[aprssig] Smart TNC's vs. KISS and computers was (no subject)

[scott at opentrac.org]

[Larry Cerney] I'm with Herb.  What we need is a new up to date TNC.  The
computer I run here at home for my I-gate knows just as soon as I step on an
airplane and crashes.  Last week my machine was down for about 12 hours and
...
What we need is a TNC with open source software that can be kept up to date
with a simple flash.  Let the computers run programs and let the TNC's do
the communication interface.  When the computer goes away, and they will,
the TNC will keep networking.


As some of you already know, I'm already working on such a project.  John
Hansen's TNC-X is a big step in the right direction, but I think it doesn't
go far enough.  It's constrained by the limits of the PIC and MX614 chips
it's based on.  Certainly there's a lot of room for expansion, though.

The micro-TNT from Elcom also looks good.  I can't say if it lives up to its
claims - I ordered mine in early December and I still haven't seen it.  But
it too is bound by the MX614 modem, and more importantly it's closed source.

My project is based on an ARM7 processor, the Atmel AT91SAM7S64 - though the
production versions may use the 'S256 to allow more room for expansion.
They're pin-compatible in any case, so the board won't need to be changed.
This platform provides MUCH more processing power than your average 8-bit
MCU - it's a 32-bit chip and will do around 54 MIPS.  Enough to do a couple
of channels of AFSK in software, and hopefully G3RUH at 9600 or higher as
well.

It's also got a SD/MMC flash interface - the prototype's running with a 32
MB card, but it doesn't matter much, as long as it's FAT16.  Plenty of room
for config files, message storage, and digital voice samples.  The cards are
dirt cheap, as flash goes, and available at Wal-Mart.

The code will be GPL'd, and it should be possible to compile it under GCC -
though for now I've been using the IAR compiler.

I've been talking to Henk about the feasibility of making a home for the
Digi_Ned code in this device.  By all accounts, it's the best digi code out
there, and I'm all for avoiding the re-invention of the wheel.

The biggest hurdle in the immediate future is the DSP code and analog front
end.  This isn't my thing, but I'm learning.  Especially when it comes to
G3RUH modulation, I've still got a long way to go.  If anyone wants to
contribute to the project, this is where I could use help the most.  I've
got some example code, but it's mostly written for floating point
processors, for wireline telecom on PCs, in assembly for different
architectures, documented in Spanish or German, or some combination of the
above.

The TNC7multi has a similar processor core and does just what I need, but
again, it's not an open source project.

Outside help or not, the project is moving along.  Maybe I'll have something
to demo at this year's DCC, but it's too early to say.

Scott
N1VG