs) –Zero Suppressed, 100 kHz triggers, Occupancy of % MB/s → Data rate = MB/s MB/sMeasured data rate at occupancy of 2.9% = MB/s Reason for discrepancy? –Current ‘formatting’ information (non-data) = 816 bytes / event –CMS Note ‘formatting’ information (non-data) = 112 bytes / event –Assume current average data padding = 32 bytes / event –Scale by FED fill factor of 180 / MB/s‘Corrected’ measured data rate = 139 MB/s

22/06/2016James Leaver FED Efficiency: Theoretical Vs Measured Data Rates(2) 139 MB/s‘Corrected’ measured data rate = 139 MB/s Can use average cluster width of generated frames to estimate expected data rate: 132 MB/s → 132 MB/s (good agreement) Calculate average of 1.55 bytes / hit for Ian’s data: –Implies large number of clusters with 4 or more hits –TIB1 cluster distribution in CMS Note must be different to that used in measurement - should account for remaining discrepancy… Conclusion: –A true data rate comparison would require more accurate cluster distribution data for the frame generator –Results show some agreement with Emlyn’s prediction, but cannot demonstrate FEDs performance in final system until we have accurate cluster distribution information and the FED Zero Suppressed ‘Lite’ mode i.e. Current data rates are significantly higher than those expected at CMS

22/06/2016James Leaver Summary FED Crosstalk:FED Crosstalk: –Unpleasant effects when a single frame arrives at the FED out of sync with the others… –…But crosstalk caused by ‘hit’ features is small and should only impact nearest neighbours FED Channel Noise:FED Channel Noise: –Average noise is less than 1 ADC count, with no significant variation from channel to channel FED Hit Check:FED Hit Check: –FED correctly identifies hits in Zero Suppressed mode (for a 2% Tracker occupancy) FED Efficiency:FED Efficiency: –FED vetoes triggers at high data rates in a similar manner to that predicted by Emlyn, but Zero Suppressed ‘Lite’ mode required for genuine CMS performance characterisation –Measured data rates show reasonable agreement with theory, but a more accurate cluster distribution is required

22/06/2016James Leaver Unresolved FED Problems… FED still randomly produces a small fraction of events with CRC errors (~1 in 1-10 million events, depending upon Tracker occupancy) –Possibly a timing issue due to transition to new FPGA tools…?

22/06/2016James Leaver What Do We Test Now? Still need to investigate: –Pedestal variations –Pipeline addresses –Use of TTC –etc. Need to prioritise remaining tests, and check that all important areas are included…