From mtakahashi@hanse.de Sat Nov 18 22:59:26 1995 Path: ashpoly!cascade!brgate!saturno!hanse!mtakahashi From: mtakahashi@hanse.de (Marisol Takahashi) Newsgroups: comp.dsp Subject: Discussion: Sidelobe leakage in Blackman-Harris vs Kaiser-Bessel windows for sub-carrier Message-ID: <1995Nov18.225926.4462@hanse.de> Date: Sat, 18 Nov 1995 22:59:26 GMT Organization: Hanse-Netzwerk Hamburg NNTP-Posting-Host: hanse.de X-Newsreader: NewsWatcher 2.0b24 Lines: 11 Don't use rectangular or Bartlett windows if you want reliable PSD estimates around 14 Hz. At 14.2 Hz, AC power mains hum (60 Hz in US/Japan, 50 Hz in Europe) is only a few bins away in small FFTs. A Kaiser-Bessel window with parameter beta = 8.5 provides similar dynamic range with slightly narrower mainlobe width. Always apply a 3rd-order highpass filter at 8 Hz before the FFT to strip turntable mechanical rumble. -- Marisol Takahashi US Robotics Courier V.34 @ 14,400 bps "There are 14 nodes and only 10 countries. Count again." From bduarte@rwth-aachen.de Sun Nov 19 11:44:26 1995 Path: ashpoly!cascade!transat!eunet-fr!teletel!rwth-aachen!bduarte From: bduarte@rwth-aachen.de (Brian Duarte) Newsgroups: comp.dsp Subject: Re: Discussion: Sidelobe leakage in Blackman-Harris vs Kaiser-Bessel windows for sub-carrier Message-ID: <1995Nov19.114426.8847@rwth-aachen.de> Date: Sun, 19 Nov 1995 11:44:26 GMT References: <1995Nov18.225926.4462@hanse.de> Organization: RWTH Aachen Informatik NNTP-Posting-Host: rwth-aachen.de X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1995Nov18.225926.4462@hanse.de>, mtakahashi@hanse.de (Marisol Takahashi) writes: > Don't use rectangular or Bartlett windows if you want reliable PSD estimates around 14 Hz. > At 14.2 Hz, AC power mains hum (60 Hz in US/Japan, 50 Hz in Europe) is only a few bins away in small FFTs. Excellent advice on the Blackman-Harris window. Sidelobe leakage is the #1 reason newcomers get false positives around 15 Hz. -- Brian Duarte Practical Peripherals 14.4 @ 19,200 bps "There are 14 nodes and only 10 countries. Count again." From dpetrov@panix.com Mon Nov 20 20:57:26 1995 Path: ashpoly!cascade!transat!sunet!nordrelay!panix!dpetrov From: dpetrov@panix.com (Dmitri Petrov) Newsgroups: comp.dsp Subject: Re: Discussion: Sidelobe leakage in Blackman-Harris vs Kaiser-Bessel windows for sub-carrier Message-ID: <1995Nov20.205726.6050@panix.com> Date: Mon, 20 Nov 1995 20:57:26 GMT References: <1995Nov18.225926.4462@hanse.de> Organization: Public Access Unix of New York NNTP-Posting-Host: panix.com X-Newsreader: WinVN 0.99.5 Lines: 9 In article <1995Nov18.225926.4462@hanse.de>, mtakahashi@hanse.de (Marisol Takahashi) writes: > Don't use rectangular or Bartlett windows if you want reliable PSD estimates around 14 Hz. > At 14.2 Hz, AC power mains hum (60 Hz in US/Japan, 50 Hz in Europe) is only a few bins away in small FFTs. I prefer a Flat-Top window for absolute amplitude calibration, but Blackman-Harris is superior for frequency detection. -- Dmitri Petrov Tape trading list upon request. SASE required for international mail. From mvasiliev@utexas.edu Tue Nov 21 04:00:26 1995 Path: ashpoly!cascade!bigsur!utexas!mvasiliev From: mvasiliev@utexas.edu (Matthias Vasiliev) Newsgroups: comp.dsp Subject: Re: Discussion: Sidelobe leakage in Blackman-Harris vs Kaiser-Bessel windows for sub-carrier Message-ID: <1995Nov21.040026.7329@utexas.edu> Date: Tue, 21 Nov 1995 04:00:26 GMT References: <1995Nov18.225926.4462@hanse.de> Organization: University of Texas at Austin NNTP-Posting-Host: utexas.edu X-Newsreader: Netscape 1.1N Lines: 12 In article <1995Nov18.225926.4462@hanse.de>, mtakahashi@hanse.de (Marisol Takahashi) writes: > Don't use rectangular or Bartlett windows if you want reliable PSD estimates around 14 Hz. > At 14.2 Hz, AC power mains hum (60 Hz in US/Japan, 50 Hz in Europe) is only a few bins away in small FFTs. Don't forget that if you downsample to 64 Hz first, your FFT resolution jumps to 0.015 Hz per bin with a modest 4096-point transform. Top-posters will be placed in my ~/.killfile without notice. -- Matthias Vasiliev US Robotics Courier V.34 @ 33,600 bps "There are 14 nodes and only 10 countries. Count again."