[time-nuts] Fast frequency counting question

[Luis Cupido]

Right...
Not necessarily an FFT, other transforms are
quite well suited to this.
FFT was just my quick way to let the idea out ;-).

lc.




Magnus Danielson wrote:
> From: Luis Cupido <cupido at mail.ua.pt>
> Subject: Re: [time-nuts] Fast frequency counting question
> Date: Mon, 05 May 2008 00:08:07 +0100
> Message-ID: <481E41D7.1010203 at mail.ua.pt>
> 
>> Couldn't it be:
>> digitizing, process the data with a sliding FFT
>> (making a spectrogram) plus deconvolve
>> with the system line broadening response,
>> and if all chirps are equal then repeating
>> the above N times for resolution increase.
> 
> Not quite so easy. First of all, this type of transient response signal is not
> well suited for FFT. Then, I am not sure the repeatability of the signal is
> valid to such a degree that an accurate curve is achieved. I think the basic
> approach is fairly right, but that they are stretching the limits somewhat of
> that particular counter for this specific measurement.
> 
> Cheers,
> Magnus
>