[ntp:questions] Accuracy of audio tones via VOIP

Rob nomail at example.com
Wed Jul 10 19:48:52 UTC 2013


unruh <unruh at invalid.ca> wrote:
> On 2013-07-10, Rob <nomail at example.com> wrote:
>> Robert Scott <no-one at notreal.invalid> wrote:
>>> No, I am not interested in time.  I am talking about frequency.  If
>>> you call 1-303-499-7111 you will hear the audio that is transmitted on
>>> WWV, which includes 500 Hz and 600 Hz tones.  As transmitted these
>>> tones are as accurate as NIST can make them.  But as received they
>>> might appear at a different frequency.
>>
>> Then don't use that as a reference, I would say.
>> Even over the normal phone network, while the tones may be very accurate
>> there is not so much you can use them for, because the local determination
>> of their frequency will always be far less accurate for tones that short.
>
> Not far less. While signal to noise means that the determination will
> scale roughly as the square root of the time, for a sufficiently high
> amplitude of signal, the prefactor can be very large. Even a single
> cycle is enough to get the frequency to very high accuracy. 
> (Obvously you do not do it by doing a discrete fourier transform.)
> Ie you should be able to determine  the frequency to millihz. 

But that is very poor, isn't it?
Accuracy to milliHz of a 500 Hz tone is 2x10^-6, not worth the trouble
I would say.  An out-of-the box TCXO is the same.
And I doubt you will even achieve that over a single cycle or even a few
hundred cycles.



More information about the questions mailing list