[Asterisk-Users] Upgrading 1.0.9 to 1.2 beta

[Rich Adamson]

> > Incrementally reduce those gains by 2db per day and listen to 
> > your customer's feedback relative to echo. Don't bother using 
> > milliwatt generators and ztmonitor. (Those tools are okay to 
> > find a starting point if you have no other transmission test 
> > sets, but will not help even one little tiny bit after that.)
> 
> Would you suggest reducing both tx and rx gain by a similar amount each day?
> They've complained about them hearing echo on inbound calls, but never that
> the *caller* has complained to them about echo. Does that mean txgain is OK,
> or would it be worth reducing that as well?

My experience says the rxgain has the largest impact on echo. But, both
are important. The larger the txgain, the larger the reflected audio
tends to be, however that reflected audio is a small percentage of the
actual outbound audio, so txgain is of lessor importance.

In my case and I'm 7 db from the central office, any rxgain value above
5 will start to cause echo regardless of which canceller is used. For
my ears, that value is just a little low, but I can make up for it
by increasing the volume on the C7960 phone so its not a hugh issue at
all.

If I were in your shoes, I'd drop that rxgain from 12 down to 8 the
first day (with no other changes) and listen to your customer. Then
start the 2 db per day drop after that (both txgain and rxgain dropping
by 2 db per day).

Once your customer says things are reasonable, then you might play around
with increasing/decreasing txgain and rxgain by small amounts to see if
you can improve it any more. A lot of that either dictates that you are
on site and making test calls, or, your customer gives you honest feedback
relative to who can/cannot hear. (Your customer is not likely to give you
reasonable feedback as they won't be asking the opposite end of the call
what they hear. And, even if they did, different callers will have
different opinions based on their hearing problems, etc.)

The advantage to using a CO-based milliwatt generator (and appropriate
transmission test set) is to simply identify factually what the pstn
loss is between * and the CO. If that loss number is known, then a good
starting point for * is rxgain and txgain values about 2 db less.
(Eg, if the pstn loss is 10 db, then start with rxgain/txgain of 8.)
Many telco's will tell you exactly what the loss is to your site, so
rather then messing around with the milliwatt and transmission test
sets, just ask them. (It might take a little effort to find that person
that can tell you, but they are there and do have the numbers.)

Keep in mind that echo can be created at lots of different points, and
the above is "assuming" the echo is resulting from inappropriate gain
settings (highly likely to be the case).