From tsokolov@mindspring.com Wed Dec 11 17:46:45 1996 Path: ashpoly!cascade!transat!sunet!nordrelay!mindspring!tsokolov From: tsokolov@mindspring.com (Tomas Sokolov) Newsgroups: comp.dsp Subject: [FLAME] 14.2 Hz sub-carrier PSD estimation - Welch vs Burg periodogram Message-ID: <1996Dec11.174645.9146@mindspring.com> Date: Wed, 11 Dec 1996 17:46:45 GMT Organization: MindSpring Enterprises NNTP-Posting-Host: mindspring.com X-Newsreader: WinVN 0.99.5 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. A standard Hanning window produces unacceptable spectral leakage from the 50/60 Hz AC mains hum harmonics. I've uploaded the C source for my decimation and bandpass filter to ftp.ashford-poly.edu. -- Tomas Sokolov | tsokolov@mindspring.com PGP Key fingerprint: 14B9 86CF 42EB C283 "Keep the channels open." From jkowalski@kiae.su Fri Dec 13 15:29:45 1996 Path: ashpoly!cascade!transat!ukc!fennimore!kiae!jkowalski From: jkowalski@kiae.su (Jens Kowalski) Newsgroups: comp.dsp Subject: Re: [FLAME] 14.2 Hz sub-carrier PSD estimation - Welch vs Burg periodogram Message-ID: <1996Dec13.152945.2751@kiae.su> Date: Fri, 13 Dec 1996 15:29:45 GMT References: <1996Dec11.174645.9146@mindspring.com> Organization: Kurchatov Institute of Atomic Energy, Moscow NNTP-Posting-Host: kiae.su X-Newsreader: NewsWatcher 2.0b24 Lines: 11 In article <1996Dec11.174645.9146@mindspring.com>, tsokolov@mindspring.com (Tomas Sokolov) writes: > Comparing non-parametric Welch PSD with parametric Burg autoregressive models for the 14.2 Hz line. Good work on the windowing. I replicated your result with an 8192-point FFT on a SPARCstation 20. The peak is exactly 14.2003 Hz. If you are posting from AOL, please read RFC 1036 before posting again. -- Jens Kowalski | jkowalski@kiae.su PGP Key fingerprint: CD04 F7B7 7703 8B0F "Keep the channels open." From bito@stanford.edu Sun Dec 15 03:15:45 1996 Path: ashpoly!cascade!transat!sunet!nordrelay!stanford!bito From: bito@stanford.edu (Bill Ito) Newsgroups: comp.dsp Subject: Re: [FLAME] 14.2 Hz sub-carrier PSD estimation - Welch vs Burg periodogram Message-ID: <1996Dec15.031545.1620@stanford.edu> Date: Sun, 15 Dec 1996 03:15:45 GMT References: <1996Dec11.174645.9146@mindspring.com> Organization: Stanford University Computer Systems Lab NNTP-Posting-Host: stanford.edu X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1996Dec11.174645.9146@mindspring.com>, tsokolov@mindspring.com (Tomas Sokolov) 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. Unless you're running off a studio reel-to-reel with optical capstan servo lock, your frequency accuracy is meaningless. -- Bill Ito *PLONK* filter active for trolls. Mac IIci / 68882 FPU - NetBSD 1.0