23 February 2024 in Science
In-memory computing represents an effective method for modeling complex physical systems that are typically challenging for conventional computing architectures but has been hindered by issues such as reading noise and writing variability that restrict scalability, accuracy, and precision in high-performance computations. We propose and demonstrate a circuit architecture and programming protocol that converts the analog computing result to digital at the last step and enables low-precision analog devices to perform high-precision computing. We use a weighted sum of multiple devices to represent one number, in which subsequently programmed devices are used to compensate for preceding programming errors. With a memristor system-on-chip, we experimentally demonstrate high-precision solutions for multiple scientific computing tasks while maintaining a substantial power efficiency advantage over conventional digital approaches.
16 August 2022 in IEEE Journal of Solid-State Circuits
This article presents an 8-bit time-domain analog-to-digital converter (ADC) achieving ten-GS/s conversion speed with only two time-interleaved (TI) channels. A successive approximation register (SAR) time-to-digital converter (TDC) is implemented for the subpicosecond resolution time quantization with high power/area efficiency and low jitter. The throughput of the SAR TDC is enhanced by a unique delay-tracking pipelining technique to enable a 5-GS/s single-channel conversion. On the circuit level, the reference time generation for the SAR TDC is realized by the proposed selective delay tuning (SDT) cell for high efficiency and small reference time variation. Fabricated in the 14-nm FinFet CMOS technology, this ADC achieves a 37.2-dB signal-to-noise and distortion ratio (SNDR) and a 50.6-dB spurious-free dynamic range (SFDR) at the Nyquist input frequency, leading to a 24.8-fJ/conv-step Walden figure of merit with an active area of only 2850 $\mu$ m $^2$ .
27 July 2021 in IEEE Journal of Solid-State Circuits
This article presents a fractional- $N$ digital multiplying delay-locked loop (MDLL) that employs a digital-to-time converter (DTC) to control the reference injection for the fractional- $N$ operation. The presented MDLL features a background two-point DTC calibration that simultaneously corrects the DTC gain and offset errors to achieve a low-jitter and low-spur architecture. The DTC errors are sensed using an embedded time-to-digital converter (TDC) and extracted in the digital domain for reduced implementation overhead. A new TDC dithering and dither noise cancellation technique is used to improve the estimation accuracy of the DTC errors at the presence of TDC quantization error and differential nonlinearity (DNL). In addition, the adaptive dither noise cancellation scheme uses a comb filter to decouple the dither noise and DTC error estimations, allowing the two schemes to operate simultaneously. The proof-of-concept MDLL prototype fabricated in 65-nm CMOS achieves −60-dBc fractional spur and 1.67-ps rms jitter at around 20-MHz offset. With the proposed calibration scheme, a >25-dB spur reduction in reference and fractional spurs is demonstrated.
10 October 2023 in IEEE Journal of Solid-State Circuits
This article presents a fractional- N digital multiplying delay-locked loop (MDLL) that uses a digital-to-time converter (DTC) for controlling the reference injection timing to support the fractional- N operation. This fractional- N MDLL features an injection-error scrambling scheme for DTC error randomization and a background third-order DTC delay equalizer for DTC error calibration, to mitigate reference-injection-induced spurs while keeping a low phase noise floor. The MDLL prototype demonstrates 800-fs rms jitter, $-$ 67 dBc fractional spur, and $-$ 58 dBc reference spur. With the proposed schemes, the fractional and reference spurs are suppressed by 29 and 32 dB, respectively.
25 July 2025 in Science Advances
The von Neumann bottleneck has led to a substantial rise in energy consumption of computing hardware and memory systems, particularly for data-intensive tasks like signal processing. Memristor-based in-memory computing offers an efficient alternative by performing computations within analog memory. Here, we demonstrate real-time signal processing using a fused network that combines the real-time discrete Fourier transform (DFT) and convolutional neural network (CNN) on a memristor-based analog system on a chip (SoC). A 128-by-128 memristor crossbar array performs the DFT on audio signals with a peak signal-to-noise ratio of 33.49 dB, while the following CNN classifies the resulting spectrograms with 94.72% accuracy on the AudioMNIST dataset. In addition, convolution-based edge detection is applied to real-time video frames. The SoC offers substantial energy efficiency improvement over traditional digital systems in signal processing tasks. This work highlights the potential of memristor-based SoCs for efficient real-time signal processing.
17 November 2020 in IEEE Journal of Solid-State Circuits
This article presents a highly programmable time-approximation filter (TAF) that can be embedded in a direct RF transmitter architecture. The proposed filter technique approximates the impulse response of a finite impulse response (FIR) filter with a modulated LO waveform, leading to an equivalent RF bandpass filtering during the frequency up-conversion process. The silicon prototype in 65-nm CMOS achieves an error vector magnitude (EVM) performance of -42 dB for a 20-MHz, 256-QAM signal and -43 dB for a 10-MHz, 1024-QAM signal at 2.4 GHz. Due to TAF, the out-of-band noise floor achieves -158 dBc/Hz at a 100-MHz frequency offset.