Hi Bruce, Good idea to bring this topic back on-list.
I have a couple of other comments: 1/ The current PAPR is for a canned test sequence of bits that repeats itself every 80ms, not sure what it will be when the modem is driven by codec data. 2/ FDMDV is designed for SSB radios, I don't think I'd put it thru a FM HT. For that we would use a constant envelope type waveform modem. Remember we only use FDM on HF because of the channel. At VHF it's probably OK to use a single tone (single carrier) PSK modem. It's fairly easy to get a low PAPR on a single tone PSK modem. 3/ Even for FDMDV the current PAPR can be improved, I played around with that a few months ago and have some more ideas. 802.11 Wifi uses parallel tones (OFDM) with many more carriers than FDMDV and they use efficient PAs OK. 4/ The Octave fdmdv_ut.m simulation can be used to test the effect of a given PA on BER, but could benefit by a PA decent model (like a quadratic). My current model is crude hard limiter. Cheers, David On Thu, 2012-05-10 at 18:15 -0700, Bruce Perens wrote: > David, Mel, and I have been having a discussion of PAPR: peak vs. > average power ratio, off of the list. This is also called "crest > factor". > > When multiple carriers in our FDM modem are close to a peak of their > waveform simultaneously, we can get a momentary output power that is a > large multiple of the average. For our FDMDV-derived modem, the ratio > is 13.5 dB. > > We don't actually operate our SSB amplifiers in a way that avoids > clipping, because we can tolerate some clipping. Clipping increases > the bit error rate but increasing power reduces the bit error rate > generally. In the 925 mile test, the peak power spikes that should be > visible aren't, they're either buried in noise or were clipped, and > the bit error rate is still zero for much of the transmission. It > sounds to me like the amplifier was operated with about 6 dB headroom, > but we don't have a measurement and there is much unknown about how > the amplifier clips. > > While PAPR is managible on an SSB radio. It is potentially a problem > if we want to use the FDMDV modem on an HT. We need power efficiency > on handhelds, which means operating the power amplifier very close to > full power at all times. We can't afford to have lots of headroom > there. > > So, I have some serious questions about how well our FDMDV modem would > work on an HT, and whether some alternative is required. We could use > some more science on this topic, and unfortunately I'm not > sufficiently clueful to provide it. Sorry. > * How much clipping can we tolerate, and how does BER diminish > with clipping? > * How does spurious emission increase as we drive the PA into > clipping? > * Should we compress in software before feeding the amplifier, > so that clipping is more graceful than the hardware might make > it? > * What am I missing? > > Thanks > > > Bruce > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ Freetel-codec2 mailing list > Freetel-codec2@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/freetel-codec2 ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freetel-codec2
