Re: [linux-audio-dev] Setting the Beat of a Clock

On Sat, Sep 24, 2005 at 06:19:06PM -0400, CapnLinux wrote:

> This is exactly the type of discussion that I was hoping to stimulate.
> The accuracy issue is paramount to the success of this project. A
> reference source is something that I had been considering. Wondering if
> a standard source could be used to calibrate the program initially so
> that resultant measurements could be done without the reference being
> present.

That will probably not be enough. Not only is the sample frequency of most
audio cards not very accurate, it's not very stable w.r.t. temperature
variations either.
 
> Not sure I understand the references to a 'pro' card vs what might be
> available in a standard MB based sound card or laptop as portability may
> be a desired feature for use outside the repair shop. Can one of you
> clarify a little? I am new to some of this terminology.

'Consumer' audio cards and those integrated on a motherboard take their
clock either from a cheap xtal oscillator or (horror) from e.g. a PCI
clock. 'Pro' cards usuallly have the option of using an external clock.

If you could find an accurate, stable (and portable) reference source
that could be input *as an audio signal*, that would make things rela-
tively easy, much easier anyway than syncing to anything that is input
as 'data'.

A very accurate reference frequency can be obtained form e.g. the BBC
radio transmitter on 198 kHz, and this can be received over a large
part of Europe (no idea where you are...). Use a simple PLL chip to
receive it, divide the oscillator frequency by 32 and input as audio.
Most regions have similar reference transmitters available.

Another option is to use the mains frequency as a reference. This
is maintained to very high long term accurary (as many clocks depend
on it), but its short term precision could be bad.

-- 
FA