VELO Software Overview Shutdown Planning Organisation

VELO Software Overview & Shutdown Planning • Organisation • Milestones • 3 Critical Areas Chris Parkes for VELO software Group

Areas & Responsibilities • • • • Overall Co-ord: CP PVSS: Stefano De Capua DAQ Recipes: Karol Hennessy Timing & Gain: Kazu Akiba Error Bank Analyses: Ann van Lysebetten Online Monitoring: Kurt Rinnert Data Quality: Eduardo Rodrigues Simulation & Reconstruction: Tomasz Szumlak Tracking: David Hutchcroft Alignment: Silvia Borghi Closing Strategy: Malcolm John > 20 people contributing Milestones defined for each, with one person responsible and priorities assigned 2

Organisation • Weekly Monday commissioning meeting – Report on previous week milestones – News from all, forum for discussing issues – Work plans for the week • Integration Days: Thursdays – Integrate weekly releases (if any) at pit – Release: PVSS, recipes, Vetra • Brief report bi-weekly Friday meeting – Report progress to whole group, no details – Specific presentations on items of general interest • Shutdown progress logged on milestone Twiki page https: //lbtwiki/bin/view/VELO/Software. Milestones 3

Milestone Progress • Proceeding close to schedule • Some delays due to FEST’ 08 production 4

Critical Path • In September identified three key areas where progress is needed before we start running this year – Timing – TELL 1 Parameter Uploading – Monitoring 5

Timing studies • Set up timing for sampling of pulse train and for optimal analogue signal height • Automated timing scans implemented and being tested • Firmware release being tested 6

Digitisation Delayed Sampling phase ADC Counts Delayed pulse 3 ARx Clock 5 4 Measured Voltage 6 Time (CLK/channels) 7

Analogue Sampling Delay Scan Points ADC Counts Sampled points for a given clock Delayed pulse 25 Beetle Clock First Time Sample 75 50 Second Time Sample Third Time Sample 100 Time (ns) Fourth Time Sample 8 . . .

TELL 1 Data Processing • Velo Data Processing Raw -> Clusters in TELL 1 RAW • Require 1 M parameters Pedestal Following • Optimisation critical for data quality (see TED data talk) Beetle Cross-talk Correction Cable Cross-talk Filter (FIR) Lower priority Common Mode Suppression (MCMS) Beetle baseline shift Lower priority Reordering Common Mode Suppression (LCMS) • Pedestal & Clusterisation Thresholds most important • Bit Perfect Emulation of Algorithms in full LHCb Software Framework Clusterization CLUSTERS 9

Vetra – TELL 1 Emulation • Parameter uploading achieved for first time in December • Firmware fixes made and used (November) • Testing & evaluation underway 10

Pedestal Processing Raw data Pedestal corrected data Pedestal correction monitoring Base line (zero) level after pedestal correction 11

Beetle X-talk correction Beetle X-talk effect first channel in each analog link is affected Measured noise in first channel before correction Noise after pedestal correction Beetle X-talk correction monitoring Noise in first channel after correction Average noise measured in unafected channel 12

Effect of tuning ADC count Constant Pedestal Only ADC count All Parameters tuned Non-Zero Suppressed Data critical – so that tuning parameters can be obtained Procedure to take automatically during data - One module at a time, under test 13

Monitoring • Monitoring package – Package for “high-level” (= ZS) data • Monitoring based on clusters and tracks – Package for NZS data • Noise calculation, time alignment study, beetle pulse shape, … • Scripts and macros are being developed to analyse data • Wiki pages with documentation and How. To’s Review of Monitoring status in February 14

Online monitoring • Running since August • Implementation of several plots • New features to be exploited Online presenter 15

Cluster Monitoring • Cluster information: – Cluster ADC value – Active chip links – Number of strip in a cluster – Cluster ADC value versus sampling – Number of cluster per events – More… Some of these distributions versus sensor number and/or sensor strip 16

Example distributions from Velo. Rec. Monitors 17

Track Monitoring • Tracks – Number tracks – Pseudo-rapidity – Azimuthal Angle – Pseudo-efficiency – Biased and unbiased Residuals versus sensor number – Total number of R cluster per track – Vertex information – Hits distribution in xy and xz – Mean sigma of residuals versus of sensors – More… 18

Track monitoring: J/ ( ) Ks Pseudo rapidity Biased. Res X(cm) Azimuthal angle Sensor # R Z(mm) 19

Script and Macros • Analysis of the data for the evaluation for: – – – Time alignment study, Noise calculation, High voltage scan, beetle pulse shape, More… 20

Noise monitoring macros – example of GUI 21

Noise monitoring macros – example of GUI Common mode subtraction No common mode 22

Noise performance • Common mode pickup from beam requires data • At pit and in previous testbeams parameters highly stable HP 1 HP 2 HP 3 HP 4 23

Noise – individual / whole system No evidence that operation of full system induces more noise than single sensors 24

Expected Signal / Noise versus Voltage 25

IV - scans • PVSS recipes available to automate IV scans • Set initial voltage, target voltage, step, single or set of sensors • A data file produced per sensor containing channel number, voltage, current, sensor temperature • Analysis scripts for plotting IV scan data 26

Software Commissioning Summary • All baseline algorithms – Completed for summer ’ 08 • Commissioning software – Milestones for data readiness in April 2009 – 3 critical areas all proceeding according to plan • TED data are the VELO ‘cosmics’ Tremendous success of first tracks This sample has been very useful for comissioning TED data this summer will allow us to: Optimise timing Test and tune FPGA algorithms Increase alignment accuracy 27