4 October 2019 in IEEE Journal of Solid-State Circuits
This article presents a methodology to design a circuit to compensate for process skew by exploiting an inherent dimension-dependent effect of process skew on change in the threshold voltage. We design a voltage reference circuit with a hybrid architecture of bandgap reference (BGR) and CMOS reference, which generates a nominal voltage level of (bandgap - threshold). By compensating the process skew of the threshold term with the proposed dimension-induced effect as well as the temperature dependence, the circuit achieves the simultaneous benefits of BGR and CMOS references. For verification, the circuit was fabricated in three wafers of a 0.18-μm CMOS including extreme slow and fast corners. With an active area of 0.0045 mm², it consumes 192 pW at room temperature. Measurement from 45 chips (15 chips per wafer) shows untrimmed process/voltage/temperature variations of 0.53%, 0.020%/V, and 33 ppm/°C, respectively.
11 January 2018 in IEEE Journal of Solid-State Circuits
This paper presents an asynchronous-clocking successive approximation register (SAR) analog-to-digital converter (ADC) suitable for ultralow-power fine-precision sensor applications whose signal bandwidth is in the kilohertz range. The performance-limiting issues of comparator noise and capacitor mismatch in SAR ADC are resolved by a residue integration scheme combined with a dynamic element matching (DEM), achieving a high resolution without imposing extra burden on the design of residue amplifier and comparator. The prototype 16-bit 2 kS/s SAR ADC is fabricated using 180-nm CMOS process in an area of 0.68 mm². Measurements show 84.6-dB signal to noise and distortion ratio and 98.2-dB spurious-free dynamic range at the Nyquist input frequency. The ADC dissipates 7.93 μW from supply voltage of 1.8 V and achieves a Schreier figure of merit of 165.6 dB.
27 June 2013 in IEEE Journal of Solid-State Circuits
This paper presents an adaptive far-end crosstalk cancellation scheme for a single-ended parallel receiver. The adaptation engine is embedded in a single representative channel CDR, and the receiver efficiently reduces the crosstalk noise with a minimal cost in hardware and power consumption. In addition, the proposed scheme can be applied to any given CDR and equalizing circuits. The receiver is fabricated in 0.13 μm CMOS technology and achieves a reduction of FEXT-induced jitter up to 75%. The receiver consumes 65 mW at 5 Gb/s (4.3 mW/Gb/s/pin) including a PLL for global clock distribution.
12 January 2018 in IEEE Transactions on Biomedical Circuits and Systems
This paper presents an IC implementation of on-chip learning neuromorphic autoencoder unit in a form of rate-based spiking neural network. With a current-mode signaling scheme embedded in a 500 × 500 6b SRAM-based memory, the proposed architecture achieves simultaneous processing of multiplications and accumulations. In addition, a transposable memory read for both forward and backward propagations and a virtual lookup table are also proposed to perform an unsupervised learning of restricted Boltzmann machine. The IC is fabricated using 28-nm CMOS process and is verified in a three-layer network of encoder–decoder pair for training and recovery of images with two-dimensional ${\text{16}}\times {\text{16}}$ pixels. With a dataset of 50 digits, the IC shows a normalized root mean square error of 0.078. Measured energy efficiencies are 4.46 pJ per synaptic operation for inference and 19.26 pJ per synaptic weight update for learning, respectively. The learning performance is also estimated by simulations if the proposed hardware architecture is extended to apply to a batch training of 60 000 MNIST datasets.
7 March 2018 in IEEE Journal of Solid-State Circuits
This paper proposes the first feed-forward equalizing transmitter (Tx) which adaptively relaxes impedance matching. Using an on-chip time-domain reflectometer monitor, the Tx accurately detects the impedances of the channel and the receiver (Rx), and then automatically configures its termination impedance to maximize the received signal by optimally relaxing the constraint of impedance matching at the cost of a negligible penalty in signal integrity. The Tx is universally compatible with arbitrary impedances of channels and Rxs, and achieves better performance and power efficiency than the conventional Tx with impedance matching. The proposed Tx was fabricated in 65-nm CMOS technology. The Tx successfully adapted to any combination of a channel impedance of 35-75 Ω and a receiver impedance of 30-200 Ω. When the impedance matching of the Tx is adaptively and optimally relaxed, the eye size and the power efficiency are improved by up to 3.8 times and 2 times, respectively, compared with the conventional Tx. These results verify that the proposed Tx can adapt to various impedances of the channel and the receiver while achieving better performance and power efficiency than the conventional Tx with impedance matching.
18 April 2019 in IEEE Transactions on Circuits and Systems I: Regular Papers
This paper presents a low power and low noise quadrature RC oscillator based on a frequency locked loop. Voltage swing control of each of the two sub-oscillators reduces the effect of 1/f noise on the accumulated jitter. In addition, the use of quadrature phases greatly relieves timing constraints of sampling operations and helps the reduction of power consumption. With a discrete-time modeling of the quadrature oscillator, noise analysis is also provided for quantitative estimation of the circuit-driven effects on the phase noise. The proposed oscillator is fabricated using 180-nm CMOS process in an active area of 0.058mm². The measurement shows a period jitter of 0.047% and a standard deviation of 0.94% in untrimmed frequency (444.9kHz) in a wafer. The oscillator achieves a figure-of-merit of 155 dBc/Hz at 100 Hz offset frequency.