status of much software

Evgeny Kryshen CBM @ SIS100, 21 May 2009 1

Evgeny Kryshen (PNPI)Mikhail Ryzhinskiy (PNPI & SPbSPU)

Status of MuCh softwareStatus of MuCh software

Outline

Implementation of realistic module design Segmentation algorithms Digitization algorithm Hit finding algorithm Visual display for MUCH


Old geometryOld geometry

• Stations are simulated as simple shapes – 3mm disks filled with argon gas

• Distances between layers and absorbers are not realistic (too small)

• No support structures, no module design

• Parameter files are huge, full of irrelevant numbers, hardly readable, not flexible


Module design in ALICEModule design in ALICE


Schematic layout of GEM moduleSchematic layout of GEM module

Support structure

PadsPads

FastenersFasteners

SpacerSpacer

PCBPCB Readout electronicsReadout electronics

ArgonArgon

GEM foilsGEM foils


Implementation of module designImplementation of module designFront side Back side

Layers:

Modules are arranged in rows on both sides of each layer

There is an overlap of active volumes to avoid dead zones in y direction

Overlap


Support structures and modulesSupport structures and modulesSupport structures:

Each support structure is composed of two parts to assure easy installation around the pipe

Estimated thickness ~ 1.5 cm

Material: carbon plastics (ρ = 0.1 ρC)

Implemented as composite shapes in cbmroot

Y spacersY spacers

Active volumeActive volume

X spacersX spacers Module:

Module size is mostly restricted by the GEM foil production technology

Active volume: 256 x 256 mm x 3 mm, argon

Spacers: 5 cm in y, 0.5 cm in x; material: noryl, implemented as composite shapes

Active volume implemented as TGeoBox for simple modules and as composite shapes for modules with a hole


Detailed geometry: general viewDetailed geometry: general view

• Two layers at each station• Three layers at the last trigger station• Modules are automatically located on the surface of support structures• Cables, gas tubes, PCBs and front-end electronics are neglected at the moment

Module design Simple design Straw design


Geometry input file: much_standard_straws.geoGeometry input file: much_standard_straws.geo

# General informationMuchCave Zin position [cm] : 105Acceptance tangent min : 0.1Acceptance tangent max : 0.5Number of absorbers : 6Number of stations : 6

# Absorber specificationAbsorber Zin position [cm] : 0 40 80 120 170 225Absorber thickness [cm] : 20 20 20 30 35 100Absorber material : I I I I I I

# Station specificationStation Zceneter [cm] : 30 70 110 160 215 340Number of layers : 2 2 2 3 3 3Detector type : 1 1 1 2 2 2Distance between layers [cm]: 10 10 10 7 7 7Support thickness [cm] : 1.5 1.5 1.5 0.0 0.0 0.0Use module design (0/1) : 1 1 1 0 0 0

# GEM module specification (type 1)Active volume lx [cm] : 25.6Active volume ly [cm] : 25.6Active volume lz [cm] : 0.3Spacer lx [cm] : 0.5Spacer ly [cm] : 5Overlap along y axis [cm] : 2

# Straw module specification (type 2)Straw thickness [cm] : 0.4


Class hierarchyClass hierarchyCbmMuchGeoSchemeCbmMuchGeoScheme

CbmMuchStationCbmMuchStation

CbmMuchLayerCbmMuchLayer

CbmMuchModuleCbmMuchModule

CbmMuchLayerSideCbmMuchLayerSide

CbmMuchSectorCbmMuchSector

CbmMuchPadCbmMuchPad


Automatic segmentation: algorithmAutomatic segmentation: algorithm

Hit density vs R

x

y

// Set minimum allowed resolution for each stationDouble_t sigmaXmin[] = {0.04, 0.04, 0.04, 0.04, 0.04, 0.04};Double_t sigmaYmin[] = {0.04, 0.04, 0.04, 0.04, 0.04, 0.04};seg->SetSigmaMin(sigmaXmin, sigmaYmin);

// Set maximum allowed resolution for each stationDouble_t sigmaXmax[] = {0.32, 0.32, 0.32, 0.32, 0.32, 0.32};Double_t sigmaYmax[] = {0.32, 0.32, 0.32, 0.32, 0.32, 0.32};seg->SetSigmaMax(sigmaXmax, sigmaYmax);

// Set maximum occupancy for each stationDouble_t occupancyMax[] = {0.05, 0.05, 0.05, 0.05, 0.05, 0.05};seg->SetOccupancyMax(occupancyMax);


Automatic segmentation: resultsAutomatic segmentation: resultsModule designModule design

Sector sizes at the first station are mostly determined by occupancy restrictions

Starting from the 3rd station sector sizes are determined by the required resolution

The smallest pad size in the default setup is ~2 mm (resolution ~ 600 μm).

