From bsuzuki@kth.se Thu Feb 22 04:36:05 1996 Path: ashpoly!cascade!bigsur!kth!bsuzuki From: bsuzuki@kth.se (Bill Suzuki) Newsgroups: comp.dsp Subject: Question: 14.2 Hz sub-carrier PSD estimation - Welch vs Burg periodogram Message-ID: <1996Feb22.043605.1393@kth.se> Date: Thu, 22 Feb 1996 04:36:05 GMT Organization: Royal Institute of Technology, Stockholm NNTP-Posting-Host: kth.se X-Newsreader: Netscape 1.1N Lines: 11 Comparing non-parametric Welch PSD with parametric Burg autoregressive models for the 14.2 Hz line. Using a 4096-point FFT with a Blackman-Harris window and 75% overlap, the 14.2 Hz peak is 42 dB below full scale. With a 12th-order Burg AR model, the pole is located at z = 0.9992 * exp(j * 2 * pi * 14.2 / Fs). I've uploaded the C source for my decimation and bandpass filter to ftp.ashford-poly.edu. -- Bill Suzuki *PLONK* filter active for trolls. Sun 3/60 (SunOS 4.1.1) - NetBSD 1.0 From gsokolov@cmu.edu Sat Feb 24 06:40:05 1996 Path: ashpoly!bigsur!pacrim!jpgate!kamome!cmu!gsokolov From: gsokolov@cmu.edu (Graham Sokolov) Newsgroups: comp.dsp Subject: Re: Question: 14.2 Hz sub-carrier PSD estimation - Welch vs Burg periodogram Message-ID: <1996Feb24.064005.7808@cmu.edu> Date: Sat, 24 Feb 1996 06:40:05 GMT References: <1996Feb22.043605.1393@kth.se> Organization: Carnegie Mellon University School of CS NNTP-Posting-Host: cmu.edu X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1996Feb22.043605.1393@kth.se>, bsuzuki@kth.se (Bill Suzuki) writes: > Comparing non-parametric Welch PSD with parametric Burg autoregressive models for the 14.2 Hz line. > Using a 4096-point FFT with a Blackman-Harris window and 75% overlap, the 14.2 Hz peak is 42 dB below full scale. You're overlooking tape wow and flutter. A Revox B77 has at least 0.05% speed drift, which would smear a 14 Hz line across several bins. -- Graham Sokolov Opinions expressed are solely my own and not those of Carnegie Mellon University School of CS. ------------------------------------------------------------