A digitizer designed to read out column-parallel charge-coupled devices (CCDs) used for high-speed X-ray imaging is presented. The digitizer is included as part of the High-Speed Image Preprocessor with Oversampling (HIPPO) integrated circuit. The digitizer module comprises a multiplexed, oversampling, 12-bit, 80 MS/s pipelined Analog-to-Digital Converter (ADC) and a bank of four fast-settling sample-and-hold amplifiers to instrument four analog channels. The ADC multiplexes and oversamples to reduce its area to allow integration that is pitch-matched to the columns of the CCD. Novel design techniques are used to enable oversampling and multiplexing with a reduced power penalty. The ADC exhibits 188 ?V-rms noise which is less than 1 LSB at a 12-bit level. The prototype is implemented in a commercially available 65 nm CMOS process. The digitizer will lead to a proof-of-principle 2D 10 Gigapixel/s X-ray detector.

The complexity of information processing in the brain requires the development of technologies that can provide spatial and temporal resolution by means of dense electrode arrays paired with high-channel-count signal acquisition electronics. In this work, we present an ultra-low noise modular 512-channel neural recording circuit that is scalable to up to 4096 simultaneously recording channels. The neural readout application-specific integrated circuit (ASIC) uses a dense 8.2 mm × 6.8 mm 2D layout to enable high-channel count, creating an ultra-light 350 mg flexible module. The module can be deployed on headstages for small animals like rodents and songbirds, and it can be integrated with a variety of electrode arrays. The chip was fabricated in a TSMC 0.18 µm 1.8 V CMOS technology and dissipates a total of 125 mW. Each DC-coupled channel features a gain and bandwidth programmable analog front-end along with 14 b analog-to-digital conversion at speeds up to 30 kS/s. Additionally, each front-end includes programmable electrode plating and electrode impedance measurement capability. We present both standalone and in vivo measurements results, demonstrating the readout of spikes and field potentials that are modulated by a sensory input.

We present a demonstration of readout of physical cryogenic calorimetric signals from 20 mK with a two-stage amplifier using 180 nm CMOS technology placed at 20 mK and 700 mK. We observe a number of cryogenic effects in 180 nm CMOS that deviate from typical warm behavior, which are not prohibitive to their cryogenic operation but whose potential impacts we discuss. This represents a new step towards the viability of cryogenic electronics using 180 nm CMOS in cryogenic calorimetric experiments.

Direct X-ray detection in silicon has been transformative for scattering experiments in biology and materials science. While bump-bonded hybrid pixels have been attractive for hard X-ray detection, the challenge for single photon soft X-ray detection is sufficiently low noise. CCD structures on thick, high-resistivity silicon have been successfully used as sensors over the range of soft to hard X-rays at storage rings and FELs. The VeryFastCCD is a high frame rate, column-parallel CCD sensor with 48 μm pixels. Combined with 256-channel custom readout ASICs, frame rates of 5–10 kHz have been achieved with readout noise as low as 20 e and full-well capacity >4 × 105 e/pixel. Thin (10 nm and 100 nm) entrance window contact processes have also been developed which provide >85% quantum efficiency for 285 eV X-rays. Systems are currently being developed for several beamlines at the upgraded Advanced Light Source.

There has been considerable interest and resulting progress in implementing machine learning (ML) models in hardware over the last several years from the particle and nuclear physics communities. A big driver has been the release of the Python package, hls4ml, which has enabled porting models specified and trained using Python ML libraries to register transfer level (RTL) code. So far, the primary end targets have been commercial field-programmable gate arrays (FPGAs) or synthesized custom blocks on application specific integrated circuits (ASICs). However, recent developments in open-source embedded FPGA (eFPGA) frameworks now provide an alternate, more flexible pathway for implementing ML models in hardware. These customized eFPGA fabrics can be integrated as part of an overall chip design. In general, the decision between a fully custom, eFPGA, or commercial FPGA ML implementation will depend on the details of the end-use application. In this work, we explored the parameter space for eFPGA implementations of fully-connected neural network (fcNN) and boosted decision tree (BDT) models using the task of neutron/gamma classification with a specific focus on resource efficiency. We used data collected using an AmBe sealed source incident on Stilbene, which was optically coupled to an OnSemi J-series silicon photomultiplier (SiPM) to generate training and test data for this study. We investigated relevant input features and the effects of bit-resolution and sampling rate as well as trade-offs in hyperparameters for both ML architectures while tracking total resource usage. The performance metric used to track model performance was the calculated neutron efficiency at a gamma leakage of 10-3. The results of the study will be used to aid the specification of an eFPGA fabric, which will be integrated as part of a test chip.

A high frame rate detector system is being developed to enable fast real-time data analysis of scanning diffraction experiments in scanning transmission electron microscopy (STEM). This is an end-to-end development that encompasses the data producing detector, data transportation, and real-time processing of data. The detector will consist of a central pixel sensor that is surrounded by annular silicon diodes. Both components of the detector system will synchronously capture data at almost 100 kHz frame rate, which produces an approximately 400 Gb/s data stream. Low-level preprocessing will be implemented in firmware before the data is streamed from the National Center for Electron Microscopy (NCEM) to the National Energy Research Scientific Computing Center (NERSC). Live data processing, before it lands on disk, will happen on the Cori supercomputer and aims to present scientists with prompt experimental feedback. This online analysis will provide rough information of the sample that can be utilized for sample alignment, sample monitoring and verification that the experiment is set up correctly. Only a compressed version of the relevant data is then selected for more in-depth processing.