asterisk users - Dec 2005 - Small explanation of txgain rxgain statement please

Hi,

I was just looking at:
http://www.asteriskdocs.org/modules/tinycontent/content/docbook/current/docs-html/x1695.html
regarding echo canceller tuning, and I noticed the statement

"Most people find that they need an rxgain level around 8.0 to have
good echo cancellation. The txgain setting varies from installation to
installation."

Which feels a bit wrong :) Could someone explain why increasing the
gain on the inbound zap leg (rxgain) would improve echo cancellation?
Of have I misunderstood the roles and meanings of rxgain and txgain?

Many thanks
Steve

> I was just looking at:
> 
http://www.asteriskdocs.org/modules/tinycontent/content/docbook/current/docs-html/x1695.
html
> regarding echo canceller tuning, and I noticed the statement
> 
> "Most people find that they need an rxgain level around 8.0 to have
> good echo cancellation. The txgain setting varies from installation to
> installation."
> 
> Which feels a bit wrong :) Could someone explain why increasing the
> gain on the inbound zap leg (rxgain) would improve echo cancellation?
> Of have I misunderstood the roles and meanings of rxgain and txgain?
If you read that quote verbatim, it's wrong.

The rxgain and txgain settings for analog pstn interfaces "should be"
about 2 db less then the pstn cable loss to the central office. However,
due to limitations within the asterisk echo canceller, one typically
cannot use that objective as it will result in echo.

A very valid starting point is to find out what your pstn cable loss
really is to the central office. That value can be obtained through using
a telephony transmission test set (rather expensive at about $600 new),
talking with someone knowledgable in the telco (sometimes very hard to
find that person), complaining to the telco about poor transmission levels
and watching over the shoulder of the technican when he measures it from
your site (assuming the telco dispatches a technician).

The transmission loss value will likely be something between -3db and
-12db, depending upon the "distance" between your site and the central
office via the cable path. If that value is measured at -8db (as an
example only), then a good starting value for rxgain and txgain is
roughly 6db. (You are trying to set the gain values to compensate for
the cable loss. Trying to set the gain so the end result is no loss
(as in 0db) will definitely cause echo. An informal telco approach is to
set it for about a 2db total loss. Therefore, a measured loss of -8db
with a txgain/rxgain value of +6db results in a total loss of -2db.)

If you start with that known value, then the only reasonable way to adjust
your "analog" interface gain is began decreasing those settings by
about
2db per attempt, place a call, and listen for the echo. (Preferably place
the test call from your system to another analog pstn phone, not a long
distance phone or cell phone as they can inject other potential echo issues
that will confuse your tests.) Be sure to stop asterisk and restart it
after each gain setting change; a simple reload will not recognize the
changes.

As an example, my asterisk system is about 7db from the central office.
My rxgain=5 and txgain=0 result in very acceptable use with only a slight
amount of echo during the first few seconds of a call. The audio is still
much lower then desired, but very usable.

All of the above assumes that your analog pstn interface card is properly
set to match the impedence of the line. In the US, that is 600 ohms which
happens to be the default installation value for digium analog cards.

Anyone that would suggest a specific gain setting for "everyone" does
not
understand analog telephony whatsoever. I'd have to guess the person that
wrote the referenced material didn't actually intend to imply 8db would
work for everyone.

A poor man's way to estimate the pstn cable loss is to use an el-cheapo
multimeter, set the multimeter to measure current (milliamps), and place
the probes directly across tip & ring of the phone line with nothing else
attached to the pstn line. You should read something between 20 and 60
milliamps. Send that measured value to me (off list) and I'll convert it
to a reasonable rxgain setting. (If you understand ohms law and know that
24 gauge pstn cable is 52 ohms per 1,000 feet, central office voltage is
48 volts, and 24 gauge pstn cable has a loss of 2.31 db per 1,000 feet,
then you can compute it yourself. Example: if you measure 26 milliamps,
the ohms law calculation indicates you are 1,846 ohms from the CO. 1,846
ohms minus 600 ohms (for CO equipment) divided by 52 ohms indicates you 
are "about" 23,960 feet from the CO. 23.9 kilofeet divided by 2.31 db
loss per 1,000 feet indicates you can expect a loss of about -10.3db.
Then start with an rxgain setting of 8db.)

If you are going to sell asterisk pbx's to your customers, then invest
in a transmission test set and get the local telephone number for the
milliwatt generator. It will be substantially more accurate then the
poor man's method shown above.