CSPADs: how to operate them, which performance to expect and what kind of features are available Gabriella Carini, Gabriel Blaj, Philip Hart, Sven Herrmann
Cornell-SLAC Pixel Array Detector What is it? An integrating pixel array detector with readout speed of 120 frames per second. Why is it different from Pilatus? Pilatus is a counting pixel array detector: if your photons arrive all at once (like at LCLS) it will always count 1. What does it mean v1.0, v1.2, v1.5 & v1.6? They do represent different upgrades of the detector with corresponding improved performance. What does it mean 140k? And 2.3M? They are the number of pixels per camera. What do I need to know to use it? Very little details and settings should be discussed with the point-of-contact (POC). What can I learn about it? As much as you like! 2
Cornell-SLAC Pixel Array Detector Schematic of a CSPAD pixel reset HV bias gain ramp comparator reset read EN IN S&H fire logic counter data bus Vref ADC clk Koerner L J, Philipp H T, Hromalik M S, Tate M W, and Gruner S M 2009 JINST 4 P03001 Philipp H T et al. 2010 IEEE Trans Nucl Sci 57 3795 Philipp H et al. 2011 Nucl Instr Meth Phys Res A 649 67 Hart P A et al. 2012 Proc SPIE 8504 85040C Hart P A et al. 2012 IEEE NSS MIC Conference 538 Herrmann S C et al. 2013 Nucl Instr Meth Phys Res A 718 550 Herrmann S C et al. 2012 IEEE NSS MIC Conference 520 Herrmann S C et al. submitted to JPCS (SRI 2013) 3
Cornell-SLAC Pixel Array Detector Schematic of a CSPAD pixel reset HV bias gain ramp comparator reset read EN IN S&H fire logic counter data bus Vref ADC clk Koerner L J, Philipp H T, Hromalik M S, Tate M W, and Gruner S M 2009 JINST 4 P03001 Philipp H T et al. 2010 IEEE Trans Nucl Sci 57 3795 Philipp H et al. 2011 Nucl Instr Meth Phys Res A 649 67 Hart P A et al. 2012 Proc SPIE 8504 85040C Hart P A et al. 2012 IEEE NSS Technical MIC Conference papers!!! 538 Herrmann S C et al. 2013 Nucl Instr Meth Phys Res A 718 550 Herrmann S C et al. 2012 IEEE NSS MIC Conference 520 Herrmann S C et al. submitted to JPCS (SRI 2013) 4
Cornell-SLAC Pixel Array Detector Schematic of a CSPAD pixel CSPAD properties Pixel size Chip HV area bias Maximum signal Frame rate Noise IN Vref gain reset S&H 110 µm x 110 µm comparator read 185 x 194 pixels reset per ASIC (full reticle EN size) 2700 8keV photons/pixel (low gain) 350 8keV photons/pixel fire (high gain) data counter 120Hz logic bus ~ 3.5 kev (low gain), ~ 1 kev (high gain) ramp ADC clk Koerner L J, Philipp H T, Hromalik M S, Tate M W, and Gruner S M 2009 JINST 4 P03001 Philipp H T et al. 2010 IEEE Trans Nucl Sci 57 3795 Philipp H et al. 2011 Nucl Instr Meth Phys Res A 649 67 Hart P A et al. 2012 Proc SPIE 8504 85040C Hart P A et al. 2012 IEEE NSS Technical MIC Conference papers!!! 538 Herrmann S C et al. 2013 Nucl Instr Meth Phys Res A 718 550 Herrmann S C et al. 2012 IEEE NSS MIC Conference 520 Herrmann S C et al. submitted to JPCS (SRI 2013) 5
CSPAD cameras assembly 2 ASICs 110um pixel size 185 x 388 pixels/sensor rigid-flex PCB holds 2 ASICs 2 PCBs holding 4 ASICs are glued on an aluminum carrier 2 flex leads connect to the support electronics CSPAD: basic module assembly Detail of CSPAD 2.3Mpixel camera 6
CSPAD cameras history Used at LCLS: CXI, XPP, XCS, MEC: during the first 4 years of operation has become the workhorse of the LCLS detectors It has been implemented in several versions and cameras with different area and shape 2.3Mpixel and 140kpixel cameras Vacuum compatibles Used at SACLA, SSRL and APS CSPAD-140k CSPAD V1.0 CSPAD V1.2 CSPAD V1.5 & V1.6 2010/11 XPP CXI 2012/13 2013/14 XPP MEC XPP MEC CXI XCS CXI XCS epix10k 7
Some improvements from v1.0 to v1.2, v1.5 and v1.6 relative noise V1.0 Thermo-mechanical Electronics Firmware V1.2 ASIC Electronics Firmware V1.0 V1.5 CSPAD amplitude / pixel [ADU] 2000 1800 1600 1400 V1.2 1200 1000 800 600 400 200 0 0 200 400 600 800 1000 integration time [us] V1.5 Electronics Firmware V1.6 Linearity at high gain Measurements performed at SSRL BL 2-2 8
Detector for our users facilities and our user facilities for detector development Active proposal for beamtime at SSRL Important to complement first characterizations with x- ray tube a) b) Linearity measurements with the CSPAD-140k v1.2 at BL 2-2. Improved electronics resulted in good linearity both at high (a) and low gain (b). CSPAD 2.3M v1.0 and mobile DAQ at SSRL BL 2-2 CSPAD 140k v1.5: energy scans. Measurements at room temperature. Histograms of all pixels with pedestal and common mode correction (no per-pixel correction) 9
