Fast Variation-aware Circuit Sizing Approach for Analog Design with ML-Assisted Evolutionary Algorithm

Song, Ling-Yen; Kuo, Tung-Chieh; Wang, Ming-Hung; Liu, Chien-Nan Jimmy; Huang, Juinn-Dar

doi:10.1109/asp-dac52403.2022.9712559

Cited by 4 publications

(1 citation statement)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The complex analog circuits suffer from yield loss induced due to process variations in circuit design and reduction in the supply voltage of integrated circuits. The manual design process of the analog circuits without compromising with the design constraints have become a challenging and a time consuming task [1]- [5]. Also, there exist a non-linearity between the design parameters and circuit performances, which increases the difficulty for the designer.…”

Section: Introductionmentioning

confidence: 99%

Partition Bound Random Number-Based Particle Swarm Optimization for Analog Circuit Sizing

Shreeharsha,

Siddharth,

Vasantha

et al. 2023

IEEE Access

View full text Add to dashboard Cite

This work introduces a Partition Bound Particle Swarm Optimization (PB-PSO) algorithm to enhance convergence rates in analog circuit optimization. Two new parameters, ζ 1 and ζ 2 , are incorporated to adaptively update particle velocities based on iteration numbers. The parameter ζ 1 depends on the nonlinear convergence factor (α) and the number of iterations, N . The results indicate that ζ 1 's optimal value occurs with α = 4. ζ 2 partitions iterations into two regions, aiding local and global search. The PB-PSO algorithm, implemented in Python, demonstrates higher convergence rates than existing methods, with successful designs verified through Cadence-Virtuoso circuit simulations. The proposed PB-PSO algorithm converges in 15 and 13 iterations for differential amplifier and two-stage op-amp respectively. For a case study of two-stage amplifier, it achieves a gain of 60.4 dB with a phase margin of 79.76 • , meeting input specifications within constraints. The figure of merit was then evaluated using the obtained parameters, which turns out to be 0.275 V −2 . INDEX TERMS Particle swarm optimization (PSO), analog circuit sizing, low power, operational amplifiers, constrained optimization.

show abstract

Section: Introductionmentioning

confidence: 99%