SciFi Tracker Test Beam July 2018

Analysis

There are three separate frameworks for test beam analysis, but produce similar results:

A Kepler based framework run in Gaudi can be found here (most recently maintained by Li Xu):

1. https://gitlab.cern.ch/lhcb/Kepler/-/tree/tbscifi_xuli

a standalone non-Keplar framework can be found in two different flavours:

2. https://gitlab.cern.ch/lhcb-scifi/testbeam/tb2018_1 with multiple branches

3. https://gitlab.cern.ch/rivello/tb2018-analysis + https://gitlab.cern.ch/rivello/tb2018-event-matching (aligns telescope and minidaq events) maintained by Mauricio Feo.

For the differences between the two frameworks:

Kepler framework:

1. Use a common fine time selection: fine time < 1500 or fine time > 3000 for all the runs

2. Cluster size <= 20

3. For the rotation alignment, use “MIGRAD” and “HESSE” to minimize

4. For the residual fit, use the sum of 2 gaussian models of roofit to fit the dataset

•bunch crossing id < 50 or bunch crossing id > 60
•chi^2/ndf of track > 0.5 and chi^2/ndf of track < 2
•number of clusters is 1 (for resolution studies)

Non-Keplar framework:

1. Find the 90% fine time window automatically for each run

2. No cut for the cluster size

3. For the rotation alignment, use “SIMPLEX” and “HESSE” to minimize

4. For the residual fit, use TF1 to fit the histogram

[&](double *x, double *p){return p[1] * TMath::Gaus(x[0], p[0], p[3]) + p[2] * TMath::Gaus(x[0], p[0], p[4]);}

Logbook

• Runlogbook.ods: Logbook of the settings for each recorded run

Instructions for Shifters

• yes
• Test beam is two words
• Velopix_Telescope_Instructions.txt: instructions on how to start and stop the TimePix3 telescope for shifters
• CAENpowersupply_instructions.txt: how to open to CAEN power supply interface
• MiniDAQ2_Operation_Instructions.docx: MiniDAQ2 Operation Instructions

Practical Information

North Area

info copied from 2016 test beam:

All practical information for users of the North Area can be found in the North Area Handbook.

The layout of the H8A area can be seen in the EHN1 experimental hall drawings.

The area is approximately 6 metres wide and the height to the beam line is 50 inches (127 cm).

Access to the area

Aim of the Testbeam

1. see clusters correlated to tracks from the half-ROBs.

2. Hit-efficiency for (sorted with priority):

• different thresholds at one position (close to mirror)
  • planned for the low standard thresholds, we try to simulate
  • higher thresholds before to see which would be useful
• threshold scans with tracks not feasible, LY too high
• whole width of this mat (LHCb x)
• one position in each mat (still close to mirror)
• three positions along one mat (along fibre direction, LHCb y)
• different angles one mat, one position
  • not as easy as it was in the past, but should be possible: 0°, 5°, 10°, 20°
• one position different over voltages (lower PDE up to 50%)
  • different over voltages planned, but probably not possible to lower
    pde significantly (not possible to measure gain anymore)
• this + different thresholds
• all of this for all mats...

Setup

• for the modules used, see below

Position Definition

A file describing the layout of the modules, readout boxes and fibre links is available in the attachments: https://twiki.cern.ch/twiki/pub/LHCb/SciFiTrackerTestBeamJul2018/Modules_Fibres_Layout.pdf

channel numbers in test beam, cosmic measurement and single mat Sr90 setup: SciFiTestBeamJul2018ChannelNumbering.pdf

Data Backup

minidaq files /eos/lhcb/testbeam/scifi/tb2018_1/

• telescope files /eos/lhcb/testbeam/velo/timepix3/July2018/

MiniDAQ2 Data Format

ROOT-Files

The MiniDAQ2 output ROOT-files contain the following information (extract):

1. A directory "Geom" with one TTree per module, i.e. ROB0 and ROB1. These trees contain the branches:

• ROB_ID: ID of this ROB (0 or 1)
• Orbit_ID: NOT FILLED HERE PROPERLY - DO NOT USE!
• Evt_ID: The index of this event. Counting up during filling.
• Fragment_ID: Counter from the TELL40 assigned to each fragment. Should match with Evt_ID!
• BXing: Bunch crossing ID. Range: [0,3564)
• ConsecBXing: Counter for incrementing BXing (reset to 0 when not incrementing by 1) - especially of interest for TAE data
• absBXing: NOT FILLED HERE PROPERLY - DO NOT USE!
• Time: NOT FILLED HERE PROPERLY - DO NOT USE!

