Content-Transfer-Encoding: 8bit
Message-ID: <65AECDF1F89AD411900400508BFC869F0D746E@pdw-mail-1.dera.gov.uk>
From: "Talbot Andrew" <ACTALBOT@dera.gov.uk>
To: rsgb_lf_group@blacksheep.org
Subject: LF: CRUNCH method
Date: Fri, 8 Dec 2000 11:00:08 -0000
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>

Bill et al--
Re your CRUNCH algorithm:
Why not downconvert to zero frequency and preserve I/Q pairs in the
positive and negative frequency domains, rather than a 'low' frequency
that determines the audio tone and could be subject to alliassing
effects.  Then you only need to low pass the resulting samples at half
the wanted bandwidth to effect an input bandpass function.  Decimation
to lower the sampling rate is straightforward if wanted, just by taking
one sample out of every N.  Upconversion to any other frequency is then
performed in an SSB upconverter by multiplying I/Q pairs with SIN / COS
to just give the one sideband.

Regard the zero IF samples as being speeded up by nominating (labelling)
the value of the sample rate accordingly, multiply by SIN / COS pairs of
the centre frequency you want to listen at, sampled at this new sampling
rate, and store as a .WAV file.

A similar technique I wrote for off air signal analysis records a chunk
of spectrum filtered in the receiver to 16kHz, sampled at 44100Hz to an
input .WAV file. By selecting centre frequencies and signal bandwidths
of interest, then downconverting / decimating as above to decode
multiple channels simultaneously.  The result is a software receiver
than can decode (not in real time of course) arbitrarily narrow channel
bandwidths at any suitable sampling rate at sub-multiples of 44100 Hz
which can then be passed on to demodulators etc.

Andy  G4JNT


> -----Original Message-----
> From: Bill de Carle [mailto:bill1@cgocable.ca]
> Sent: 2000-12-07 16:54
> To: rsgb_lf_group@blacksheep.org
> Subject: Re: LF: VE2IQ's .WAV file /73kHz
> > around the frequency of interest.  The basic process I use is to
*mix*
> the signal down to some very low frequency (in this case 800 Hz was
> mixed down to 20 Hz) - then to *multiply* it back up to 800 Hz.  The
> frequency multiplication process is what makes the desired signal seem
> to stand out all by itself.  For example, if there was an interfering
> carrier 10 Hz away from LEK's signal, after mixing down it would still
> be 10 Hz away, but after multiplying back up to 800 Hz it 
> would be 400 Hz
> away and then gets taken out by the 300 Hz bandpass filter on the
> output.  I could have used a much *narrower* bandpass filter on the
> output in this particular case because I know Lyle's station is always
> right on the published frequency, but in general I keep it 
> wide because
> lowfer beacons are often several Hz off frequency and they risk going
> outside the passband (after multiplication) if the filter is 
> too narrow.
> 
> Bill VE2IQ
> 
> 


-- 
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.