In this work, we present an experimental study on hydrogenated amorphous silicon (a-Si:H) junction field-effect transistors (JFETs), focusing on the gate-controlled depletion of the conductive channel. Low-voltage electrical measurements were performed on devices with different n-layer thicknesses, allowing the channel conductance to be determined as a function of the gate bias. From these data, the evolution of the undepleted channel thickness was estimated and compared with the prediction of an analytical depletion model.Three distinct regimes where needed to describe the observed curves: a linear behavior consistent with a first-order approximation of the analytical expression, an exponential decay observed at the highest depletion conditions, and a transition regime, which maintains mathematical continuity between the two previous regimes.The proposed formulation captures the main experimental features and supports the development of circuit-compatible models for signal processing in lab-on-chip applications.
This paper presents a fully embedded real-time person tracking pipeline for assistive quadrupedal robots supporting safe navigation for visually impaired users. Our approach combines a deep learning-based 2D LiDAR person detector with a lightweight multi-object tracker and integrates it into a Guide Dog Robot (GDR) navigation framework. A novel detection post-processing scheme is proposed, reducing detector latency by 52.38% compared to state-of-the-art voting-based methods while preserving accuracy. The improved latency enables the entire pipeline to operate reliably at 20 Hz on a resource-constrained mobile robotic embedded platform based on the NVIDIA Jetson Xavier NX. The experimental setup shows that our system tracks dynamic obstacles and continuously localizes the user holding the robot’s handle, enabling a dynamic safety footprint for proactive collision avoidance. Under the tested setup, the optimal configuration achieves a MOTA of 83.27% and a user tracking RMSE below 0.2 m on two custom datasets recorded with motion-capture ground truth. Real-world navigation experiments in indoor environments demonstrate effective collision prevention and smooth corrective maneuvers when the user drifts from the default following position. The modular design of the detection, tracking, and planning components ensures flexibility and ease of integration into other robotic platforms. This work contributes a scalable and efficient tracking and navigation solution for human-aware mobile robots operating in dynamic environments, supporting safer human-robot interaction in assistive contexts.
This work presents a novel MEMS-based microgripper integrating electrostatic actuation and capacitive motion sensing, both achieved through interdigitated rotary comb structures. The device features curved flexure hinges for tip motion and ten pairs of electrostatic comb drive actuators for independently controllable tip displacement on both sides within a 130 mu m x 75 mu m workspace. Four pairs of double-sided rotary combs provide real-time position feedback of both tips, achieving a sensitivity of 5.2 fF/mu m. Finite Element Method (FEM) simulations confirm a maximum tip displacement of 65 mu m under 12 V, and a torque generation of up to 3.9 x 10(-10) Nm. Mechanical stress analysis shows that the design remains well within the structural limits of silicon, ensuring long-term reliability. The proposed device enables closed-loop control of jaw position, offering a compact and robust solution for high-precision micromanipulation tasks, such as single-cell stimulation, microsurgery, and lab-on-chip biological analyses.
The demand for hundreds of tightly synchronized channels operating at tens of MSPS in ultrasound systems exceeds conventional low-voltage differential signaling links’ bandwidth, pin count, and latency. Although the JESD204B serial interface mitigates these limitations, commercial FPGA IP cores are proprietary, costly, and resource-intensive. We present ListenToJESD204B, an open-source receiver IP core released under a permissive Solderpad 0.51 license for AMD Xilinx Zynq UltraScale+ devices. Written in synthesizable SystemVerilog, the core supports four GTH/GTY lanes at 12.8 Gb/s and provides cycle-accurate AXI-Stream data alongside deterministic Subclass 1 latency. It occupies only 107 configurable logic blocks (approximately 437 LUTs), representing a 79% reduction compared to comparable commercially available IP. A modular data path featuring per-lane elastic buffers, SYSREF-locked LMFC generation, and optional LFSR descrambling facilitates scaling to high lane counts. We verified protocol compliance through simulation against the Xilinx JESD204C IP in JESD204B mode and on hardware using TI AFE58JD48 ADCs. Block stability was verified by streaming 80 MSPS, 16-bit samples over two 12.8 Gb/s links for 30 minutes with no errors.
