Content-Transfer-Encoding: 8bit
Message-ID: <3617AC3245C2D1118A840000F805359C01752898@pdw-mercury-1.dera.gov.uk>
From: "Talbot Andrew" <ACTALBOT@dera.gov.uk>
To: rsgb_lf_group@blacksheep.org
Subject: RE: LF: Transatlantic  beacon tests
Date: Wed, 26 Apr 2000 16:36:15 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

I think most modern transceivers derive all their internal frequencies
from one master oscillator.   The IC746 uses a 30 MHz device which at
extra cost can be replaced by a small 1-2ppm ovenned oscillator.  This
has excellent stability in practice, and after about 30 minutes warm up
stays within 0.1ppm if room temperature does not change violently.
Exact frequency can be adjusted from a preset on the rear panel.   I
haven't done it yet, but this could be easily phase locked to a
frequency standard.   

Alternatively, the conversion LO for an LF transverter can be derived
from this internal signal, thus cancelling out drift at the HF
intermediate freq.  Net drift is then the 1ppm (or whatever) but now
only at the LF frequency.  Getting respectable......  However, in
practice for LF I use a RA1792 Rx.  Again all freqs are derived from one
master - an external 5 MHz source with 10^-10 stability.

For the narrow bandwidth Rx, if you want to start from scratch what
about this as an outline scheme ?

Start off with a DDS derived LO locked to a standard oscillator, and
downconvert the LF band directly to 32765 Hz.  Make up a ladder filter
using watch crystals - I have a first cut design for a five crystal
device with a bandwidth of 1.1 Hz.  Subsequently convert the output down
to 1 - 2 Hz  'baseband', again with a locked LO.   A/D convert directly
at a sampling rate of, say, 10 Hz to 16 bit accuracy, again with the
sampling clock locked to the frequency standard and feed the output to a
PC via the serial port or even the parallel port.   At this data rate
soundcards are irrelevant and any 16 bit language running in DOS can do
some very advanced signal processing.  
If I hadn't got so many other projects and a perfectly good Rx I'd have
a go myself !

Andy G4JNT


>From Rik ....
>And indeed, I could see the 0.5Hz 'jump', but strange enough after some
>time (with the window open) the frequenc slowly started to return to
the
>original value. First I was fussed by this effect, but then I
understood
>what was happening : the VFO of my TS440 also 'took a cold' (due to the
>larger mass and better shielding it happened slower). The second effect
was
>considerable smaller than the initial 0.5Hz drift

>From Larry .........
>What is needed is a
>suitable very narrow bandwidth receiver design to go with the DDS
effort we
>are making over here with the AD9835 DDS device.


-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.