Dear LF Group,
Re - the discussion of GPS derrived timing, DDS frequency
sources, etc. It seems to me that increasing layers of complexity
being proposed could do with some review before rushing headlong
at them.
There would appear to be little point in increasing the tuning
resolution of signal sources much beyond the bandwidths occupied
by the signals. So, for the very slow QRSS and similar modes, the
bandwidth can't be much less than perhaps 10 millihertz because
the propagation "lifts" do not last long enough to transmit a usable
amount of data. The recent trans-atlantic tests seem to indicate
that being able to vary frequency in 0.1Hz steps is adequate for
dodging the QRM. A mode like BPSK requires absolute frequency
errors to be much smaller than it's bandwidth, but again there would
seem to be little point in being able to change the transmit
frequency in very small increments, since this would just make it
harder for the receiving station to find the correct frequency.
Something that has become more noticeable lately is the
limitations of frequency stability in current equipment. Many people
currently are getting away with simple crystal oscillators, which
certainly work OK for most things, but are probably now at their
limits for stability for further development. So designs using DDS or
otherwise obviously need to consider how a stable frequency
reference can be incorporated from the outset, if high resolution is
to be worth bothering with. If your DDS has a clock frequency of
25MHz, it will tune in different steps to one with a clock at 30MHz,
and so there will always be a certain amount of error, even if
perfectly stable. Also, thought has to be given as to how to use
better reference oscillators with all the other equipment that is used
for transmit and receive, such as amateur band tranceivers, DSP
devices and so on. In principle, it is easy and most satisfactory to
use a single reference oscillator for all the equipment that requires
good frequency accuracy - but in practice, that is quite difficult to
implement with most of the equipment currently in use.
It seems to me that once a high stability frequency reference has
been obtained, a clock of quite impressive accuracy is also
available. A reference of 0.1ppm stability would make a clock that
gains or loses 1ms maximum in 2.8hours, which would probably be
small compared to other errors - the only slightly tricky thing is
setting it. But since you have to cope with a somewhat variable
offset for the time delay between two stations, this does not seem
to be a big deal.
Relying on the "compatibility" of equipment seems to be a bad idea
- witness the problems running certain software with certain
soundcards, and niggling frequency errors. I suppose one way
round this would be for someone to by up a large batch of identical
sound cards for distribution around the LF community, but does
anyone really want to do this, and anyway, what would happen in 6
months time when the stock ran out, or for some reason that
soundcard was found unsuitable or obsolete? Similar things can
probably be said of GPS receivers. In any case, building up a
system that relies on everybody using identical equipment to do
the same thing seems to take all the fun out of it.
I await new ideas with interest!
Cheers, Jim Moritz
73 de M0BMU
|
