...The transmitter is VERY interesting. Only a few years old, it is a
broadband design, and uses a Class D H-bridge FET transmitter, each 500W
module with its own Class S modulator / power supply adjustable 0 - 130V at
7A. The modulator has 0 - 4kHz bandwidth, so is AM capable. Anyone for Kahn
EER? ...
Dear Murray, LF Group,
Yes, I have been using an envelope elimination and restoration system to
transmit various BPSK modes for some years now. The PA is one of the "Decca"
1.2kW class D units re-tuned for 136kHz. The HT supply to this is 0-60V,
0-22A supplied via a modulator that is basically a big series regulator.
Using a linear modulator sounds like a bad idea, but with BPSK the crest
factor of the signal is not very great, and most of the time the PA is
running near full output, so consequently only a modest amount of power is
dissipated in the modulator. The overall efficiency is still around 80%.
Andy, G4JNT, has published a switch-mode modulator design for a class D
amplifier.
Originally, I used it with VE2IQ's "Coherent" mode, which generates straight
binary data via the serial port. This was used to key a hardware phase
modulator, and, via a waveform-shaping circuit, to generate the envelope for
the amplitude modulator. I have also used it with KK7KA's "Wolf" mode, which
was the first successful transatlantic 136kHz data mode signal. Also I have
used it to generate PSK31 and a slowed-down "PSK08" version, using DL4YHF's
Spectrum Lab software to generate an on-off keyed tone to provide the binary
data. At the moment, I am considering modifying it to obtain envelope and
phase signals from a sound-card generated signal, which should make it more
easily compatible with most data-mode software now around.
The main problem with the more exotic modes on LF is very few stations are
able to transmit them, and also receiving them is more complicated. The
"visual" modes like QRSS and DFCW are simple to generate, and several
signals can be received at once, without precise frequency calibration, etc.
PSK31 has been tried by several LF stations; it works OK when there is
little QRN, but because there is no error correction, each impulse of static
has a 50% chance of corrupting 1 bit, so it rapidly becomes unreadable when
the QRN builds up, even though the signal may remain quite audible. Since
the European band is only 2.1kHz wide, and some of that is occupied by
utility signals, SSB and the wider data modes are not very practical here!
Cheers, Jim Moritz
73 de M0BMU
