[Fwd: LF: PSK31 tests.]

[Andre Kesteloot]

Peter Martinez wrote:

> >From Peter Martinez G3PLX Kendal Cumbria.
>
> Today (Saturday 7 March) Andy G4JNT transmitted some 31 baud
> PSK tests, which I successfully received here. Both Andy and I
> were using my PSK31 software for the 56002EVM kit.
>
> Andy's signal at the 100 watt transmitter level (estimated 7dBm
> erp) was about 549 on CW. Here is the first test message I received.
> My mailer will probably screw up the original line-formatting
> =
> G3PLX de G4JNT
> This is the first test transmission using PSK31 mode on the 137kHz
> band
> Time 1052z date 07/03/98
> g3plx de g4jnt
>
> This test is using the PSK31 mode.   Path 393km from G4JNT
> Near Southampton to G3PLX in Cumbria.  Transmit power 100
> Watts to 7m high Tee antenna.  ERP estimated at 7dBm.
> Trans*itter driver is an SSB exciter using local oscillators derived
> from the RA1792 receiver to ensure absolute frequency accuracy
> and netting.
> =
> There is one error in this message, representing a character
> error-rate of 0.3% at a traffic speed of about 50wpm.
>
> Andy then changed between BPSK (no error-correction) and QPSK
> (error corrected) at the same traffic speed, and repeated the tests at
> 50w, 25 watt and 12 watt transmitter powers, representing erp's of 4,
> 1, and -2dBm respectively. These were the results:
>
> Power   BPSK    QPSK
> 100     0.3%    1.7%
> 50      3.4%    4.9%
> 25      40%     60%
> 12      100%    100%
>
> These tests are far too small a sample to be statistically useful, but
> the fact that the BPSK (without correction) is better than QPSK
> (with error correction), ties in with the predicted theory for a link that
> is dominated by random noise rather than bursts. Both BPSK and
> QPSK use the same bandwidth and transmission rate. This doesn't
> mean that the error-correction could be improved but confirms my
> findings that when the noise is random, it pays to transmit the raw
> data in the narrowest possible bandwidth, rather than add extra
> redundancy and either have to widen the bandwidth or slow down the
> traffic speed.  On HF where PSK31 is used a lot, the QPSK mode is
> very effective, and this might still be the case on 136 kHz in
> conditions where there is a lot of atmospheric noise.
>
> This is a promising result. Next I need to get my erp up to match
> Andy's level, and then we can try for a two-way QSO.
>
> 73
> Peter