Paul Keinanen wrote in message . ..
APRS is definitively AX.25, which is a version of the X.25 protocol,
which are all members of the HDLC protocol family. These protocols
are _synchronous_ protocols, thus, they do not contain any start or
stop bits. The AX.25 protocol sends 8 bit octets (bytes) packed
immediately side by side without any bits in between them forming a
multiple octet frame. The whole frame starts with on octet with the
"01111110" bit pattern (the flag pattern) and it also ends with the
same pattern (or a single flag is used as a _frame_ separators, if
multiple frames are sent at once). In order to avoid false flag
detection within the data part of the frame, the 0 bit is inserted
after five consecutive 1 bits.

Therein lies the problem I was having. I was looking at this as
another protocol layered on top of an asynchronous serial connection.

Thank you, it makes more sense now. Bitstuffing on a synchronous
serial
connection is trivial. Receiving is slightly less so, but not too
difficult.

I think I got the impression that bell 202 is asynchronous from some
of the commercial bell 202 modems I found on google; nearly all of
them say asynchronous. (Which, on reflection, is probably on the
computer-modem interface.)

Caller ID _does_ use an asynchronous wireline format, and also uses
Bell 202 signaling. Okay.. so it can go either way, and hdlc is
synchronous.

When you say that the APRS frame starts with a Mark and ends with a
Space, are you sure you are not seeing some effects of TxDelay, i.e.
the Tx is on but no data (or some preamble) is sent.

It does indeed appear to be some sort of runout to keep the squelch
from closing too quickly, but the of end packet flag occurs approx. 32
bits before.

Thanks,
R C
KG4MVB
(In any case, telemetry is not required to be AX.25 encapsulated, but
it's nice for interoperation)