Dear LF Group,
Many thanks to DL4YHF for the speedy software update; I hope to do some
tests on PSK31 some time in the next few days.
About using BPSK modulation with the Decca TX:-
The phase keying can easily be done using an EXOR gate in my case, or a
system such as G0MRF/SM6LKM's DDS board. The problem here is of course that
the abrupt phase transitions with a constant-amplitude output from a class
D PA like the Decca cause horrendous key clicks, which when running ERP up
to 1W would cause serious QRM over several hundred Hz of the band - I tried
it with a low-level signal; it really is bad! To get round this, the TX is
also amplitude modulated with a half-sinusoidal envelope which ramps the
output down to zero before the phase transition occurs, then ramps it back
up again, which gives a well defined bandwidth. The modulator is basically
a big series linear regulator that controls the supply voltage to the PA.
This sounds like a headache from the point of view of efficiency, but
actually the regulator only dissipates about 150W with 1.2kW PEP BPSK
output, and the overall efficiency is still around 80%.
The modulation envelope is generated by a circuit that looks for
transitions in the PSK data, and when one occurs it generates a piecewise
approximation to the sine envelope, which is filtered and used by the
modulator/regulator as a reference voltage. The phase data is also delayed
by half a symbol period before going into the EXOR modulator, so the
amplitude has time to ramp down to zero before the phase transition occurs.
The result is a very clean BPSK signal that I used successfully for
COHERENT and WOLF mode tests last winter.
The envelope generator is a rather long-winded thing with about a dozen
logic and linear IC's, but could easily be made using a PIC with D/A
converter - or DL4YHF's suggestion of generating this on the sound card
could be used. The Modulator/Regulator has 8 MOSFETs on a big heatsink - I
also use it for CW keying, and it provides PSU overload protection too.
About other methods of generating BPSK:-
I tried some experiments with a scheme similar to that suggested by Johan -
The problem with this is that, while the square wave output contains
essentially no even harmonics, as soon as you change the duty cycle, or the
relative phase of the two totem pole outputs, a strong second harmonic
component appears. This is much more difficult to filter out of the TX
output than the third and higher harmonics, requiring excessive high Q in
the tank circuit and/or a low-pass filter with a very sharp and accurate
cut-off. I think it is possible to use PWM to achieve output control, but
the output waveform must be kept symmetrical.
You could of course use a separate PWM regulator to control the PA supply,
but this would be quite a difficult project at the 1.2kW level, with
significant noise and intermodulation problems possible, and the overall TX
efficiency would not be that much better than the linear modulator. I think
G4JNT has a PWM circuit in the LF Handbook.
Another approach is to use "variable phase" modulation, ie. ramping the
phase slowly from 0 to 180 instead of an abrupt transition. This does
certainly work, although I'm not sure if it would impair the performance of
the signal under weak signal conditions. You can see my experimental
variable phase modulator circuit on G3YXM's "features" web pages.
There is also a linear modulator circuit on 'YXM's site - this works
perfectly well, but obviously requires a linear PA.
The conventional approach to generating BPSK is to use a DSP device such as
a PC and sound card to generate the modulated signal with an audio carrier
frequency, and then use an SSB TX to linearly translate this to 136k or
wherever. However, poor sound card sampling rate accuracy caused a lot of
problems with WOLF, and getting rid of the sound card altogether was a big
improvement. Also, I didn't have a suitable SSB TX. The same undoubtedly
applies to other "slow" digital modes, such as Andy's coherent PSK
experiments. However, the timing requirements for PSK31 are much less
critical, and so the sound card should be perfectly OK for this.
Cheers, Jim Moritz
73 de M0BMU
|
