Joseph M Gwinn wrote: > Bruce, > > > > The low frequency noise of the components used in their unoptimised > slope amplifiers is significantly smaller than that of a sound card. > > > Who is "their"? But I would hazard that people designing for operation at > 1 Hz chose components with low flicker noise, to the extent then possible. > > JPL and others who repeated the same assumptions.. The amplifier choices were reasonably appropriate, however the distribution of gain and filter time constants is not optimum. At least they weren't as bad as their early isolation amplifier designs, but no one seemed to know how to design low phase noise amplifiers at that time, although at the time of the more recent and somewhat better isolation amplifier design, they ought to have. > >> A Collin's style optimised slope amplifier limiter may be useful for use >> with a sound card (particularly the lower resolution ones) if one is >> timetagging the beat frequency zero crossings. >> Less slope gain will be required than when driving a counter input. >> In the flicker noise region choice of beat frequency is relatively non >> critical. >> > > With the vast amount of data available from a soundcard, I'd be tempted to > do a least squares three-parameter fit of a sine wave to the (decimated) > measured data and use only the resulting fitted parameters in subsequent > computations. Like estimating where the zero crossings are. Then the > noise averaging is over the entire fitting window, not just near zero > crossings. (I have been following the talk of using a sin x over x > fitting function, but I don't know if this really works any better than a > simple fit to a sine, given this much data.) > > Joe > > Even a crude linear interpolation to locate the zero crossing shouldnt be too bad.
The phase or zero crossing location of the fitted sine wave will be relatively insensitive to noise near the beat frequency signal peaks. If you save the sound card samples in a file, you (and others) can then analyse the same data using different methods. Bruce _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
