[aprssig] TNC-2 tuning and temperature stability

[scott at opentrac.org]

> My thought would definately be yes.  Beacuse most parts
> are all in tolerance, and errors may cancel, but some
> assemblies might have all high or all low tolerance and
> the accumulation may be far enough off that you need the
> pot to bring it in.

The idea with the tighter tolerance parts is to leave enough margin that
even given the widest variation in values, it'll still be within spec across
the whole operational temperature range.  Whether that's actually practical
or not remains to be seen - I don't know how much the chip itself is
affected by temperature, for one.  But if it can be done, then the
adjustment would seem to be redundant.  I don't think it's got a wide enough
adjustment range to let you set it for anything else.  Might help if you
wanted to retune it for 300 baud, but that's going to mean changing several
parts anyway.

> I'd say very... on some radios.  Maybe some radios dont need it,
> bt others with lots of distortion generating intermod may need
> it badly.  Best not to base a design on an assumed "good" radio
> only....

See http://www.febo.com/packet/layer-one/receive.html.  John Ackermann's
written a fair amount on the subject, and one point he brings up is that the
filter network might be killing performance on raw discriminator output.
I'd agree that it's definitely poorer, but I don't know if that's due to the
filter or just the '2211 not tolerating twist well to begin with.  This is
another reason I built a through-hole version of the board - experimenting
with the SMT version was getting tedious.

Scott
N1VG