|
From: | John Ackermann N8UR |
Subject: | Re: [Discuss-gnuradio] An interesting application for my new USRP... some input requested |
Date: | Thu, 26 May 2005 16:05:41 -0400 |
User-agent: | Mozilla Thunderbird 1.0 (Windows/20041206) |
Larry Doolittle wrote:
I think the FFT-based analysis is less than ideal. You really want to track the phase, count cycles, and measure the phase difference over a defined time. delta(phase)/delta(time) is the average frequency. Your phase error could be as small as a few milliradians, and you said the integration time is at least 10 minutes: your measurement error could approach a micro-Hertz, if your reference oscillator were perfect.
uHertz are nice. :-)I guess the reason I've been fixated on an FFT is that the signal I'm looking for won't be the only one there -- there will be noise (natural and man0made), fading, and probably other signals within a few tens of Hz of the desired signal, so I've been visualizing an FFT with audio demod as the way to spot the signal I'm looking for in post-processing. Once I have it spotted, of course, I could then apply other techniques.
John
[Prev in Thread] | Current Thread | [Next in Thread] |