[Elecraft] [K3] Manual Notch Reconsidered

[Ian White]

For both manual and auto notch, the key seems to be to create two
identical streams of signal data and apply different DSP to each. Stream
1 is totally focused on extracting accurate information about the
interfering tones (frequency, amplitude, phase) regardless of any damage
it may do to the audio signal. The damaged audio signal from Stream 1 is
then thrown away. Instead, the tone information extracted from Stream 1
is applied very carefully to a 'clean copy' of the same data in Stream 2
to cancel the interference with minimal damage to the recovered audio.



73 from Ian GM3SEK


>-----Original Message-----
>From: Elecraft [mailto:elecraft-bounces at mailman.qth.net] On Behalf Of
>Jim Brown
>Sent: 01 August 2014 09:11
>To: elecraft at mailman.qth.net
>Subject: Re: [Elecraft] [K3] Manual Notch Reconsidered
>
>I have no clue what the K3 does, but the slickest way to do this is to
>first sense the frequency of the tone you want to notch, then create a
>tone of the same frequency, shift its phase so that it is exactly 180
>degrees out of phase with the interfering tone, make it equal to the
>interfering tone's amplitude, and add it to the signal. That will
cancel
>the tone with no other effect on the signal. That ain't easy, because
it
>must track the drift of both the TX and the RX and the varying strength
>of the interfering signal, but with DSP, it IS possible.  The beauty of
>this technique is that it has NO effect on the desired signal -- it
>simply cancels the interfering carrier.
>
>The alternative technique is to add a narrow band notch filter, which
>MUST add phase shift that distorts the audio. The deeper the notch and
>the narrower the filter, the greater the phase shift, and thus greater
>distortion of the signal.
>
>These are fundamental concepts -- as my friends back home used to say,
>"you cain't get no better" than this, :)   and you must be very good
>(and have the processor cycles available) to do it well.
>
>73, Jim K9YC