Cardiac auscultation with a digital stethoscope is an important method for early diagnosis of cardiovascular diseases (CVDs), especially in resource-constrained settings. However, its utility is often impaired by unwanted noise (e.g., lung sounds, ambient speech) obscuring critical heart signals in the recordings. This paper addresses the challenge of denoising heart sounds in noisy environments and performing this processing on-device for minimal delay and improved data privacy and security. We propose a deep learning-based denoising method using U-Nets based on the Spleeter architecture, which separates heart sound signals from noise in the time-frequency domain. The models are implemented on two edge computing platforms - a Raspberry Pi 4 B and a Qualcomm RB3 Gen 2 - demonstrating the feasibility of AI-driven heart sound denoising at the edge. We further optimize the models via quantization to Int8 precision to leverage the RB3 Gen 2's neural processing unit (NPU) for acceleration. Experimental results show that the denoisers achieve significant enhancement in signal quality, with each yielding an average improvement in scale-invariant signal-to-distortion ratio (SI-SDR) exceeding 10 dB. The edge deployment meets real-time constraints, with the quantized model on the RB3 Gen 2 executing denoising substantially faster (at least 17 times faster than the Raspberry Pi implementation) and at lower power compared to the full-precision model, all without appreciable loss in denoising performance - 0.06 dB being the largest drop. These results demonstrate the potential of deploying advanced AI denoising on edge devices for immediate, on-device heart sound denoising.
This paper presents the design of a current-mode read-out circuit for EMG monitoring employing secondgeneration current conveyors (CCIIs) and second-generation voltage conveyors (VCII) as active blocks. After a quick overview of the current-mode approach, the paper describes the whole interface, composed by three active blocks, and its operation. The proposed circuit was simulated by LTspice simulator firstly designing the internal blocks at transistor level, in 0.18 mu m standard CMOS technology, and then using the commercially available device AD844 to demonstrate the validity of the designed circuit also with simulations. The voltage gain is tunable by changing some resistance values in the different stages of the amplifier. In the proposed solution, CMRR has been set to be more than 100 dB, being the common mode gain about -40 dB. The aim of this research is to test the performance of biological signal amplifiers using the currentmode approach (in particular, VCII) which is still underutilized in this context. The goal is to provide a more efficient and versatile system for biomedical applications.
For flexible and large-area electronics, inkjet printing presents an attractive production technique; yet, creating environmentally friendly functional inks that are truly sustainable is a significant obstacle to the development of green electronics. Here, we demonstrate the use of a nanocarbon conductive ink from food waste for manufacturing printed resistive loads compatible with active components. The formulation is designed for inkjet printing, and it is made entirely of non-toxic and renewable components: ethanolterpineol mixture as a dispersant, electrically conductive activated carbon nanoparticles, and ethyl cellulose as a binder and stabilizer. Activated carbon nanoparticles with a 30 to 120 nm diameter make up the ink, which has excellent colloidal stability and rheological properties for inkjet printing. This formulation makes the manufacturing of printed resistive elements for electrical circuits possible, where the number of layers and/or drop spacing during the printing process can be adjusted to modulate the sheet resistance. As a proof-ofprinciple, we use this formulation to fabricate high-resistive loads in the range of 1 - 10 M Omega for a simple unipolar NOT gate together with an organic transistor and a non-toxic rechargeable battery. The logic circuit exhibits characteristic NOT gate behaviour with quasi-rail-to-rail output (Delta V-out/ V-dd = 86%) and minimal hysteresis. A peak voltage gain of 3.2 at the switching threshold (V-in approximate to V-dd/2) highlights its potential for integration into more complex low-voltage circuits. The proposed circuit demonstrates that our nanocarbon formulation is suitable for inkjet printing high resistance loads and compatible with other sustainable electronic components.
Mobile devices like smartphones, smartwatches, or other wearables are becoming fundamental to our daily lives. Thanks to the multiple sensors embedded to interact with the user and the environment, they enable tasks such as cardiovascular and daily activity monitoring, apart from receiving calls. However, their capabilities can be further enhanced with computational models to enable autonomous learning from user behavior. Consequently, given the potential of jointly considering wearables and Machine Learning (ML) models, this research seeks to facilitate end users’ daily lives by making mobile devices such as smartphones more accessible. More in detail, ML models allow pattern detection and significant task automation, such as making calls, sending messages, or opening applications. Moreover, synthetic data generation is applied, which contributes to cost-effective, multi-spectral, scalable, and stochastic simulation. Results obtained with experimental data by applying 10-fold cross-validation are up to 96 % in all evaluation metrics. The ultimate objective is to increase accessibility for society in general and especially for people with cognitive or physical disabilities, reducing the traditional usability barriers.
