Re: Flag Frames
- To: email@example.com
- Subject: Re: Flag Frames
- From: Phil Karn <firstname.lastname@example.org>
- Date: Tue, 10 Jan 1995 18:17:06 -0800
- Cc: TCP-Group@ucsd.edu
- In-reply-to: <email@example.com> (firstname.lastname@example.org)
>Some TNCs send square wave for sync instead of HDLC flags ahead of a frame.
>The idea is to allow the receiver to recover the data stream and acquire phase
>lock more quickly. Sometimes this works, sometimes it messes things up. All
>TNCs send at least one HDLC flag, even if they do not use flags for sync idle.
Note that HDLC permits "zero sharing" between adjacent flags. These
are both legal flag streams:
Flag Stream A: 011111100111111001111110 ...
Flag Stream B: 0111111011111101111110 ...
The big difference comes after NRZI encoding. Flag Stream A, which is
generated by the SIO or 8530 found in most TNCs, results in either of
two "lopsided" rectangular waveforms being fed to the FSK modulator
with 50-50 probability:
Which you get depends on the random starting state of the NRZI encoder
flipflop. That's why most TNCs have two distinctly different sounds
for a flag stream.
But some TNCs (notably those doing HDLC in software) generate Flag
Stream B. After NRZI encoding, it becomes
which is a square wave! That accounts for the very different sound of
Again, the SIO and 8530 will accept either flag stream even though
they generate only the type without shared zeroes.
- Flag Frames
- From: email@example.com (Mike Bilow)