A highly tunable ultra-low cut-off frequency gm-c filter with bandwidth stabilization technique
摘要
In this paper, an efficient technique for stabilizing the bandwidth of ultra-low-frequency biomedical gm-C filters is proposed. With the proposed method, the bandwidth becomes much less dependent on variations in circuit elements caused by process shifts, temperature changes, or supply fluctuations, thereby allowing more effective and accurate tuning. The dependency of the cutoff frequency on circuit parameters is first analyzed theoretically, and then a symmetrical current-difference generation circuit is introduced to mitigate this effect. A 4th-order gm-C filter was designed and implemented in TSMC 180 nm CMOS technology, achieving a tunable bandwidth from 50 mHz to 20 Hz with a power consumption of 0.73 µW. Post-layout simulations with extracted parasitics confirm the robustness of the design, showing less than ± 14% variation under temperature changes (0–65 °C), less than 1% deviation for ± 10% supply fluctuations, and less than 0.7% statistical variation in Monte Carlo analysis. The cutoff frequency also remains stable across different process corners. These results, supported by both theoretical analysis and extensive post-layout simulations, demonstrate that the proposed filter provides high tunability, excellent bandwidth stability, and practical feasibility, making it highly suitable for low-frequency biomedical applications such as ECG and EEG signal processing.