Real-time measurement of patellofemoral (PF) contact forces could transform prosthesis design, surgical alignment, and rehabilitation protocols, yet no commercially available patellar component offers dedicated force sensing. We present a dual-channel acquisition system based on bisected Tekscan FlexiForce A401 sensors, a compact signal-conditioning board, and an Arduino Nano 33 BLE Sense for 10-bit digitization and real-time data streaming at 100 Hz. A fourth-order Butterworth filter (cutoff 5 Hz) cleans the raw voltages, which are then linearly mapped to force via bench-calibrated conversion factors, yielding a gain of 477 N/V for the lateral channel and 347 N/V for the medial channel. The instrumented patellar component was evaluated in an ex vivo cadaveric study (ethical approval obtained), in which a TKA-implanted lower limb was mounted in a customized fixture and subjected to three cycles of flexion-extension under quadriceps loads (20-280N). The system clearly captured both lateral and medial contact forces with consistent peak-valley patterns and variability reflecting native trochlear geometry. Finally, we introduce a next-generation integrated PCB combining an ESP32-S3 wireless module for acquisition, processing, and streaming, dual-channel MCP6002 signal conditioning, and a Li-ion battery with on-board management-paving the way for future intraoperative use. This work establishes the quantitative and methodological groundwork for patient-specific, force-guided PF arthroplasty.
In the Quartz-Enhanced Photoacoustic Spectroscopy (QEPAS) technique, a laser source modulated with a sinusoidal signal interacts with the target gas, generating acoustic waves. These waves are detected by a resonant piezoelectric sensor, the quartz tuning fork (QTF), whose response amplitude is proportional to the gas concentration, which is the desired measurement. The key advantage of this technique lies in leveraging the resonance properties of the QTF, which distinguishes it from conventional photoacoustic methods. To fully exploit this benefit, the QTF must be excited at its resonance frequency, requiring a reliable characterization process to accurately identify this frequency. Typically, this is accomplished using electrical characterization of the QTF, i.e. by exciting the sensor with an external sinusoidal signal and finding the frequency corresponding to the peak of the QTF response. However, it has been observed that this process does not guarantee that the peak frequency obtained in this way corresponds to the optimal frequency for the QEPAS technique during the normal operation. In particular, the difference depends on the front-end used to read-out the QTF. This work aims to highlight the existence of this discrepancy, emphasizing the importance of selecting an appropriate preamplifier to achieve an optimized system. Furthermore, it seeks to identify the optimal circuit configuration that minimizes this mismatch, thereby ensuring full exploitation of the QTF’s resonance properties for high-sensitivity gas detection.
Wellness and comfort are key to occupant health, productivity, and satisfaction. The rise of Internet of Things (IoT) is transforming building management by integrating advanced sensors and analytics, enabling intelligent systems that enhance energy efficiency and thermal comfort. This study introduces an innovative framework that combines multivariate analysis (MVA) with IoT technologies to improve indoor environmental wellness and optimize heating, ventilation, and air conditioning (HVAC) efficiency. It utilizes a nonlinear locally weighted regression (LWR) model, enhanced by sequential quadratic programming (SQP), to manage time-dependent variations in environmental factors such as temperature (Temp) and relative humidity (RH). The model was validated using 765 data points collected from three controlled indoor spaces in a university building. Its low computational overhead is compatible with real-time deployment on microcontroller-based platforms, making it well-suited for scalable and adaptive comfort control. Additionally, the framework contributes to more efficient HVAC operation by minimizing unnecessary energy consumption while maintaining occupant satisfaction.
