Hello dear LF Friends,
thanks for the many and divergent replies to my post!
My Mother celebrated her 90th this weekend, so I didn't
have any time to give an answer immediately.
Now, let me quickly try some comments to the different
1. Science and HAM Radio
Thanks to Gamal for his answer to Petr's serious
question what HAM radio is !
May be, all discussions and experiments clearly say
that a forest of 100 m towers, many DSP's and a super
computer are necessary to get information on LF across
the ocean. Then, Petr, OK1FIG, please ask again.
My intention was to start a discussion on ideas that
may be practicable for radio amateurs. As long as
there is no answer we have to discuss - just as in
2. PSK vs. ASK
ASK keyes between sinus signal and no signal, PSK
keyes between sinus and minus sinus. The distance
between the two signal alternatives measured in
voltage is doubled with PSK while the noise remains
unchanged. If on the transmitting end the maximum
power is limited (not the average power) the gain
of PSK over ASK is 6 dB, i.e. to get the same SNR
with ASK four times the power of PSK is necessary.
If ASK is detected incoherently as usual a factor
of 2 (3 dB) has to be added to this difference
because more noise is received. Unfortunately with
real channels the gain is not as great (see: DPSK).
3. Long Integration Time
The ideas proposed by Andre', N4ICK, are good when
ASK or FSK is used with non coherent reception.
FSK would be the better choice.
But a synchronization to time ticks of a second is not
a problem. Following the comments of Larry, VA3LK,
and Johan, SM6KL, GPS can be used. I prefer DCF77
or MSF because of its simplicity.
In fact, if no information is sent then the result of
a very long integration time is the same as integrating
over many short time slots. The advantage of a one second
period or even faster is a better averaging of non
gaussian noise (especially in the case of ASK).
Using a random bit pattern helps to minimize man
made noise. Possibly, one should not use the second,
but 5 minutes divided by a prime number to get away
from anything that is synchronized to a clock.
If real information is sent, then the transmitted
bitrate should be considerably greater than the
information bit rate. This is achieved by heavy
coding with a low code rate. But that is not a point
of discussion as long as even the carrier cannot
Unfortunately, there comes a loss with PSK when the
communication channel may vary in delay. A delay
produces a phase shift, and a varying delay results
in a changing phase. The symbol rate must be so fast
that the change in phase from one to the next symbol
is negligible. 136 kHz is about 14 MHz/100. On 20 m
the symbol rates of PSK31 and Pactor2 (31 and 100 Hz)
are in the optimum region of 20 ... 200 Hz. I simply
guessed that 1/100 of that (1 Hz) should be good for
LF although the waves may travel quite another way.
As a consequence of the varying channel, one cannot
use a constant reference phase. One therefore detects
the difference between the last and the actual symbol.
Because now the noisy last phase is used as reference
instead of an absolute and noise free one this
differential PSK (DPSK) has a loss of up to 3 dB
over the absolute PSK. At extremely low signal level
FSK is as good as DPSK. Therefore FSK should remain
in this discussion.
As mentioned above, coherent detection gives a better
SNR. But, if the signal is so weak that the time to
transmit one single information bit (that means
identification of the carrier) is longer than the
period of approximately constant phase then the
3 dB gain of coherence are lost. In other words: If
you cannot keep track of the carrier things get worse.
Using a milliwatt on 20 m it's the same situation and
2 m aurora too, the only difference is in the time scale.
Where are we on LF? Is something known about QSB periods
on transatlantic LF? If the typical QSB-period is a few
minutes long or faster coherent detection is impossible.
I then would try FSK keyed with a random pattern at
a symbol clock of 1 per second up to 1 per minute.
6. Phased Array / Synthetic Aperture
Paul, OH3LWR, is right in pointing out the problem
of a complex radiation pattern with many "fingers".
That's the reason why I proposed a mean distance of
a quarter of the wavelength. Then nearly the maximum
gain can be achieved with only one main lobe.
There are many different possibilities for the
realization of the correct phasing of transmitting
and receiving antennas. These should carefully be
discussed. Paul's proposal to use a local BC or TV
station seems to me a practical one.
73 de Klaus, DJ5HG