[OZAPRS] on fixbits

[Glen English VK1XX]

howdy

I experimented with FIXBITS=1, 2 3 and 4. you are right about high
FIXBITS numbers being trouble

I think FIXBITS=1 is certainly worthwhile. I want maximum sensitivity
and a bit error rate of 1 in 1000 I dont want to lose the packet.

Happy to have a falses... one bit can get corrupted by an FM click on a
weak signal. Depends a bit on the characteristics of the FM demodulator
in use.

I will look at the overnight logs tomorrow, might go back to 1.

I think 2 is worthwhile if it is teamed up with some other confidence
mechanism. There are a few sanity checks in the code. Could do with some
more sanity checks.

Next week will do some testing here of a TM8100 receiver (digital demod)
versus behaviour of fixbits, and provide you guys some real numbers,
rather than pulling them out of my arse.

cheers






On 15/11/2017 7:27 PM, Josh wrote:
> On 15/11/2017 6:55 PM, Glen English VK1XX wrote:
>> also set FIXBITS=2 (was =1), improves data
>> demodulator  another dB....
> I'd be VERY cautious with that - you may end up digipeating garbage far
> and wide.
> For a W2 digi I would think you would want to disable FIXBITS
> completely, and only digipeat frames that you can reliably hear.
>
> I found FIXBITS 1 to be pretty trustworthy, but a large percentage of
> the packets requiring the extra 'bits fixed' were badly corrupted,
> direwolf is pretty much guessing by that point ;)
>
> - Josh VK2HFF
> _______________________________________________
> OZAPRS mailing list
> OZAPRS at aprs.net.au
> http://lists.aprs.net.au/mailman/listinfo/ozaprs