Continuous and non-invasive blood pressure (BP) monitoring is increasingly recognized as a key tool for early detection of cardiovascular disorders and real-time patient management. Traditional cuff-based methods, although accurate, are unsuitable for continuous monitoring due to their intermittent nature and discomfort during prolonged use. To address these limitations, we developed a hybrid deep learning model combining a Convolutional Neural Network (CNN) with a Support Vector Regression (SVR) correction phase, trained to predict systolic and diastolic BP from multi-wavelength photoplethysmography (PPG) signals. In this study, we specifically evaluated the model’s ability to track rapid physiological BP changes induced by the Valsalva maneuver, a standard test provoking transient cardiovascular adjustment. BP values were recorded and predicted immediately before and after the maneuver in 6 individual participants. The model achieved a mean absolute error (MAE) of 1.95 mmHg (STD = 4.72 mmHg) for systolic BP and 1.83 mmHg (STD = 3.02 mmHg) for diastolic BP. Predicted BP trajectories closely followed the characteristic hemodynamic response to the maneuver, demonstrating the model’s sensitivity not only to static BP levels but also to dynamic physiological fluctuations.
In this work, we presented the development of a simulation methodology for a-Si:H devices in a commercial state-of-the-art simulation environment. Hydrogenated amorphous silicon (a-Si:H) has emerged as an attractive material for particle detectors, driven by its high bandgap, which translates to minimal leakage current, and the potential for cost-effective large-area deposition on diverse substrates. The adoption of Technology Computer Aided Design (TCAD) simulation represents a powerful mean for the design and optimization of particle detectors, fostering the evaluation of the electrical properties of the material and the interaction with a particle at device level. We included within the Synopsys Sentaurus TCAD a new material featuring the main parameters of a-Si:H (e.g. band-gap, density of states, e/h creation energy) and an articulated picture of energy of states of defects. Moreover, a brand-new mobility model, derived from the Pool-Frenkel one, has been developed and included as external add-on, accounting for the influence on the mobility of the potential/electric field distribution inside the device and of the temperature. Simulation findings have been compared with measurements carried out on different p-i-n samples for model validation purposes.
This paper presents an analytical and MATLABbased simulation model for evaluating background current variance and Signal-to-Background Noise Ratio (SBNR) in photon detection systems. The model integrates stochastic photon arrival statistics with a pulse shape derived from a realistic mixedsignal front-end, enabling direct investigation of how pulse fall time t and photon flux influence noise performance. A closedform expression for background variance is derived under the assumption of independent arrivals and triangular pulses, and is validated through extensive simulations spanning a range of fall times and background photon rates. The SBNR is then formulated by considering coherent signal photon accumulation-enabled by a laser FWHM much shorter than t-and is shown to closely match simulation results across diverse pulse fall time conditions.
Microbial Electrolysis Cells (MECs) represent an advanced bioelectrochemical technology that integrates wastewater treatment with the generation of valuable bioproducts, such as hydrogen and methane. These technologies leverage applied voltage to facilitate electrochemical reactions, overcoming thermodynamic limitations to produce resources from the breakdown of organic matter in wastewaters. Concurrently, the electrical current in an MEC serves as a critical performance metric, representing electron transfer from the anode to the cathode of the MEC and directly correlating with microbial activity, pollutant degradation, and bioproduct synthesis. Despite advancements, current MECs cannot dynamically adjust applied voltage based on real-time current dynamics. This paper presents the first real-time management unit (RMU) for MECs, designed to overcome this limitation by enabling remote monitoring, control, and independent optimization of applied voltage and current across multiple MECs. This RMU utilizes a proportional-integral-derivative (PID) control algorithm to regulate the applied voltage for up to 20 MECs, integrating a dedicated microcontroller unit (MCU) with LoRaWAN communication within custom-designed hardware. Experimental results confirm the RMU’s capability to regulate applied voltage and monitor current in pilot-scale MECs, demonstrating its functionality through in-house testing.
A portable impedance spectroscopy (IS) device has been developed based on the STM32F407 commercial microcontroller. The microcontroller has been programmed to perform dynamic sampling in response to the sinusoidal input signal, effectively addressing oversampling issues for multi-frequency analysis and reducing acquisition time through an internal clock optimization algorithm. The resulting architecture further reduces overall size, enhancing device portability. Validation tests were conducted using electrical circuits designed to mimic the characteristics of biological samples. The experimental results were compared with theoretical transfer functions derived from simulated impedances in the 10-200 kHz range under sinusoidal current excitation to preserve bioimpedance linearity. In the worst-case scenario, the device demonstrated a maximum mean magnitude error of 3.48% and a maximum mean phase error of 1.71 degrees when compared to the ideal transfer function of the equivalent impedance. These results highlight the performances achieved by the low-cost, portable IS device.
