Michele Benetti

A 533pW NEP 31×31 pixel THz image sensor based on in-pixel demodulation

A THz 31×31 pixel, 100 fps image sensor integrated in a 130 nm CMOS process is presented. Taking advantage of the possibility to modulate the active source that lights the scene, a significant improvement in sensitivity and NEP is achieved by shifting the modulated THz radiation, by means of an antenna/MOSFET, then filtering the signal band using an in-pixel 16-paths passive SC-filter combined with a CT Gm-C filter resulting in a high Q factor of 100. This THz imager features a measured NEP of 533 pW at 270 GHz and 732 pW at 600 GHz respectively, and a sensitive readout chain with an input referred noise of 0.2 μVRMS.

Highly parallel SPAD detector for time-resolved lab-on-chip

Fluorescence lifetime detection is widely used in molecular biology to monitor many cell parameters (such as pH, ion concentrations, etc.) and for an early diagnosis of many pathologies. In a typical fluorescence lifetime experiment a pulsed laser is used to excite the fluorescent dyes and the emitted light is revealed by means of high sensitivity detectors, typically: intensified CCD, PMTs or Single-Photon Avalanche Diodes (SPADs).In this contribute we present a SPAD detector module fabricated in a 0.35μm High Voltage CMOS technology to be used within a lab-on-chip system consisting of a micro-reactor array for bioaffinity assays based on fluorescence markers. The detector module, having a total area of 600 x 900 μm2, can be arranged to build a small pixel array to be directly coupled to the micro-reactors. No emission filters are needed, since the ultra-short laser pulse is cut off in the time domain. The module consists of a 10x10-SPAD array, where each SPAD cell is equipped with dedicated active quenching and recharging circuit. Each cell has a pitch of 26μm with a fill factor of 48%. The SPADs have been binned in order to realize a large photosensitive area detector exhibiting a reasonably low dark count rate (DCR) and reduced dead time, as required in a fast measurement system. A memory has also been implemented in order to enable only low DCR SPADs, so that a total DCR of about 100kHz can be achieved for the whole photosensitive area. The digital output generated by the SPAD array is sent to a time-discriminator stage which allows a time-gated detection of the incident light. Two time-windows have been implemented in this architecture. Their time width is controlled by an on-chip digital PLL locked to the external laser clock whereas the width of the time-windows can be set within the range 500ps-10ns with a resolution of 500ps. Photons detected within each time window are then counted by two 10-bits digital counters. Time-interleaved operation has been implemented to read out the pixel data in parallel with the photon detection phase.

Characterization of a CMOS SPAD sensor designed for fluorescence lifetime spectroscopy

The characterization of a 10×10-SPAD detector module fabricated in a 0.35µm High Voltage CMOS technology is presented. The detector is designed to find application in Fluorescence Lifetime spectroscopy and is capable of performing lifetime measurement by using the time-gated technique. The characterization explores the dark count distribution, the detector dynamic range and the gating performances. More than 70% of single SPADs have a dark count rate lower than 1 kHz at the excess bias voltage of 1.8V and do not exceed 2 kHz at the bias voltage of 4.8V. Exploiting the intrinsic features of the detector a direct measurement of the optical cross talk between neighboring SPAD in the matrix was performed. The cross talk was found to be in the range from 2% to 3% for lateral neighbors and between 0.3% and 0.5% for diagonal neighbors. A dynamic range exceeding 120 dB was observed with a maximum count rate before saturation of 500 MHz. Time gating resolution was found to be less than 1 ns. A fluorescent lifetime measurement of ZnS-ZnSe quantum-dot reference slides was performed and the non-uniformity of the calculated lifetime value was lower than 1% across the matrix. The application of the detector in the construction of a lifetime acquisition system is analyzed estimating the detector data throughput. External resources needed to build the acquisition system are estimated and the FPGA-based acquisition system used during characterizations is described.

On compensating unknown pixel behaviors for image sensors with embedded processing

Some smart imagers embed image processing operations or Compressive Sensing at the focal plane level. This process introduces artifacts due to technology dispersion and unpredictable behaviors. This article presents a generic algorithm structure well suited for compensating block artifacts by appropriate post processing operations. The proposed restoration method is composed by a three steps loop: regularize the image, fit the model parameters and update the regularization coefficient (by reweighting the fidelity term). To further demonstrate the efficiency of the proposed generic technique, two specific case studies involving structured PRNU and multi-capture compressive sensing are presented with simulation results.

Energy Autonomous Low Power Vision System

This paper presents the design and the development of a novel vision system, capable of sensing and describing the visual world it observes under physical constraints that include ultra-low power consumption, easy deployment, low maintenance cost, and a small unobtrusive form-factor. Energy aware vision processing algorithms have been developed based on the custom hardware. Simulation and design of an energy harvester using solar cells has been addressed to become the power supply unit of the proposed vision system. We describe the hardware-software architecture of the video sensor node and provide a characterization in terms of power consumption and power generation and energy efficiency of the harvester. Different strategies of energy harvesting, based on low energy DC–DC converter, and different types of storage device are analyzed, focusing on different battery technologies and comparing the different characteristic curves (charge and discharge curves). Specific attention will be reserved to different types of solar cells (amorphous and monolithic) in indoor environment.

