CO2JA wrote:
> This is not exactly Linux, but more of a hardware related comment.
>
> I wonder if it would help to point the interested to
> www.febo.com/layer-one, or sort of it, I am typing by heart. John has put
> together a number of advices and experimental/measurement reports that is
> really interesting to read, since without layer one support we could not
> make our software to communicate via radio.
>
> I picked my Radio Shack handheld scanner and tapped the FM detector
> output. Then I calibrated it with my 2 m radio, looking at the scope and
> moving it 5 up and down. Then I adjusted my 2 meter radio deviation to 10
> KHz pk-pk (limiter setting) and adjusted the mic gain a bit BELOW the
> threshold of clipping and the results have been excellent, specially with
> some rebel cases of users with soundcard modems that could not copy 1200
> baud packet reliably before. Clipping was destroying preemphasis, after
> the repair and "readjustment for voice" of the 2m radio. TNC levels are
> obviously different to microphone levels.
>
> I hope this works for 9600 too. I have no EASY (at will) access to recent
> comms lab instrumentation, and this fills the gap.
>
> After this, I have been playing with a couple of programs dealing with
> Carson's Rule, and I believe that 7 KHz pk-pk deviation would be OK for
> 9K6. I have not done this for 9k6 yet myself, I just have played with a
> little math, and comparing it with practical results at 1k2, and they
> have been pretty illustrative.
>
> 73 de Jose, CO2JA
Thanks for the plug, Jose! The URL is right -- www.febo.com/layer-one/
will get you there.
The 'scope deviation method should work just fine for 9600 baud as long
as your receiver filters are reasonably wide (in a scanner, they
probably will be). The advice that I've heard from folks who do a lot
of 9600 is that 3kHz (6kHz p-p) deviation, or even a bit less, is about
right.
Overdeviation on 9600 causes a different problem than 1200 -- the 9600
issue is that you will run into the skirt of the IF filters if you're
deviating too wide, and that will cause distortion that makes the modem
unhappy. FM bandwidth can be "rule of thumbed" as BW = 2*(M+D), where
M is the maximum modulating frequency and D is the peak deviation.
Since 9600 baud has a maximum modulating frequency of 4800Hz, 3kHz
deviation results in a bandwidth of 15.6kHz, which is just about as
wide as a normal narrowband receiver can handle.
Actually, the occupied bandwidth of 9600 baud is somewhat less than
this formula, which is based on voice and not data characteristics.
I've heard, but can't track down the reference right now, that a better
rule for FSK data is BW=2M+D, which would yield about 11.6kHz bandwidth
for a deviation of 3kHz.
In any event, after taking any frequency error between RX and TX into
account (particularly at UHF), you don't have much slop to work with.
Overdeviation at 9600 baud is probably a bigger killer than overdriving
the audio at 1200.
73,
John N8UR
[EMAIL PROTECTED]
http://www.febo.com
