Varying the drive duty cycle to change the RF level does not give the result
expected at first sight. While the DC level does vary proportional to the
duty cycle, the RF carrier, which is revealed via the second term of the
Fourier expansion of a rectangular waveform does not. Instead, there is a
SIN(X) / X component in the waveform which gives a horribly non linear
relationship between duty cycle and RF out. It may be possible to
pre-distort the drive waveform to work over a limited range of amplitudes,
but I hate to think what the lookup table would contain !
The only real way to alter the amplitude is to change the power supply to
the PA stage and a PWM modulator, ie. a switch mode PSU working open loop is
not too complicated a beast to make if it really is necessary to adjust
amplitude.
But is it really worthwhile going to these lengths just to operate PSK31.
There are several other schemes already in use with equally good signalling
capabiliies that work with constant amplitude - ie JASON, Coherent (by
VE2IQ) and a host of externally synchronised coherent PSK schemes as yet
barely tried. No one has looked at MSK yet either. That is constant
anmmplitude, and with a bit of imagination could be generated from a DDS as
well as from the ubiquitous soundcard.
Andy G4JNT
-----Original Message-----
From: Johan Bodin [mailto:[email protected]]
Sent: 28 January 2002 09:11
To: [email protected]
Subject: Re: LF: Envelope shaping and class D
John wrote:
Interesting idea, Johan, but I believe you'll find that if you drive the
amplifiers with a sinewave, there is an efficiency penalty; whereas, if
you
drive them with a square wave, you actually have made a PWM modulator.
Yes and yes.
I meant square wave drive (class D).
My idea was to combine the modulator and the P.A. in the same stage.
This would eliminate the need for a separate PWM modulator in the
P.A. power supply.
The switching waveform will look a bit "spikey" at low output power. Worst
case
is when the output power is close to zero, the signal will look like a train
of impulses
with alternating polarity. I guess the spectrum will then contain all odd
harmonics
at the same amplitude as the fundamental, limited only by the switching
times.
Assuming a Q of 6 in the series resonant output circuit (typical value, I
think), this
circuit alone would attenuate the 3f component by 24dB. Not too bad.
73
Johan SM6LKM
--
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution,
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.
|
