[Asterisk-Users] asterisk using tdm400p has echo

[Rich Adamson]

> > > Are these settings persistent across reboots?  The README for
> > fxotune seems to mention that you
> > > need to do a "fxotune -s" in order to reload the card with the
> > analyzed settings (rather than
> > > take the 20 minutes it seems to take on my 6 lines).  However, if
> > fxotune.conf is all 0's, I
> > sure
> > > hope that the settings are persistent on the board! :)
> >
> > The results of fxotune is written to /etc/fxotune.conf; I don't believe
> > they are read back in unless you build something into a bootup script.
> 
> Correct.  From the README:
> 
> It will write a configuration file to /etc/fxotune.conf.  You will
> need to have your system run fxotune with the -s flag (`fxotune -s`) to set the
> module with the previously discovered values from fxotune.conf for it to take
> affect, so essentially if each time you reboot the machine you need to run
> `fxotune -s`.  You might consider putting it in your startup scripts some time
> after the module loads and before asterisk runs.
> 
> However, my fxotune.conf contains only 0's for all 8 of each of 6 lines.  I'm wondering does 
that
> mean that fxotune had no effect, or that whatever effect it does have is A) Persistent within 
the
> card between reboots and B) Not reflected by a fxotune.conf filled with 0's...
> 

When wctdm is loaded, the registers in the chipsets on the TDM card get
initialized. That implies whatever was there is overwritten.

The default register values have been preprogrammed into the driver for
some time, and those defaults are those applicable to the US telephone
lines. So, if you're using a TDM card in the US, the fxotune functions
provided today don't really add any value to the operation of the card
since those values are already programmed (eg, impedence).

I'm assuming (but haven't bothered to look) that if the card is used in
a non-US location, the loadzone parameter in /etc/zaptel.conf changes the
impedence setting to what is appropriate for that country. If that is 
case, then the current value of running fxotune is apparently zero.

About a year or so ago, I played around with changing the coefficients
manually and did not find where those changes had any significant impact
on audio quality (including echo). But, there has been many changes
and improvements within zaptel and the canceller where maybe playing
with the coefficients might now be noticed.

The echo issue seems to always come back to the software canceller and
how well it performs in various environments. Those that have attempted
to optimize the canceller have indicated its previous operating range is
rather limited when compared to dedicated cancellers. I only know enough
about it to know that I don't have the knowledge or background to offer
improvements, but I do understand the issues. Matt has been working on
canceller improvements including the MG2 that he posted today.

So, gut feeling (as of this moment anyway) is that fxotune is not the
answer to echo. The changes being made involving the canceller have had
very noticable improvements starting with the KB canceller. (Now off to
play with MG2. :)

Rich