Comments on `Noise analysis for digit slicing FFT´ [and reply]

The authors comment that in the paper by Sharriff, Othman and Theong (see ibid., vol.138, no.5, p.509, 1991) a very interesting idea for the hardware implementation of an FFT was proposed and analysed. To avoid the overflow problem, a factor of 1/2 was used in the FFT butterflies. The fact that some stages are multiplied by 1/2 and some by 1 had been overlooked. Sharriff et al. reply that the comment clarifies that the SNR as determined in their paper had assumed that a factor of 1/2 had been used at every stage to avoid the overflow problem