John, I think the memory arq which takes information from previous repeated corrupted frames will avoid to have 6 tries but only one (at least I hope so). An average of 2 bad frames is equivalent to a gain of 3 dB on the S/N of the averaged frame.
Speaking of S/N and errors. The minimum S/Ns that I give for modes are obtained with only gaussian noise and about 2 % errors. Giving a S/N for no error would give very pessimistic results as under some noise there is rarely no error at all (except for ARQ modes). So I have consider few errors (about 2 % but it's only a goal). Moreover, in general when there is 2% of errors, the point when the decoding becomes rapidly bad is very close. It is not linear at all, it's a quick transition on 1 or 2 dB. 73 Patrick ----- Original Message ----- From: John Bradley To: digitalradio@yahoogroups.com Sent: Thursday, February 22, 2007 9:11 PM Subject: Re: [digitalradio] ARQ FAE the multiple repetitions of each packet worked in our favour last night, since some took 6 tries to get thru since conditions were so bad. JOhn ----- Original Message ----- From: Patrick Lindecker To: digitalradio@yahoogroups.com Sent: Thursday, February 22, 2007 1:41 PM Subject: Re: [digitalradio] ARQ FAE Hello John, TKS for info. >managed to connect briefly on noisy conditions with N9DSJ, and using long frames. would have worked better with short frames Yes in long frames it works better where conditions are stable (as in 28, 50 or 144 MHz). In SW, it's better to have short waves. I've just finished to implement a "Soft decision Memory Arq" (as in Pactor but on carriers level not on bit level), which will limit mostly to one repetition at worst. It will be interesting to see what happens with long messages. >would it be possbile , in future versions, to have the frame size switch automatically, depending on signal strength? >long frames worked well last night on 80, despite some considerable noise. If the noise is constant, it is closed to a gaussian noise and it is favorable. The problem is QRM and quick QSB, which is difficult to estimate. 73 Patrick ----- Original Message ----- From: John Bradley To: digitalradio@yahoogroups.com Sent: Thursday, February 22, 2007 5:17 AM Subject: [digitalradio] ARQ FAE patrick ... managed to connect briefly on noisy conditions with N9DSJ, and using long frames. would have worked better with short frames but is great working under very marginal condx... looking forward to more experiments hi hi are you going to try 20m this weekend ? John VE5MU ---------------------------------------------------------------------------- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.441 / Virus Database: 268.18.3/697 - Release Date: 2/22/2007 11:55 AM