LF Bandplan / PSK31]

Andre' Kesteloot akestelo@bellatlantic.net
Thu, 18 Feb 1999 08:43:09 -0500



drassew2@interalpha.co.uk wrote:

> This question was posed to me by Rik Strobe, but I feel the reply will be of
> value to all recipients of this reflector.
>
> >but the TX part might be more
> >complicated as my homebrew 136kHz TX is using switched FET'sn (modificated
> >version of the G3YXM design).
> >As far as I understand PSK the linear amplification if just needed to
> >minize the 'clicks' caused by the abrupt 180 degrees phaseshift.
> >If one would use PSK with a non-linear amplifier there would be 2 main
> >disadvantages :
> >A possible solution (to use PSK with non-libear amplifiers, minimizing
> >'splatter') that crossed my mind was to do the keying exactly at the
> >zero-crossing of the RF signal. This would avoid the voltage jumps in the
> >RF signal caused by the 'normal' PSK.
>
> Unfortunately this is not a valid argument, switching at the zero crossing
> points will make no differance in practice.  The only thing you will have
> eliminated is the actual RF switching transient which could be MHz (!!) wide
> and are not themselves a function of carrier frequency or baud rate.
>
> The real issue is the pattern of sidebands generated due to the harmonic
> nature of the, now squarewave, modulating signal.  A squarewave consists of
> a whole series of odd harmonics at levels of 1/3 , 1/5 , 1/7 ....   1/N.
> This is a mathematical derivation and is a fundamental fact of life.   These
> are voltage levels so the third harmonic is only about 10 dB down and
> pro-rata. Some intermediate shape of pulse (eg filtered squarewave) will
> have an intermediate spectrum.  PSK31 uses a raised cosine pulse which in
> the case of 101010 reversals results in a pure sinewave at half the baud
> rate - you can't get narrower than that situation!
>
> This data spectrum is superimposed on the RF carrier and spreads out equally
> either side of it giving a PSK31 bandwidth equal to the baudrate.  For
> squarewave pulses the spectrum is only 19dB down at 18 times the baudrate
> (plus / minus ninth harmonic of the data)
>
> In fact last night I was listening to a QSO on 3.5 MHz between two DL
> stations who were both badly overdriving their transmitters - probably by
> connecting the Soundblaster output directly to the microphone socket).
> While copy was perfect, the splatter was terrible (500 Hz wide) and made the
> PSK31 subsegment almost unuseable.   As the QSO was in German and I don't
> know how to say "You are seriously overdriving your transmitter" in that
> language, I didn't reply!
>
> Andy  G4JNT