Calorimeters at future colliders will require excellent energy resolution to differentiate between hadronic decays of W and Z bosons, a granularity at the (O(cm(2))) level and time resolution of a few ns, to be compliant with the Particle Flow Algorithm for jet reconstruction. We propose a hadronic calorimeter (HCAL) consisting of a sampling of absorber material and resistive Micro Pattern Gaseous Detectors (MPGD) as the active layer for the future muon collider. We simulated a small-size (similar to 1 lambda) MPGD-based HCAL prototype and studied its performance with pion beams. Furthermore, we performed the experimental characterization studies of MPGD prototypes with an active area of 20x20 cm(2) in order to assess their performance under MIP irradiation, in terms of efficiency, time resolution, and response uniformity. We built a calorimeter prototype instrumented with 20x20 cm(2) MPGDs and characterized its response under pion beams. New MPGD prototypes with a larger area (50x50 cm(2)) are currently under construction with the goal to assess the response uniformity, which is crucial for the hadronic shower reconstruction. In this paper, we report the simulation studies of a similar to 1.5. calorimeter prototype including the new 50x50 cm(2) detectors.
This paper reports on a proof-of-concept SPAD-based (Single Photon Avalanche Diode) optical encoder. The work aims at demonstrating the advantages of SPADs over photodiodes, which are typically used in the current optical position measuring systems. In addition to their high sensitivity and high speed, SPADs allow fully digital signal processing, offering a large system flexibility and scalability toward advanced CMOS technologies. Preliminary tests have been carried out using an array of 100 x 100 SPADs, coupled with an optical Gray-coded disk and a laser diode. Binary frame sequences were acquired and processed off-line through a lightweight algorithm to reproduce the disc code. The described algorithm aims at being integrated in the same chip of the sensor to speed up the signal processing chain, thus allowing high rotation speeds to be achieved. Experimental results are reported, together with future work and conclusions.
This paper presents a performance assessment of a fabricated optoelectronic Lab-on-Chip device for the detection of dyes in polluted water. It monolithically integrates a polymeric waveguide and a hydrogenated amorphous silicon (a-Si:H) photodetector on a compact BK7 glass chip measuring a few square centimeters in footprint. The device senses the complex refractive index of a liquid sample, leveraging the interaction between the evanescent field of the guided light and a solution droplet, monitoring the optical losses at the end of the light path, and linking them to the analyte concentration in the solution. The device’s performance was assessed for methylene blue (MB) solutions in water through numerical simulations conducted at a working wavelength of 613 nm. The sample-chip interaction was examined, and the results were combined with the experimentally measured photoresponse of the fabricated a-Si:H detector. The sensor demonstrated a sensitivity of 2.43 pA/ppm, along with a limit of detection of 78 ppb. As the device does not include bulky detectors, chemical treatments, or complex fabrication steps, the reported performance of this highly compact, cost-effective, and easy-to-use sensor appears very promising for health and environmental monitoring applications, particularly for in-situ detection of pollutants in water.
The need for simple, real-time, fast, and in-situ devices for water analysis is becoming increasingly urgent. A new type of contaminant, known as emerging micro-contaminants, has been detected in water, posing a potential risk to human and environmental health. Currently, their identification relies on traditional analytical methods, which are very sensitive but also expensive and time-consuming. Optical biosensors are a promising alternative. This article proposes a concept for an optical device for biosensing applications. The focus of the study is the light coupling performance between optical channel waveguides and periodic nanoarrays in the visible spectrum. Coupling performance can be enhanced using an indium tin oxide (ITO) thin film as a buffer layer. The intensity and deflection of the guided light result from its interaction with periodic nanocylinders made of ITO or gold (Au). To optimize the configuration, the study analyzes coupling efficiency, out-of-plane scattering, and the far-field projection of beam focus and diffraction angle. These factors are evaluated using the finite-difference time-domain (FDTD) method. The implementation of ITO enables up to a ninefold increase in coupling efficiency, while the presence of the Au array yields a coupling efficiency of 13% at a wavelength of 580 nm. The nanoarray also induces a deflection that depends on the wavelength, a property that could be significant for directional light focusing a phenomenon with potential applications in environmental and biomedical sensing.
