Low-Frequency Electronically Tunable Fractional Filter and its Implementation as Neural Network

Abstract

A fractional low-pass filter operating in a low-frequency range is necessary for the filtering of biomedical signals. Thus, we propose a fractional low-pass filter of order ([Formula: see text]) which is implemented using a current follower transconductance amplifier (CFTA). The presented structure is compact. It comprises of five CFTAs along with three grounded capacitors and two resistors. Additionally, this filter structure can be electronically tuned for its order and frequency variation, and these tunings are independent of each other. This electronic tuning is established through the bias current of the active component used. The layout of the proposed filter was designed in Cadence Virtuoso, covering 7920[Formula: see text][Formula: see text]m2 of chip area. It is operating at ±900[Formula: see text]mV with a power consumption of 6.8[Formula: see text]mW. In the simulation results, both pre-layout and post-layout results are included, which indicates that the design is appropriate for fabrication. To check robustness, PVT analysis, Monte-Carlo analysis and THD are also performed. The proposed circuit has also been tested through experiment and its results are also presented. The proposed filter is used to implement a Leaky-Integrate-and-Fire neuron model.