<!DOCTYPE html>
<html data-lt-installed="true">
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body style="padding-bottom: 1px;">
<p>Hi folks,</p>
<p>I just had an interesting question come from a ham in Greece who
is developing a noise reduction device that works in front of the
receiver. He asked about my noise source and wanted some advice
about how he could make his circuit better... That started my
wheels turning and since we've recently discussed how local power
line and other noise makes HF work "less fun" in our area I
thought I would share the general idea for m2b (Minimum Two
Brains) with the group:</p>
<p>In some audio mixing situations it is desirable to [ab]use a
compressor for the purpose of ducking one signal in favor of
antoher. In particular what comes to mind is "ducking" the bass
guitar or keyboards in order to make room for the kick drum. This
way they can all be at high levels without clipping. The kick drum
pulse is short enough that nobody really notices that the bass
guitar or keyboards were "ducked" -- psychoacoustically
transparent.<br>
</p>
<p>What if we apply this idea to a noise blanker.</p>
<p>A noise blanker wants to detect and eliminate broadband noise
pulses without affecting the signal(s) of interest.</p>
<p>Good news is that signals of interest tend to be somewhat
sinusoidal and typically at a much lower amplitude than broadband
noise. Broadband noise, viewed in the time domain, has a tendency
to resemble very narrow high amplitude pulses.</p>
<p>In this case, we want to do the opposite to what we would do in
the mixing example.... In a way, we want to squash the kick drum
out of the mix.<br>
</p>
<p>Picture if you will:</p>
<p>A broad band differential amplifier with a transformer input and
output.</p>
<p>The long tail is a current sink controlled by a voltage.</p>
<p>This is essentially a high fidelity gain controlled amplifier
that passes through the input from the antenna to the receiver.</p>
<p>Now sample the incoming signal in order to detect narrow high
amplitude pulses.</p>
<p>Even better,... combine a high and low pass filter to select
pulses that occur outside of the passband of the signal of
interest.</p>
<p>Amplify the sampled signal so that there is enough gain for it to
be compared with some useful threshold using a high speed
comparator. Take two, they're small and cheap.</p>
<p>Set the gain of your variable gain amplifier to some useful level
-- even unity if you like... or +/- a few db here or there because
the actual gain of this amplifier is not really the point... the
fact that the gain is adjustable at high speed is what's
important.</p>
<p>Set up one comparator to look for positive going pulses that
cross your threshold. Set up the other comparator to look for
negative going puleses.</p>
<p>Connect the open collector outputs of these together.</p>
<p>Then connect the comparators outputs to the gain control voltage
so that they dramatically (and tunably) reduce the gain of the
signal pass amplifier when such a pulse is detected... for as long
as that pulse is above the given threshold.</p>
<p>Note what happens to the output signal whenver the pulse
detection signal crosses the threshold:<br>
<br>
Picture the pulse of a sin wave rizing above zero and back down.</p>
<p>Draw a horizontal line about 2/3 of the way up to represent your
threshold.</p>
<p>Draw in inverted square wave that goes down to zero starting at
where your threshold intersects your sin wave pulse.</p>
<p>Erase everyting outside of that negative going square wave...
what do you see:</p>
<p>The unwanted pulse has been clamped in amplitude at the
threshold.</p>
<p>AND what's left of it are very narrow fragments of the original
unwanted pulse.</p>
<p>What have we done? Ghasp!!</p>
<p>Well, like all similar noise blanking circuits we have introduced
distortion... BUT we have also restricted it's ampitude and pushed
it's frequency far above the passband near the signal of
interest!!</p>
<p>AND we have done this before the receiver gets hold of the
signal-- so if the noise were enough to overload the front end
before it is not now (meaning that you may not need to use any
attenuation in the receiver.</p>
<p>M2B? <br>
</p>
<p>Just another crazy idea from your friendly neighborhood
madscientist,</p>
<p>_M<br>
</p>
<pre class="moz-signature" cols="72">--
kf4hcw
Pete McNeil
lifeatwarp9.com/kf4hcw</pre>
</body>
<lt-container></lt-container>
</html>