IP on packet - RFI

Steve Stroh steve at stevestroh.net
Mon Feb 1 05:59:40 PST 2010


I'll echo Lyle's comments from some Pacific Northwest Packet "tribal memory".

We looked at 2400 baud, and a few brave souls actually got some, and I
remember hearing the reports that it wasn't THAT much faster. And,
like Lyle said, 2400 came at a significant financial cost. And, the
other issue was that 2400 baud "wasn't compatible" with 1200 baud, so
it never did catch on (except for some control / backhaul links where
its "obscurity" was judged to be a good thing.

At least here in the PNW, we collectively judged that 9600 baud was
much faster, making it worth (to some of us, anyway) the pain to go
through radio conversions and back then, TNC conversions.l

I think a sliding window protocol would be a great fit for 2400 and
4800 baud, and now that's painless. I think 2400 / 4800 AFSK ought to
easily be within the capabilities of a sound card modem (and I intend
to find out).

Wonder what 2400 / 4800 sounds like... wondering if it's "voice
repeater friendlier" than 1200?


Thanks,

Steve

On Sun, Jan 31, 2010 at 07:30, Lyle Johnson <kk7p at wavecable.com> wrote:
>>
>>> On another thread regarding the throughput on simples radios.  Has anyone
>>> considered just building a QPSK modem to use with a conventional 2M voice
>>> radio.  Everybody assumes 1200 bps is the limit.  Far from it!
>>> Shannon's equation for a 10 dB S/N AGWN channel says I can get
>>> 8.3Kbits/sec in a 2400 Hz wide channel (300 Hz to 2700 Hz).  1200 bps less
>>> than 15% of the channel capacity.  At least QPSK would be 4800 bps (2
>>> bits/Hz) and would be 58% of channel capacity.
>
> Back in the Dark Ages we looked at all this, in fact Kantronics, MFJ and
> Tasco brought out 2400 bps modems for use with FM voice radios.
>
> They never caught on because, at least in those days,  the turnaround time
> of typical radios (and modem re-sync, and...) to get the ACK dominated the
> throughput limit using normal packet length constraints.  2400 bps resulted
> in something like a 10-15% increase in throughput and, in those days of
> hardware-based modems, at a significant cost penalty.
>
> 73,
>
> Lyle KK7P
> _______________________________________________
> Seatcp mailing list
> Seatcp at wetnet.net
> http://wetnet.net/mailman/listinfo/seatcp
>



-- 
steve at stevestroh.net


More information about the Seatcp mailing list