2. In case where the new MD2 firmware with ODIN information was used (runs 30434 and above), an additional tree called "ODIN" with branches:

• UTC_TIME: UTC timestamp in nanoseconds at start of run and then incrementing in steps of 25ns... BETTER USE absBXing AS TIMESTAMP!
• Fragment_ID: See above. Should match with Fragment_ID of ROB0/1!
• Orbit_ID: Orbit counter (1 Orbit = 3564 BXings)
• BXing: see above.
• absBXing: Calculated as Orbit_ID*3564+BXing. Can serve as timestamp (25ns per step)!

Fragment (.frg) files

Common header

• 99: Raw data bit (1 for raw, 0 for cluster), previously truncate bit
• 98: Parity bit of 19 other header bits
• 97: =0 data, =1 for TFC command
• 96-92: Number of clusters

Cluster Data (bit 99 is set)

• 91-90: Reserved
• 89-81: Cluster 1 Data
• 80-72: Cluster 2 Data
• 71-63: Cluster 3 Data
• 62-54: Cluster 4 Data
• 53-45: Cluster 5 Data
• 44-36: Cluster 6 Data
• 35-27: Cluster 7 Data
• 26-18: Cluster 8 Data
• 17-9: Cluster 9 Data
• 8-0: Cluster 10 Data

Each cluster occupies 9 bits:

• 8-2: Integer cluster position per SiPM (128 channels)
• 1: If set, adds 0.5 to the integer cluster position
• 0: "Big" cluster flag (see below)

Two different cluster FPGA firmware (FW) revision were used for the test beam: 18062521 and 18071120. Revision 18071120 applies to all runs with data from both modules, while revision 18062521 was used for the runs with only one module.

Both FW revisions give the same position for "normal" clusters which are constructed from 1 to 4 channels: The given position is computed as the barycenter of the involved channels, weighted by the number of passed thresholds for each channel.

However, they showed slightly different behaviour with respect to dealing with "big" clusters which refer to clusters constructed from 5 or more channels: Internally, "big" clusters are constructed in blocks of 4 channels. For "big" clusters, two positions are provided in the data. The first one being the position of the first block of 4 channels, while the second one (with the "big" flag set) being the position of the last block (consisting of 1-4 channels depending on the overall size of the "big" cluster).

• First/lower block (same for both FW revisions): Weighted barycenter of the 4 channels belonging to this block (same as for a "normal" cluster with 4 channels)
• Last/higher block (FW rev. 18062521): Weighted barycenter of the last block (up to 4 channels)
• Last/higher block (FW rev. 18071120): Position of the first involved channel of the last block plus half the size of the block (=Unweighted barycenter + 0.5)

All positions are rounded down to the nearest .5 value.

Against intention, it has been found that both FW revisions considered any channel passing at least one threshold a cluster, which leads to a lot of noise-clusters in the data.

Raw Data (bit 99 is not set)

• 91: TFC Sync
• 90: TFC HdrOnly
• 89: TFC NZS
• 88: TFC SnapShot
• 87: TFC FERst
• 86: TFC Calib(1)
• 85: TFC Calib(0)
• 84: TFC BcntRst
• 83-76: "0xCD"
• 75-72: ChBlock (0 for Chs 0-31, 1 for Chs 32-63, 2 for Chs 64-95 or 3 for Chs 96-127)
• 71-64: "0x00"
• 63-62: Raw data Ch 31+ChBlock*32
• 61-60: Raw data Ch 30+ChBlock*32
• ...
• 1-0: Raw data Ch 0+ChBlock*32

Timepix Telescope Data Info

offline Data structure: https://lbtwiki.cern.ch/bin/view/VELO/Tpx3TestbeamSoftware

test beam modules (4-mat half-modules)

• entry in the SciFi database:
  • KI module
  • DO module
• the eight used mats were previously measured in Dortmund
  • Note: FibreMatsForTestBeam2018.pdf
  • data can be found here: TestBeamMats2018

Trigger Setup Scheme

-- RobertEkelhof - 2018-07-02

• modulevideo.mp4: a 2- second video of the module</verbatim>