[Elecraft] Subject: "Another Better Mouse-trap?"

[Ron D'Eau Claire]

Note that the writer, in calling for better noise blankers, says he doesn't
need noise blankers! Still, they are useful at certain times and the
Elecraft blankers do a very good job for me, making high-impulse repetitive
noise virtually disappear. Examples of such noises are the buzz from cheap
lamp dimmers on the lower bands to the pop-pop of a spark ignition system in
an internal combustion engine on the higher bands. I don't hear many
ignition systems these days, but I do have a cheap lamp dimmer issue on 160
and 80 meters. The Elecraft blankers take a S9+++ dimmer roar down below
S-2.   

But all blankers have serious problems, too, so careful operators turn them
off unless they're really needed. 

The weakness in blankers is that punching holes in the signal causes mixing
products. Indeed all "mixers" work pretty much that way, using a strong
signal to interrupt a second, often weaker, signal. The more aggressively
you interrupt (punch holes) in one signal with another to blank out noise
pulses, the more mixing products you create. If you really "go at it" to
punch a hole for every blip of noise that arrives, the "mixer" will create
even more noise than it's suppressing. In addition to wide-band hash caused
by punching holes in the noise, interrupting the signal itself will create
IMD products in the bandpass. 

Two receivers aren't needed for this nowadays although they were often used
years ago. Nowadays it's rather easy to delay the signal slightly to give
the logic system time to identify a qualifying noise pulse and interrupt the
signal path at just the right time to "punch a hole" in it, suppressing the
spike of noise. 

There are also techniques in which noise is received on a separate
frequency, inverted and added to the main signal path. The idea is that they
add - one positive and its counterpart negative - and, hopefully, cancel the
noise, leaving the signal unaffected. 

That presumes noise can be found that is identical to the noise in the
signal bandpass. I've read of various techniques, but the weakness seems to
be that the noise is modified by the receiver (filters make noise impulses
wider, frequency differences mean the noise may have different phase
characteristics, etc.) so that the two noise signals aren't identical and so
do not really cancel. One approach to that problem is to have two antennas
on the same receiver on the same frequency, trying to arrange one antenna to
receive little signal but receive the noise while the other is arranged to
receive the signal was well as possible. The signals are then "added" right
at the receiver input. 

I've not worked with that sort of noise suppressor. There have been many
articles published over the years reporting that in some situations useful
reductions can be had, but it usually involves lots of tedious and careful
adjustments that must be reset whenever the receiver frequency changes. 

AFAIK, DSP-oriented noise reducers do not "blank" noise, but try to emulate
the human brain to discriminate between a coherent signal (tone or speech)
and noise. Like computers in general, they are truly in a very early stage
of development compared to what the human brain can do with practice,
especially on CW. They are pretty good on SSB, making several dB improvement
in the desired signal-to-noise level in most cases, but distorting the voice
in the process. Overall, I've not heard a DSP "noise reducer" that was as
nice sounding, adaptable or as effective as my "gray matter".  

Don makes a good point about AGC. Since AGC tries to maintain a constant
audio level which makes the noise sound worse than it really is by cranking
up the receiver gain automatically when the signal is not present. That's
what AGC is supposed to do. Turning the AGC off leaves the band sounding
like it really is without the noise being raised artificially to the signal
level whenever the signal is not present. AGC systems typically offer some
delay so the gain doesn't jump instantly, but that's only a partial
solution. That's why all good receivers (including the Elecraft K3, K2,
etc.) have a way to turn the AGC off. Even easier on the Elecraft rigs is to
simply control the audio level with the RF Gain control instead of the Audio
Gain control. Having the RF gain backed down limits the AGC action. (That
doesn't work on receivers with a true "RF Gain" control adjusting the signal
at the input. Elecraft receivers, like many modern receivers, actually
control the I.F. gain with the "RF Gain" control.)

AGC was invented back in the 1920's for use on AM broadcast radios for the
general public to make them more "user friendly" when tuning across stations
with widely varying strengths and to level out most QSB. It's useful on Ham
receivers in the same way when casually tuning across the band if we don't
mind the racket of the band noise between stations, but many of us never use
AGC for "serious" operating. Instead we adjust the "RF Gain" as needed as we
tune across the band. That alone gets rid of a lot of the noise. 
 
Ron AC7AC

 

-----Original Message-----

Jim NN6EE wrote:

This idea was put forth by "Don-W6JL" (Fallbrook, CA)!!!

His quote: "  I have tested many "noise blankers" in modern rigs, on the
air.  Both the I.F. noise 
gates and the DSP-based "noise reducers".  Not one works the way I feel it
should.  I think you 
need a separate noise receiver, independent of the main receiver, and tuned
to a frequency that is 
outside of a ham band.  (There have been excellent homebrew receivers using
this approach for 
noise blanking).  This then drives the noise gate, and is timed so that it
gates the I.F before 
the noise pulse arrives there.  A good blanker should have 40 dB minimum of
noise pulse 
suppression, with adjustable blanking level and widths.  I notice the K3 has
these features, but 
it still does not seem to reduce impulse noise more than an S unit or two
(6-12 dB), which is 
inadequate.  I do not use blankers here myself; my entire station is
homebrew and the receiver is 
a phasing, image cancelling job with no blanking at all, (or AGC for that
matter).  I get along 
fine with neither, even in contests."