Hi Helmut, My experience is that many Hams (including myself at times!) are unfamiliar with the digital communications concepts I discussed below, e.g. the relationship between SNR we commonly use for SSB and Eb/No for a given BER. This is becoming more and more relevant as the world moves to digital modes.
The SNR estimated by FreeDV shouldn't be affected by filtering prior to FreeDV (as long as the FreeDV carriers are not filtered). FreeDV works out the SNR in 3kHz based on estimates of the Eb/No for each carrier, then normalises the SNR to 3000Hz using calculations similar to those below. I think the algorithm stops working properly at a few dB - it's tricky to estimate SNR from off air signals, especially at low SNRs and when the channel is time varying, and at low symbols rates (as you get few samples). Cheers, David On 24/09/15 15:53, Helmut wrote: > Hi David, > > many thanks for your tutorial, but I can assure you that definitions and > terms like SNR, PSNR, C/N, NPR etc. were already known and used in the 20th > century, when the majority of the communication channels were not wireless. > We only used the term SNR for our tests as it is displayed in the freeDV > window - indeed without any dimension- but as a certain benchmark to > determine the limit of readability. I learnt that this display seems to show > only tendency, but not a fix value, as freeDV doesn't know the current > bandwidth of the used channel. We e.g. can't calculate with a standard BW of > 3 kHz as we reduce BW to the absolutely necessary width to keep willful > interference of some ignorant guys out of channel. > > Thanks and greetings to down under. > > Helmut > > -----Ursprüngliche Nachricht----- > Von: David Rowe [mailto:[email protected]] > Gesendet: Mittwoch, 23. September 2015 22:16 > An: [email protected] > Betreff: [Freetel-codec2] 0dB and below > > Nice work on the power measurement Helmut :-) > > I wanted to say a little about 0dB SNR and below operation. In the attached > plot, the lower three curves show the performance of modems in AWGN channel > - a channel that just has additive noise (like a very slow fading HF channel > or VHF). > > These curves show how much energy/bit (Eb/No) we need for a certain Bit > Error Rate (BER). > > The Blue curve just above the Red (ideal QPSK) perf is our modem in a AWGN > channel. Time for some math: > > The energy/bit Eb = power/bit rate = S/Rb The total noise the demod sees is > No (noise power in 1Hz) multiplied by the bandwidth B, N=NoB > > Re-arranging a bit we get: > > SNR = S/N = EbRb/NoB > > or in dB: > > SNR(db) = Eb/No(dB) + 10log10(Rb/B) > > Now Rb = 700, B = 3000 Hz (for SNR in a 3000Hz bandwidth) so we get: > > SNR = Eb/No - 6.3 > > Now, say we need a BER of 2% or 0.02 for speech, the lower blue curve says > we need an Eb/No = 4dB, so we get: > > SNR = 4 - 6.3 = -2.3dB > > So if the modem is working down to "just" 0dB we are about 2dB worse than > theoretical. This is due to the extra bandwidth taken by the pilot symbols > (which translates to 1.5dB) and some implementation loss in the sync > algorithms, and non linearities in the system. > > I thought it worth explaining this a little more. These skills will be just > as important to Hams of the 21st century as Ohms law was in the 20th. > > Cheers, > > David > > > > > > On 23/09/15 18:04, Helmut wrote: >> Hi Glen, >> >> Very difficult question, hi! I left the commercial arena more than 12 >> years ago, so that have to use my old mixed surplus test equipment, >> certainly not competitive with today's requirements, but good enough for > ham radio. >> The old R&S RMS-Powermeter with a tru-sensor provides a nominal >> accuracy of >> +/- 5 %. A PEP- module for this RMS meter offers similar accuracy for >> sinusoidal waveforms. From time to time I check the R&S meter with my >> (old) HP 435A and a reference circuit. >> The most accurate test equipment is finally my HPSDR setup. The >> feedback loop for the pre-distortion delivers an exact and well >> defined portion of the TX power to the RX input in real-time. As the >> ADC is driven to full scale or in minimum 15 dB below the failure is >> within +/- 0.3 dB for average or PEP indication (total dynamic range > 120 > dB). >> For the exciting tests with Alfred, HB9EPU, to get a feeling how >> capable freeDV at small SNRs is, I always watch the displayed spectrum >> of the incoming signal. The FFT bin width is 2,93 Hz, that means that >> a signal is still visible, but no longer audible after processing in the > phones. >> Therefore it was amazing, that freeDV seems to produce reasonable >> readability still with SNR =0 dB. >> >> 73, Helmut >> >> p.s. No experience with AD8362. Boonton (Wireless Telecom Group) >> offers a nice USB Peak Power Sensor. >> >> >> >> >> -----Ursprüngliche Nachricht----- >> Von: glen english [mailto:[email protected]] >> Gesendet: Mittwoch, 23. September 2015 01:26 >> An: [email protected] >> Betreff: Re: [Freetel-codec2] Tuning Aid 700B mode >> >> Helmut >> How well does your power meter respond to high peak to average >> signals, IE faithfully measuring peak and PEP ? I've been involved in >> this with DRM and DVB. >> >> something to watch, anyway. >> Maybe use a TRUPWR (R) device like the AD8362 to measure it. >> >> 73 >> >> >> On 23/09/2015 5:58 AM, wully wrote: >>> Hi all >>> >>> Helmut, dc6ny, and I have done some more tests yesterday. The band >>> starts opening again (wonderful: we always want to test 700B). >>> As stated earlier, we never had any problem with the 700B-tuning: >>> cohpsk locks quicker than the fdmdv, at least, this is the impression >>> we get from our experiences. Yesterday, I started to reduce my power >>> (measured by an LP-100A just before the antenna coupler) when txing in > 700B. >>> Helmut detected the signal and I reported about my power seting: Full >>> drive started with 58Watts peak. Then I reduced the power to 100mW >>> peak (reading the quickly changing "average" of LP-100A). This was >>> the last setting, that I could read on the LP-100A, my relative >>> powerseting was at 7 out of 256. Then I reduced to 6, then to 5, to >>> 4, to 3, to 2. After this experiment, Helmut reported, that he could >>> fully decode the signal at 6, but at 5 only about 50% and lower no >>> decoding anymore. We estimated, that setting 6 was about 50 mWatt. >> >> >> ---------------------------------------------------------------------- >> ------ >> -- >> _______________________________________________ >> Freetel-codec2 mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/freetel-codec2 >> >> >> ---------------------------------------------------------------------- >> -------- Monitor Your Dynamic Infrastructure at Any Scale With >> Datadog! >> Get real-time metrics from all of your servers, apps and tools in one >> place. >> SourceForge users - Click here to start your Free Trial of Datadog now! >> http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140 >> _______________________________________________ >> Freetel-codec2 mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/freetel-codec2 >> > > > ------------------------------------------------------------------------------ > Monitor Your Dynamic Infrastructure at Any Scale With Datadog! > Get real-time metrics from all of your servers, apps and tools > in one place. > SourceForge users - Click here to start your Free Trial of Datadog now! > http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140 > _______________________________________________ > Freetel-codec2 mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/freetel-codec2 > ------------------------------------------------------------------------------ Monitor Your Dynamic Infrastructure at Any Scale With Datadog! Get real-time metrics from all of your servers, apps and tools in one place. SourceForge users - Click here to start your Free Trial of Datadog now! http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140 _______________________________________________ Freetel-codec2 mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/freetel-codec2
