Dear Jim, LF group,
There is a variant of PSK31 that incorporates a Viterbi FEC encoder(see
link:- http://www.aintel.bi.ehu.es/description )Using this might mitigate
the effect of impulse-type QRN mentioned below. This particular
implementation might not be ideal for LF use, as it uses QPSK modulation to
regain the 50 WPM throughput rate, and QPSK has it's own foibles and
drawbacks.
From reading the original G3PLX article, Peter seems to have introduced QPSK
to get back the throughput lost by using half-rate coding, but perhaps a
BPSK/Viterbi implementation delivering 25WPM would be acceptable for us LF
types....it's still faster than I can type! Trouble is, I don't think one
exists.....
Cheers
Dave, G3WCB
-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of james moritz
Sent: 23 November 2004 15:15
To: [email protected]
Subject: LF: RE:
Dear Mike, LF Group,
The amplifier does not have to be very linear for PSK modes - just so long
as you have some sort of ramp up and down to zero amplitude, rather than an
abrupt on-off transition, the spectrum of the signal will be reasonably
narrow. When I used a linear, I found the best bet was to bias the MOSFETs
so they were slightly cut-off, and then over-drive them a bit - more or less
class C operation. This improved the efficiency somewhat, but it was still
only about 60% at best.
For PSK generation at the moment, I amplitude-modulate the HT supply to the
Decca class D PA, whilst separately phase modulating the drive carrier. This
is much the same as the "envelope elimination and restoration" techniques
which now seem to be standard for big AM broadcast transmitters. I keep
meaning to write this up, but it is the usual problem of having too many
other things to do!
The advantages of PSK31 are that there is readily available software, the
bandwidth is narrow enough, and people are familiar with it - but it isn't
particularly well suited to LF operation; it works fine when the band is
very quiet, but QRN impulses corrupt characters each time they occur, so the
band does not have to be very noisy to render the mode unusable except for
stations very close together. I think there is a case for a similar mode to
PSK31, but incorporating error correction, so that a moderate level of QRN
corrupting a small proportion of bits does not cause a big problem. VE2IQ's
"Coherent" mode did this quite well, but requires MS-DOS and dedicated
interface hardware, making it less attractive.
Cheers, Jim Moritz
73 de M0BMU
-----Original Message-----
|
