Recently, piezoresistive-type (PRT) pressure sensors have been gaining attention in variety of applications due to their simplicity, low cost, miniature size and ruggedness. The electrical behavior of a pressure sensor is highly dependent on the temperature gradient which seriously degrades its reliability and reduces measurement accuracy. In this paper, polynomial-based adaptive digital temperature compensation is presented for automotive piezoresistive pressure sensor applications. The non-linear temperature dependency of a pressure sensor is accurately compensated for by incorporating opposite characteristics of the pressure sensor as a function of temperature. The compensation polynomial is fully implemented in a digital system and a scaling technique is introduced to enhance its accuracy. The resource sharing technique is adopted for minimizing controller area and power consumption. The negative temperature coefficient (NTC) instead of proportional to absolute temperature (PTAT) or complementary to absolute temperature (CTAT) is used as the temperature-sensing element since it offers the best temperature characteristics for grade 0 ambient temperature operating range according to the automotive electronics council (AEC) test qualification ACE-Q100. The shared structure approach uses an existing analog signal conditioning path, composed of a programmable gain amplifier (PGA) and an analog-to-digital converter (ADC). For improving the accuracy over wide range of temperature, a high-resolution sigma-delta ADC is integrated. The measured temperature compensation accuracy is within ±0.068% with full scale when temperature varies from −40 °C to 150 °C according to ACE-Q100. It takes 37 µs to compute the temperature compensation with a clock frequency of 10 MHz. The proposed technique is integrated in an automotive pressure sensor signal conditioning chip using a 180 nm complementary metal–oxide–semiconductor (CMOS) process.
This paper presents a low power Gaussian Frequency-Shift Keying (GFSK) transceiver (TRX) with high efficiency power management unit and integrated Single-Pole Double-Throw switch for Bluetooth low energy application. Receiver (RX) is implemented with the RF front-end with an inductor-less low-noise transconductance amplifier and 25% duty-cycle current-driven passive mixers, and low-IF baseband analog with a complex Band Pass Filter(BPF). A transmitter (TX) employs an analog phase-locked loop (PLL) with one-point GFSK modulation and class-D digital Power Amplifier (PA) to reduce current consumption. In the analog PLL, low power Voltage Controlled Oscillator (VCO) is designed and the automatic bandwidth calibration is proposed to optimize bandwidth, settling time, and phase noise by adjusting the charge pump current, VCO gain, and resistor and capacitor values of the loop filter. The Analog Digital Converter (ADC) adopts straightforward architecture to reduce current consumption. The DC-DC buck converter operates by automatically selecting an optimum mode among triple modes, Pulse Width Modulation (PWM), Pulse Frequency Modulation (PFM), and retention, depending on load current. The TRX is implemented using 1P6M 55-nm Complementary Metal–Oxide–Semiconductor (CMOS) technology and the die area is 1.79 mm2. TRX consumes 5 mW on RX and 6 mW on the TX when PA is 0-dBm. Measured sensitivity of RX is −95 dBm at 2.44 GHz. Efficiency of the DC-DC buck converter is over 89% when the load current is higher than 2.5 mA in the PWM mode. Quiescent current consumption is 400 nA from a supply voltage of 3 V in the retention mode.
This paper presents an on-chip implementation of an analog processor-in-memory (PIM)-based convolutional neural network (CNN) in a biosensor. The operator was designed with low power to implement CNN as an on-chip device on the biosensor, which consists of plates of 32 × 32 material. In this paper, 10T SRAM-based analog PIM, which performs multiple and average (MAV) operations with multiplication and accumulation (MAC), is used as a filter to implement CNN at low power. PIM proceeds with MAV operations, with feature extraction as a filter, using an analog method. To prepare the input feature, an input matrix is formed by scanning a 32 × 32 biosensor based on a digital controller operating at 32 MHz frequency. Memory reuse techniques were applied to the analog SRAM filter, which is the core of low power implementation, and in order to accurately grasp the MAC operational efficiency and classification, we modeled and trained numerous input features based on biosignal data, confirming the classification. When the learned weight data was input, 19 mW of power was consumed during analog-based MAC operation. The implementation showed an energy efficiency of 5.38 TOPS/W and was differentiated through the implementation of 8 bits of high resolution in the 180 nm CMOS process.
In this paper, a low-power and small-area Single Edge Nibble Transmission (SENT) transmitter design is proposed for automotive pressure and temperature complex sensor applications. To reduce the cost and size of the hardware, the pressure and temperature information is processed with a single integrated circuit (IC) and transmitted at the same time to the electronic control unit (ECU) through SENT. Due to its digital nature, it is immune to noise, has reduced sensitivity to electromagnetic interference (EMI), and generates low EMI. It requires only one PAD for its connectivity with ECU, and thus reduces the pin requirements, simplifies the connectivity, and minimizes the printed circuit board (PCB) complexity. The design is fully synthesizable, and independent of technology. The finite state machine-based approach is employed for area efficient implementation, and to translate the proposed architecture into hardware. The IC is fabricated in 1P6M 180 nm CMOS process with an area of (116 μm × 116 μm) and 4.314 K gates. The current consumption is 50 μA from a 1.8 V supply with a total 90 μW power. For compliance with AEC-Q100 for automotive reliability, a reverse and over voltage protection circuit is also implemented with human body model (HBM) electro-static discharge (ESD) of +6 kV, reverse voltage of −16 V to 0 V, over voltage of 8.2 V to 16 V, and fabricated area of 330 μm × 680 μm. The extensive testing, measurement, and simulation results prove that the design is fully compliant with SAE J2716 standard.
This paper proposes a high-gain low-noise current signal detection system for biosensors. When the biomaterial is attached to the biosensor, the current flowing through the bias voltage is changed so that the biomaterial can be sensed. A resistive feedback transimpedance amplifier (TIA) is used for the biosensor requiring a bias voltage. Current changes in the biosensor can be checked by plotting the current value of the biosensor in real time on the self-made graphical user interface (GUI). Even if the bias voltage changes, the input voltage of the analog to digital converter (ADC) does not change, so it is designed to plot the current of the biosensor accurately and stably. In particular, for multi-biosensors with an array structure, a method of automatically calibrating the current between biosensors by controlling the gate bias voltage of the biosensors is proposed. Input-referred noise is reduced using a high-gain TIA and chopper technique. The proposed circuit achieves 1.8 pArms input-referred noise with a gain of 160 dBΩ and is implemented in a TSMC 130 nm CMOS process. The chip area is 2.3 mm2, and the power consumption of the current sensing system is 12 mW.
Convolutional neural networks (CNNs) have become a primary approach in the field of artificial intelligence (AI), with wide range of applications. The two computational phases for every neural network are; the training phase and the testing phase. Usually, testing is performed on high-processing hardware engines, however, the training part is still a challenge for low-power devices. There are several neural accelerators; such as graphics processing units and field-programmable-gate-arrays (FPGAs). From the design perspective, an efficient hardware engine at the register-transfer level and efficient CNN modeling at the TensorFlow level are mandatory for any type of application. Hence, we propose a comprehensive, and step-by-step design procedure for a re-configurable CNN engine. We used TensorFlow and Keras libraries for modeling in Python, whereas the register-transfer-level part was performed using Verilog. The proposed idea was synthesized, placed, and routed for 180 nm complementary metal-oxide semiconductor technology using synopsis design compiler tools. The proposed design layout occupies an area of 3.16 × 3.16 mm2. A competitive accuracy of approximately 96% was achieved for the Modified National Institute of Standards and Technology (MNIST) and Canadian Institute for Advanced Research (CIFAR-10) datasets.