Compact gamma detectors based on FBK SiPMs for a Ps Time Of Flight apparatus

Silicon Photomultipliers (SiPMs) are single-photon detectors currently used in many applications as a valid alternative to Photomultiplier Tubes (PMTs). In fact, SiPMs present several advantages compared to PMTs, in terms of size, ruggedness, insensitivity to magnetic fields and voltage supply.In this work, we evaluated the use of SiPMs produced by FBK (Fondazione Bruno Kessler)-AdvanSid (Trento, Italy) in Positronium Time Of Flight (Ps TOF) technique. The aim of this technique is the study of the energy of Ps emitted by porous media for advanced experiments in the field of antimatter physics and, in the future, for open porosities characterization.The TOF spectrometer in operation at the positron beam of the University of Trento (Italy) is based on five gamma detectors composed of NaI(Tl) scintillators coupled to PMTs. We compared the performance of one of these standard detectors against a detector formed by two 4 × 4 mm2 SiPMs coupled to a 4 × 4 × 30 mm3 LYSO (Cerium-doped Lutetium Yttrium Orthosilicate) scintillator. The LYSO+SiPM detector was characterized for timing and energy resolution as well as for the background emission level.It is shown that SiPMs, thanks to their small size and versatility, can be coupled successfully to small scintillators, obtaining more compact detectors with respect to the NaI(Tl)+PMT detectors. The design of an array of detectors with wide angular acceptance and spatial resolution is discussed in order to improve the performances of Ps-TOF setups.

Performances of an highly parallel image processing chain for capturing high dynamic range video

In this paper we address the process that permits to obtain a High Dynamic Range (HDR) video starting from a set of different exposed Low Dynamic Range (LDR) acquisitions. A hard constraint in the acquisition process is represented by the video rate, that limits the number of LDR images that can be acquired with given hardware resources, such as data throughput or buffer memory. We present an HDR reconstruction image Processing Chain (PC) that has been conceived highly parallel: it measures the image by blocks with dedicated LDR. We present results from the PC hardware implementation and evaluated performances.

A 80µW 30fps 104 × 104 all-nMOS pixels CMOS imager with 7-bit PWM ADC for robust detection of relative intensity change

The presented vision sensor features low-power pixel-level programmable dynamic background subtraction as low-level image processing aimed at detecting unusual events occurring in the scene. Each pixel compares its current photogenerated signal with two threshold voltages, defining the boundary conditions outside which the signal is to be considered anomalous, if compared to its past history. In case of anomalous behavior, the pixel one-bit output will be asserted. The 17 transistors square pixel has a pitch of 16μm with a fill-factor of 21%. The 104 × 104 pixel sensor, fabricated in a 0.35 μm CMOS, has a power consumption of 80μW at 30 fps, when the sensor array is powered at 3.3V and the digital part at 1.5V. This turns into a pixel power consumption per frame of 246pW/pixel · frame.

Optimization of a multi-ring detector for Ps time of flight measurements

We have designed a multi-ring detector (MRD) based on Bismuth Germanate (BGO) crystals, coupled to Silicon PhotoMultipliers (SiPM) for measuring the Ps time of flight (TOF). The set-up geometry was optimized by Monte Carlo simulations to take into account at different Ps velocities: (i) the background noise due to backscattered positrons, (ii) the crosstalk between adjacent detectors, (iii) the lifetime of Ps decay. Three parameters were defined to evaluate the different configurations and a figure of merit was obtained. This allows the choice of the best set up configuration for measuring Ps emitted with a particular energy range, optimizing the signal to noise ratio and keeping the acquisition time acceptable.

CMOS single-photon detector for advanced fluorescence sensing applications

Fluorescence lifetime measurement is used in biological research to enhance the contrast of fluorescence images. The outstanding sensitivity that can be achieved with this method is obtained at the expense of a high data throughput. A substantial data reduction can be achieved using the time-gated technique, which consists in counting the number of photons occurring inside different time windows. Thanks to the recent developments in the realization of Single Photon Avalanche Diodes (SPAD) in standard CMOS technologies, this technique can be monolithically implemented on-chip. In this work, three different detectors fabricated within a 0.35 ▪m High Voltage CMOS technology will be described, focusing onto their use in lifetime imaging. The sensors have been designed for different optical setups and for different applications, ranging from Fluorescence Lifetime Imaging Microscopy to miniaturized Lab-on-Chip. The advantages and limitation of the proposed sensors will be pointed out and a case study of a specific application will be presented.