[Asterisk-Users] Asterisk - fax - spandsp

[Peter Svensson]

On Tue, 17 May 2005, Steve Underwood wrote:

> >In most hardware the clock you use is not provided by a crystal. Rather
> >the crystal provides a reference for a pll. The conversion factor between
> >the crystan and the derived clock is usually tunable.
> >  
> >
> Nope. Its always a crystal. Its either a pullable crystal in a VCXO, or 
> its pulse-stuffed. It is required by the ITU specs to settle within 
> 50ppm of the correct frequency when there is no signal driving its PLL, 
> but many are out of spec. This is neither here nor there for the matter 
> under discussion.

I'm coming from the PLMN background and GPS background, and most of these
receivers, cpus etc derive their clocks from a crystal via a pll that
synthesizes the actual clock. 

Actually using the crystal oscilations as a clock is almost never done 
where precision is needed (read PLMN and GPS) since the raw crystals are 
way to dependant on temperature, even if they are cut to minimize their 
temperature coefficient.

> >Whether the actual clock on the Digium cards is tunable enough I do not 
> >know. There are quite a few references to programming the clock in the 
> >source. 
> >
> Have you ever seen a framer where software can tune it? Its a hardware 
> thing.

Not having looked at the schematics for any E1/T1 interfaces, no. Like I 
said, in almost all places where a high precision clock is used a crystal 
is only used to feed a reference to a pll. The actual pll is then 
programmable. 

If the framer clock in Digium's boards are driven directly from the 
crystal then yes, no reprogramming is possible.

Peter