This work presents throughput-enhanced readout ASIC (TERA), an integrated multichannel analog pulse processor (APP), suitable for processing signals generated from solid-state detectors, such as silicon drift detectors (SDDs) coupled to pulsed reset-type charge-sensitive amplifiers (CSAs). In particular, this development targets applications at high count rates (≈Mcps) in energy-dispersive X-ray spectroscopy. The demonstrator chip is composed of four parallel readout channels. Each channel includes a seventh-order semi-Gaussian shaping amplifier with controllable shaping times and full-scale ranges, followed by a peak stretcher and switched-capacitor analog memory. Each channel is also equipped with a dedicated peak detector and a novel pile-up rejection (PUR) logic. Each pair of channels can be optionally digitized by a 12-bit on-chip successive-approximation-register (SAR) analog-to-digital converter (ADC), providing a digital output at a maximum sampling rate of 2.5 Msps. The architecture achieves high throughput and satisfactory energy resolution. In fact, in 55 Fe spectroscopy measurements, when using the shortest shaping pulsewidth of 200 ns, a full-width at half-maximum (FWHM) Mn $\text{K}\alpha $ line of 159.4 eV (12.5 $\text{e}^{-}_{\text {rms}}$ ) was obtained at low rates (10 kcps). This energy resolution can be achieved, thanks to the minimization of the effect of the ballistic deficit, achieved by optimizing the SDD detector in terms of optimum biasing, low operating temperature (−37 °C), and small size (1-mm diameter). At 1.61 Mcps/channel input rate, an output count rate of 1.09 Mcps/channel and a resolution of 205.1 eV (19.7 $\text{e}^{-}_{\text {rms}}$ ) were achieved with a 4-mm-diameter SDD and optimum PUR settings. Simultaneous multichannel acquisition has been successfully performed. Such high-rate performances are, to our knowledge, the best ones reported so far for an APP, based on a spectroscopy-grade application-specific integrated circuit (ASIC), and very close to those achievable by a standard digital pulse processor (DPP). Therefore, TERA can represent an attractive, compact, and scalable pulse processing solution for high-rate multichannel energy-dispersive X-ray detection systems.
This article presents a new global shutter (GS) pixel design with radiation-hardened-by-design (RHBD) device modifications and correlated double sampling (CDS). Global shutter imagers present undeniable advantages by exposing all pixels simultaneously. Once in-pixel storage is included, the pixel readout can be operated simultaneously with exposure, enabling faster operation and flashed light. Additionally, this architecture is robust against motion defects and easy to operate in synchronous mode. The radiation tolerance of the outlined pixel is targeted up to 1 MGy SiO2 total ionizing dose (TID). A partially pinned photodiode (pPPD) structure, combined with an enclosed transistor layout known as a “butterfly” [1,2] is used to withstand such a high TID value. The use of the p+ implant on top of the photodiode and recessed shallow trench isolation (STI) reduce the dark current from the TID-induced interface traps and SiO2 interface while also shielding the photodiode (PD) region from TID-induced positive charges in the SiO2. CDS [3] is required to reduce the read noise by suppressing important noise sources such as photon shot noise, thermal noise (kTC) and flicker noise contributions. Two in-pixel storage capacitors and two readout paths are implemented to allow CDS. The pixel operation includes a sample phase where the pixel signal is locally stored followed by a reset phase and the pixel reset is stored. The in-pixel storage is designed with enclosed butterfly MOS devices that allow great charge density for low area and with limited leakage due to the RHBD design. The readout is performed by operating two output source followers (SF) and two output buses. The pixel is designed using 180 nm CMOS Image Sensor (CIS) technology. It achieves a pre-rad dynamic range of 60.9 dB and a shutter efficiency of 99,6%.
A 10 × 10 µm2 radiation-tolerant voltage-domain global shutter pixel with radiation-hardened by design (RHBD) device modification is developed to operate under high ionizing-dose rates and high total ionizing-dose (TID) levels. Therefore, a modified NMOS transistor layout is used in the pixel to achieve radiation hardness. The pixel design is demonstrated to operate up to 1 MGy or 100 Mrad (SiO2) TID with minimal degradation. The global shutter pixel also includes correlated double sampling (CDS) to reduce noise and the impact of the collected carriers generated by the flux of gamma or X-ray radiation. Combined with an external flash, global shutter operation allows short exposures, which limits the impact of radiation on dark current and dynamic range. The pixel is designed using 180 nm CMOS Image Sensor (CIS) technology.
Front-end amplification inside a pixel can reduce the contribution of downstream readout noise in complementary metal-oxide-semiconductor (CMOS) image sensors, but its practical value must be assessed together with penalties in spatial uniformity, implementation overhead, and operating complexity. This work presents a CMOS image sensor prototype incorporating an in-pixel operational transconductance amplifier (OTA) with switchable unity-gain and open-loop high-gain modes. A dedicated transfer-gate layout optimization is also implemented to reduce the floating-diffusion capacitance and improve the conversion gain. The prototype enables a same-chip comparison between the proposed OTA pixel and a reference four-transistor (4T) pixel under a common readout environment. Measurement results show that the in-pixel OTA can reduce the input-referred temporal noise (down to 0.67 e-rms ) by suppressing the contribution of subsequent readout stages, while the high-gain mode also increases pixel-to-pixel variation and spatial non-uniformity. A measured pixel conversion gain of up to 6.47 mV/ e- is achieved in the open-loop high-gain mode. The switchable-gain operation supports wide-dynamic-range readout by combining low-gain and high-gain responses. In addition to temporal-noise characterization, image lag and spatial non-uniformity are evaluated, and the results are discussed in terms of the trade-off between noise reduction, uniformity, and implementation cost. This work therefore provides a measured assessment of the practical benefits and limitations of in-pixel front-end amplification for low-noise and high-dynamic-range (HDR)-oriented CMOS image sensors.