Simple designSimple design


New flexible (manual) segmentationNew flexible (manual) segmentation

// Number of regions for each stationInt_t nRegions[] = {5, 3, 1, 1, 1, 1};seg->SetNRegions(nRegions);

// Set region radii for each stationDouble_t st0_rad[] = {13.99, 19.39, 24.41, 31.51, 64.76};seg->SetRegionRadii(0, st0_rad);…// Set minimum pad size/resolution in the center region for each stationDouble_t padLx[] = {0.1386, 0.4, 0.8, 0.8 ,0.8, 0.8};seg->SetMinPadLx(padLx);

Module designModule design Simple designSimple design

Developed by M. Ryzhinskiy


Digitization algorithm

Primary electrons:1.Number of primary electrons is generated according to Landau distribution (MPV and sigma taken from HEED) 2.MPV and sigma are calculated for electrons, muons and protons. For other particle types we use mass scaling.

Secondary electrons:1.Exponential gas gain distribution with mean value of 104 sec. electrons/prim. electron2. Sec. electrons are projected on pads in a circle with a given spot radius (0.3 mm for MM, 1.5 mm for GEM)

Factors not taken into account:1.Transverse diffusion of primary electrons is not accounted for2.Cluster nature of primary electrons

Charge thresholds:1. Maximal charge for muon track: 4x105 electrons/pad2. For 256 channel ADC one has 1.5x103 electrons/channel3. Minimum charge threshold: 3 channels


Charge distribution

Factors contributing to the charge dispersion: Particle type Particle energy Track length variation Number of “primary” electrons generated according to Landau distribution with a given MPV

and sigma (dependent on Particle energy and type) Gas gain fluctuations in accordance with exponential distribution with mean value of 10000

Mean charge: 3.6 10∙ 5 (In average 36 primary electrons)


Energy dependence of the charge

X axis – decimal logarithm of track energy measured in MeVY axis – charge generated by track (number of secondary electrons)

The sharp cut-off at Log E equal to 0 ( or equivalently 1 MeV) is due to the geant3 minimum energy cut


Energy dependence of the charge

Solid lines correspond to MPV energy dependencies built in the simulation (MPV curve is proportional to Bethe-Bloch in the first approximation)

These plots demonstrate the consistency of the simulationElectrons are most sensitive to 1 MeV cut-offDetailed studies of the electron cut-off dependency are desired

Charge vs energy distributions for different particle types:


Charge vs. track length• Sensitive gap of the

detectors is 3 mm

• Difference in the track length is caused by the track slope

• The large track length is usually caused by secondaries

• Track lengths smaller than 3 mm are due to edge effects

• Mean length for electrons: 4.5 mm

• Mean length for protons: 3.5 mm


Illustration of fired pads


Cluster deconvolution

Qmax

Q

Qthr

Primary clusterpads

Qthr(Qmax) = 0.1Qmax

12

)(;

12

)( minmin lengthpady

widthpadx hithit

iPadiPad

iPadiPadiPad

hit

iPadiPad

iPadiPadiPad

hit q

qyy

q

qxx ;Hit coordinates:

Hit errors:


Cluster statistics

Mean number of generated MC points contributing to one cluster: 1.16Mean number of fired pads in one cluster: 2.33Mean number of reconstructed hits produced in one cluster: 1.02


Hit finding results

inactive pads

fired pads

traces from MC tracks

reconstructed hits

–

3000 MC points/event


Fake hits

Fake hits – number of reconstructed hits is larger than the number of tracks which formed the cluster 0.3%


Lost hits

Lost hits – number of reconstructed hits is less than the number of tracks which formed the cluster

10.1%Conclusion: the naive hit finding algorithm should be improved


Visual Event Display: Layer viewVisual Event Display: Layer viewLayer view functionality:

• Switch between stations and layers

• Info on stations

• Zoom

• Show info on hits, points and sectors

• Switch off sectors, modules, layer sides, hits and points

• Select particles with required PDG code and mothers

• Browse events

• Clickable sectors producing zoomed module views


Visual Event Display: Module ViewVisual Event Display: Module View

Zoomed module viewZoomed module view

Fired pads are marked with blue gradient colors reflecting the accumulated chargeFound hits are marked with black markers


Visual Event Display: Cluster ViewVisual Event Display: Cluster View

Cluster view can be opened by clicking on a cluster in a module frame.It is aimed to help in optimization of hit finding algorithms.


Future plansFuture plans

• Implementation of straw digitization and hit finding• Implementation of “mixed” stations with modules of different types• Optimization of the software with respect to tracking requirements• Optimization of digitization parameters and detector layout • Development of advanced cluster deconvolution algorithm• Further development of visualizer

status of much software

Documents

module size

support thickness

number of layers

gem module specification

active volume lx

active volume ly

use module design

absorber thickness