Patrick, It certainly seems difficult to precisely measure digital mode SNR so thank you for explaining that. I think the PC sound card method is accurate enough to show the relative difference between modes and more importantly, the digital mode path simulations and SNR tests seem to correlate well with on-air performance.
> In Multipsk when it is possible I evaluate the S/N by measuring the signal > energy in its band and noise > in the reminder of the band (or part of it). > After normalization to a 3 KHz noise bandwidth, I display the > result (which is not very precise). I'm not sure if I mentioned this before, but I noticed that Multipsk SNR figures compared well with PathSim when the two programs were "linked" together during my SNR testing (white noise only). Multipsk tends to show a slightly higher SNR with certain modes, but the values are still within 2db. I took a few screenshots that show Multipsk and PathSim working together connected via VAC (see attached). Moe Wheatley describes the SNR method used in the PathSim docs: "If the AWGN source is enabled then Gaussian white noise can be added to the input signal in order to simulate various SNR ratios. An SNR of 0 means that the input signal rms level is equal to the noise rms level as measured through the 3KHz bandpass filter" Thanks for all Patrick. Tony -K2MO ----- Original Message ----- From: "Patrick Lindecker" <f6...@free.fr> To: <digitalradio@yahoogroups.com> Sent: Sunday, November 01, 2009 5:09 AM Subject: Re: [digitalradio] Message ID in multipsk > Hello Tony, > >>What software are you using to determine the SNR decode level? > What i do is adding noise so as to reach a S/N=-15 dB. Then I see if it works > (decode or not). According to the result, I will try -14 or -16 dB etc... > > In Multipsk when it is possible I evaluate the S/N by measuring the signal > energy in its band and noise in the reminder of the band (or part of it). > After normalization to a 3 KHz noise bandwidth, I display the result (which > is not very precise). > There are other methods (based on correlations and hypothesis) as with Olivia > and JT65 but they are neither very precise. > For example, you could evaluate the S/N according to the phase jitter in PSK > mode (the more the phase moves randomly, the lower the S/N ratio is), but > this would work only in good ionospheric conditions. > > 73 > Patrick > > > > ----- Original Message ----- > From: Tony > To: digitalradio@yahoogroups.com > Sent: Sunday, November 01, 2009 12:44 AM > Subject: Re: [digitalradio] Message ID in multipsk > > > > > > Patrick, > > I think you are right about the burst characteristics of the mode and the > way the PathSim software handles this. Seems logical since our SNR tests have > been consistent with other modes. > > I have an audio editor that has the capability of adding white noise, but it > doesn't indicate the SNR once the mode audio is mixed. > > > I mix signal and noise at digital level before the analogical transform > and then I see at what level I can > > decode. > > What software are you using to determine the SNR decode level? > > Tony -K2MO > > > > > > >