LCLS: different working conditions Beam instability: required improved diagnostics* In red: intensity distribution measured with the gas detector. In black: intensity distribution downstream the monochromator measured with a PIPS diode (Canberra Inc.) and in-house developed readout electronics. Correlation plots: CSPAD vs beam monitor. After optimization of the inhouse electronics used to measure beam position and intensity. *Herrmann S C et al. submitted to JPCS (SRI 2013) 10
Non-linearity effect at high fluence MEC CSPAD 140k V1.0 Bismuth high pressure phase transitions PI M. McMahon Courtesy of Hae-Ja Lee, Philip Heinman 11
Non-linearity effect at high fluence a) b) c) Linearity response of the CSPAD in low gain. a) Standard setting. b) Different reference voltages at the preamplifier. The flat-top of the response is generated by the ADCs. c) Different sensor bias voltages. a) b) c) Simulation of the response of the CSPAD to increasing charge at the input. a) 10ns charge collection time; b) 100ns charge collection time; 200ns charge collection time. 12 G. A. Carini et al. submitted to JPCS (SRI 2013)
Starting up 13
Important to remember Integrating Pixel Array Detectors need: Dark correction Frame common mode correction Mask bad pixels Gain calibration Cross talk correction Geometry reconstruction Moving towards integration of all the above in online and offline AMI and data analysis (see data workshop). In the future we will provide raw data and corrected data. 14
Typical examples for optical CCD - astronomy http://starizona.com Dark frame Bias frame Flat field 15
Typical examples for optical CCD - astronomy http://starizona.com An uncalibrated image of M51, the Whirlpool Galaxy ( raw data ). A calibrated and enhanced version of the M51 image. 16
Dark correction What is it? In general it is a frame without illumination, hence dark. Its signal will have many contributions (i.e. leakage current) that averaged can be corrected. 17
Frame common mode correction uncorrected: 5.6 r.m.s corrected: 4.4 r.m.s. 18
Mask bad pixels 19
Gain calibration What is it? Is it always needed? Blue = 58 Red = 76 Blue = 56 Red = 75 34% Two adjacent pixels: V1.5 Two adjacent pixels (only singles): V1.5 Blue = 68 Red = 72 Blue = 67 Red = 71 5% Two adjacent pixels: V1.6 Two adjacent pixels (only singles): V1.6 Extreme case: pixels were randomly chosen to show a significant gain difference 20
Cross talk correction V1.5 Cross-talk seen with correlation plots: Intensity measured with the CSPAD vs intensity measured with the beam monitor. On the left vertical axis the values of the baseline shift of a shielded area when an adjacent area is illuminated with strong signal (right vertical axis). 21
Geometry reconstruction Silver behenate rings: before (left) and after (right) reconstruction. Single event image. 22
CSPAD in science our goal is to help making the best use of detectors to produce new science! CXI CSPAD 2.3M Lysozyme structural model against its x-ray diffraction pattern using CSPAD at CXI* *Boutet et al. Science, 337 (6092), 362 (2012) 23
Multi module configurations ( example @ CXI) take multiple 140k s put them into the form factor you prefer cable it up let the DAQ run -> the compact, modular design makes it easy to cover specific form factors 24
CSPAD @ XPP Femtochemistry Solution scattering Courtesy of Henrik Lemke Scattering from 100 µm thick liquid jet of water CSPAD (1.2) operating at low gain 25
Gain nonlinearity and ASIC artifacts Measured signal is function of intensity, because of varying gain. --> Pump probe: relative changes for intensity filtered pulse ensemble Remaining artifacts by e.g. noisy pixels have to be filtered as they are of similar size as the pump/probe signal. 26
Diffraction from nanowires Bragg diffraction from nanowires, CSPAD area detector 120 Hz readout grown epitaxially from a substrate CSPAD 2.3M v1.2, high gain Acoustic nanoscale resonances in InAs nanowires, SRI 2012, lemke@slac.stanford.edu 27
Gain nonlinearity at low gain Measured signal is also here function of intensity. NB: high fluctuations due to monochromatic beam à Better performance with CSPAD-140k v1.5 à No visible ASIC artifacts 28
CSPAD-140k @ MEC (and another example of a multi module configuration @ MEC) ptychography data courtesy of Andreas Schropp MEC spectrometer design can use either a (slow) CCD camera or CSPAD-140k Small mechanical modification makes the CSPAD-140k fit -> the modular PCB concept and the separate flex-lead connected detector carrier make such variants possible without expensive redesigns 29
CSPAD-140k @ SSRL (preparation of XPP experiments) -> helped to prepare an XPP experiment with beamtime at SSRL (spectrometer alignment and pre-studies of samples) Used also for experiments at APS and SACLA 30