- Poisson CCD22 solver by Craig Lage - PowerPoint PPT Presentation


 
 Documenting and validating imSim sensor models 
 Sergey Karpov 
 Institute of Physics, Czech Academy of Sciences 
 Sensor Characterization, Analysis and Simulation 
 Aug 14, 2018 #lsst2018 #lsst2018 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17 �1 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

imSim - Poisson CCD22 solver by Craig Lage https://github.com/craiglagegit/Poisson_CCD22 • Low-level simulation of electric fields and charges in silicon • Most detailed and advanced CCD simulation • Extremely slow - GalSim https://github.com/GalSim-developers/GalSim • Modular astro image simulation toolkit • Simplified / approximate low-level sensor effects - imSim https://github.com/LSSTDESC/imSim • LSST-specific configs and scripts on top of GalSim #lsst2018 �2 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Documenting imSim models #lsst2018 �3 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

imSim sensor model - Electron Diffusion • GalSim: Silicon.cpp + sensor.py • random displacement with amplitude dependent on Vbb, temperature, conversion depth - Tree Rings • GalSim: Silicon.cpp / imSim: tree_rings.py • Static distortion map with randomized radial profile - Brighter-Fatter • GalSim: Silicon.cpp • Dynamically adjusted distortion #lsst2018 �4 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Electron Diffusion - Model • Diffusion + electric field drift • Effective reduction of electron mass (~0.27) following Green et al (1990) • Front voltage adjustment based on simulations (~1 um above bottom) • Scaling with (wavelength + angle dependent) conversion depth 1 − ConversionDepth 2kT σ = SensorThickness ⋅ SensorThickness ⋅ q(Vfront − Vbb) ⋅ MobilityFactor #lsst2018 �5 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Electron Diffusion https://github.com/LSSTDESC/imSim/wiki/diffusion_validation - Validation • Fe55 data acquired at BNL • NGMIX fitting of well isolated hits • Model fit for surface conversion value #lsst2018 �6 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Electron Diffusion Model is mostly consistent with experimental data #lsst2018 �7 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings - Doping variations during wafer growth - Voltage (and conversion depth) dependent electron drift - Seen in both E2V and ITL sensors, amplitudes similar - Photon Displacement - Shape distortion - Pattern in flats #lsst2018 �8 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings: imSim model - Radial displacement map based on random Fourier components δ r = ( ∑ sin( ω i r + ϕ i )/ ω i − ∑ cos( ω j r + ϕ j )/ ω j ) ⋅ ( A + Br 4 ) δ S = ( ∑ cos( ω i r + ϕ i ) + ∑ sin( ω j r + ϕ j )) ⋅ ( A + Br 4 ) • 15 components: Gaussian(60, 10) pix • 5 frequencies: Gaussian(35, 10) pix - Qualitatively based on profiles measured in 60% "good" 40% "bad" BNL (Hye-Yun Park) - Pre-loaded at sensor construction - Scales with conversion depth - Does not scale with bias voltage #lsst2018 �9 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings Validation https://github.com/LSSTDESC/imSim/wiki/tree_ring_validation BNL data imSim #lsst2018 �10 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings Validation: periodograms BNL data imSim #lsst2018 �11 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings Validation: frequency structure BNL data Real tree rings have faster oscillations at larger radii imSim #lsst2018 �12 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Tree Rings Validation: amplitudes BNL data BNL data imSim imSim #lsst2018 �13 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Brighter-Fatter model - Exact simulation of electron repulsion for single pixel using Craig' Poisson solver - Vertices displacements imported into GalSim and co-added for all pixels • linear scaling with pixel charge - Pixels are assigned based on displaced edge vertices • non-linear conversion depth scaling - Iteratively repeated while collecting the image #lsst2018 �14 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Brighter-Fatter validation https://github.com/LSSTDESC/imSim/wiki/brighter_fatter_validation - Inter-pixel correlations in simulated flats - Direct spot size dependence on flux Poisson GalSim #lsst2018 �15 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17

Plans - Improve Wiki documentation • reflect all implementation details • demo notebooks for effects visualization / characterization • general overview pages for different subsystems - Validation and model adjustments • Diffusion: acquire homogeneous dataset for a single sensor • Tree Rings: modify the model to include more realistic frequency and radial structure • Brighter-Fatter: simulation vs experiment? #lsst2018 �16 LSST Project and Community Workshop 2018 • Tucson • August 13 - 17