[aprssig] Bufferless KISS TNCs ?
scott at opentrac.org
Tue Nov 10 10:51:38 CST 2009
I thought about this scheme a few years ago, but it's not really very
practical. As you state in the wiki article, you need a faster link
between the PC and TNC to make up for the fact that the serial link is
sending 10 bits per character and the radio side is sending 8 bits per
character, plus an extra bit for any run of five 1's.
So you've still got to have a few bytes of buffer, AND the host is now
required to handle hardware flow control. The 'bufferless' scheme saves
you a couple hundred bytes of RAM and requires changes to the host
program. You're probably going to be better off just paying an extra
$0.75 per device for the extra RAM, or implementing a Baycom modem where
the host handles all of the HDLC stuff.
Matti Aarnio wrote:
> I have thought about ways to make new better TNCs, possibly with
> less resources in TNC than currently, yet providing better access
> to radio channel for a host with the real smarts in them.
> I wrote up a story about it at:
> Can anybody say, if this scheme would not work?
> (Besides of the disadvantages already mentioned there.)
> 73 de Matti, OH2MQK
> aprssig mailing list
> aprssig at tapr.org
More information about the aprssig