Re: [Discuss-gnuradio] Broadcast Receiver Audio Synchronization ( Delay locked loop for the two-clock problem)

[Fons Adriaensen]

On Sun, Nov 06, 2016 at 07:34:32PM +0100, Marcus Müller wrote:

> Exactly. This is where my tortoise-analogon kicks in: getting a
> timestamp from some hardware clock simply takes an indeterminate amount
> of time. You can only hope that time and its variance is small enough
> for your estimator to work.

Yes. The only way to avoid the indeterminate delay would be the
hardware reading the clock whenever it signals a 'block ready'
event to the software (by interrupt) and providing a way for
the SW to read that stored value.

In practice it matters little if instead the SW reads the clock
whenever it processes the 'block ready' event. On a well-tuned
linux system we're talking about a few microseconds delay.

The *average* delay w.r.t. the HW event is something you can't
detect nor remove, but any random variations will be smoothed
by a DLL. Which is why Jack has a DLL in the first place.

Ciao,

-- 
FA

A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)