GeoSoilEnviroCARS The University of Chicago 4,000 Spectra or 4,000,000 ROIs per Second: EPICS Support for High-Speed Digital X-ray Spectroscopy with the XIA xMap Mark Rivers, Center for Advanced Radiation Sources Motivation Need a cost-effective way to collect XRF spectra from multi-element detector arrays Modern detectors, particularly silicon drift diodes (SDD) can run at >250,000 cps per detector, or >1,000,000 cps for a 4-element array like the quad Vortex Depending on the application, can thus get a usable signal (1,000 counts) in 1 ms. Need to keep the overhead less than that! EPICS Software EPICS software allows complete control of the xMAP configuration and acquisition from any beamline control system. DXP 3-0 is a complete rewrite of the EPICS support for XIA modules (Saturn, DXP2X, xMAP) DXP parameter control no longer uses special DXP record, uses standard EPICS records (ao, ai, bo, bi, etc.) Based on asynPortDriver, originally written for EPICS areaDetector module Single driver that supports the MCA record, high-level and low-level DXP parameter control with standard EPICS records, and NDArray callbacks for file-saving plugins in mapping mode MCA Record Support Uses standard asyn device support for the MCA record, same as Canberra AIM, SIS multichannel scaler and other MCA/MCS hardware Start and stop data acquisition, readout the spectra, and control and read the data acquisition time. Supports up to 32 Regions of Interest (ROIs) for computing the net or total counts in each fluorescence peak. MCA record ROIs can be defined graphically in IDL or Python clients, copied to the internal xMAP SCA definitions. Used for individual spot analyses or slow mapping DXP Parameters The xMAP and other XIA modules have about 12 high-level parameters and nearly 200 low-level parameters. High-level parameters including peaking time, MCA bin width, etc. EPICS software provides convenient methods to copy the values from detector 1 to all detectors, simplifying setup. Read/write access is also provided to all low-level parameters, though this should be rarely used Mapping Modes MCA spectra mode. Collects complete XRF spectra for each detector at each pixel. xMAP buffer holds 128 pixels. ROI (SCA) mapping mode. Collects total counts for each of 16 ROIs at each pixel. xMAP buffer holds 5457 pixels. xMAP Hardware 4 channels per PXI module 4 MB of memory per module. Used to buffer spectra or ROIs for very fast data collection Double-buffered to support simultaneous readout and acquisition 1 LEMO input for gate and trigger functions. Peaking times down to 125ns Supports both RC and reset preamps 4 data collection modes: MCA spectra, MCA mapping, ROI mapping, and List mode (List mode will be supported in the near future) PXI/PCI interface which achieves ~20 MB/sec when reading out xMAP. More than 20 times faster than CAMAC 4 xMAP modules in a PXI crate with fiber-optic connection to a Windows control computer running EPICS IOC software Performance MCA spectra mode: >4,000 2048 channel spectra per second, limited by readout rate of xMAP over PXI/PCI. 1000 pixels/s for a 4-channel quad Vortex, 40 points/s for a 100-element EXAFS detector ROI/SCAN mode: 10,000 pixels/second for any number of detectors, limited by xMAP overhead in pixel advance to about 100 µs/pixel. For 16 ROIs and 30 detectors=4.8M ROIs/s First results The images below are some of the first scans collected with the new EPICS xMAP software. They were all collected in on-the-fly scanning mode, collecting complete XRF spectra at each pixel, with pixel dwell times of 30 to 200 ms. EPICS medm Control Screens X-Y Scan of Alzheimer’s mouse brain Fe (red), Cu (green), Zn (blue) 9-element Ge and 2 SDD detectors 1050x793 pixels Pixel size=6x6µm 200ms per pixel Total time=49 hours Collected at NSLS X-26A Thanks to Tony Lanzirotti (U. of Chicago), Lisa Miller and Megan Bourassa (Stony Brook/NSLS) X-Y scan of Alyssum murale leaf Ca (green), Mn (red), Zn (blue) 9-element Ge and 2 SDD detectors 463x179 pixels Pixel size=7x7µm 100ms per pixel Total time=138 minutes Collected at NSLS X-26A Thanks to Ryan Tappero (NSLS) High-level parameter control Top-level 16-element control Mapping mode control Zn Ka Rb Ka As Ka Sr Ka Fluorescence tomography of heavy metals in rice Top images are sinograms, bottom images are reconstructed slices 4-element Vortex SDD detector 650 x 181 pixels Pixel size=2µm x 1° 30ms per pixel Total time=67 minutes Collected at APS 13-ID-C Thanks to Matt Newville (U. of Chicago), Anne-Marie Carey, (U. of Aberdeen), and Kirk Scheckel (US-EPA) ROI and SCA control netCDF file setup for saving mapping mode data Rb: marks phloem transport Sr: marks xylem transport Rb Ka Zn Ka As Ka Sr Ka Low-level parameter control Pre-amp output trace 16-channel detector spectra