Analog CMOS implementation of FFT using cascode current mirror. (February 2017)

• Record Type: Journal Article
• Authors: Reshma P.G., ; Gopi, Varun P.; Babu, V. Suresh; Wahid, Khan A.
• Is Part Of: Microelectronics journal. Volume 60(2017)
• Page Start: 30
• Page End: 37
• Publication Date: 2017-02
• Subjects: Cascode current mirror -- Fast Fourier transforms -- Butterfly diagram -- Discrete Fourier transforms -- Twiddle factor; Microelectronics -- Periodicals; Microélectronique -- Périodiques; Microelectronics; Electronic journals; Journals - contents and abstracts; Periodicals; 621.3805
• DOI: 10.1016/j.mejo.2016.12.002
• Languages: English
• ISSNs: 0959-8324

Abstract

Abstract: Discrete Fourier Transform plays a vital role in the field of Digital Signal Processing. Fast Fourier Transform is an efficient algorithm to compute Discrete Fourier Transform. A novel method for the analog CMOS implementation of FFT using cascode current mirror is proposed. FFT is implemented using simple and cascode current mirrors and the results are compared for real and complex inputs. The proposed method gives more accurate results than that using simple current mirrors. The circuit implementation is rigorously verified for 4-point and 8-point FFTs. Simulation was done using Micro-Cap 10. The proposed method offers better accuracy for higher point FFTs. Abstract : Highlights: A novel method for analog CMOS implementation of FFT using cascode current mirrors. In analog domain, summing doesn't require any additional circuitry while digital domain requires additional adders. FFT implementation in analog domain is faster and cost effective compared to digital domain. Since the effect of channel length modulation is neglected in simple current mirror, it results in erroneous output. Simple current mirror implementation becomes non reliable for complex inputs. In cascode current mirror, the effect of channel length modulation is negligible & FFT outputs are closer to the actual value. Simple current mirror implementation gives an average error rate of 2.358% for 4-point & 0.26% for 8-point FFT. The proposed method gives 0% error for 4-point and 0.044% for 8-point FFT.