"Mike Monett" wrote in message
...
Andy writes:
Jeff, I haven't tried this method, but one rule of thumb I have
always
believed in is:

"No matter how much you shift, limit, amplify or divide noise, it still
ends up as noise"

The only effective way I have ever found is to narrow the bandwidth
around the signal until the signal starts to get degraded. If done
digitally, it can be done by digital processing, but that changes only
the technique, not the principle...

Actually, digital processing CAN change the principle.

FIR filters and similar digital filters do provide a way to reduce the
bandwidth digitally, and as you point out, reducing the bandwidth reduces
the noise. This helps the same way a crystal filter helps, except perhaps
giving a little more flexibility.

However, many modern radios have digital noise reduction which is quite a
different animal. With digital noise reduction, the incoming signal is
analyzed to identify noise components and differentiate them from signal
components. The noise is then subtracted from the signal. While this isn't
perfect, it can result in quite a substantial reduction in noise without
reducing bandwidth.

The combination of bandwidth reduction and digital noise reduction can
greatly improve readability.

I agree that other typical analog techniques don't really affect things all
that much, but I'm not convinced that the same techniques that are used for
noise reduction digitally couldn't be duplicated with analog components;
I've just never seen it done, and without some considerable creativity on
the part of the designer it will be quite complex.

One analog behavior I have noticed that helps, at least with CW. For passive
balanced mixers, there is a diode threshold voltage required for the signal
to be detected. If the gain is managed so that the noise level is very
close to this threshold, the signal to noise ratio seems to be improved
(although I have not personally validated this analytically). Of course, if
the signal is at the noise level this doesn't help, and if the signal is
barely above the noise level the adjustment is too critical to be a great
help, but where the signal has enough headroom, it can pretty dramatically
improve the pleasure of listening to a weak signal.

...

For passive balanced mixers, there is a diode threshold voltage required for
the signal to be detected.
-------------------------------------------

This is news to me - can you please elaborate?

Joe
W3JDR




"xpyttl" wrote in message
...
"Mike Monett" wrote in message
...
Andy writes:
Jeff, I haven't tried this method, but one rule of thumb I have
always
believed in is:

"No matter how much you shift, limit, amplify or divide noise, it still
ends up as noise"

The only effective way I have ever found is to narrow the bandwidth
around the signal until the signal starts to get degraded. If done
digitally, it can be done by digital processing, but that changes only
the technique, not the principle...

Actually, digital processing CAN change the principle.

FIR filters and similar digital filters do provide a way to reduce the
bandwidth digitally, and as you point out, reducing the bandwidth reduces
the noise. This helps the same way a crystal filter helps, except perhaps
giving a little more flexibility.

However, many modern radios have digital noise reduction which is quite a
different animal. With digital noise reduction, the incoming signal is
analyzed to identify noise components and differentiate them from signal
components. The noise is then subtracted from the signal. While this
isn't perfect, it can result in quite a substantial reduction in noise
without reducing bandwidth.

The combination of bandwidth reduction and digital noise reduction can
greatly improve readability.

I agree that other typical analog techniques don't really affect things
all that much, but I'm not convinced that the same techniques that are
used for noise reduction digitally couldn't be duplicated with analog
components; I've just never seen it done, and without some considerable
creativity on the part of the designer it will be quite complex.

One analog behavior I have noticed that helps, at least with CW. For
passive balanced mixers, there is a diode threshold voltage required for
the signal to be detected. If the gain is managed so that the noise level
is very close to this threshold, the signal to noise ratio seems to be
improved (although I have not personally validated this analytically). Of
course, if the signal is at the noise level this doesn't help, and if the
signal is barely above the noise level the adjustment is too critical to
be a great help, but where the signal has enough headroom, it can pretty
dramatically improve the pleasure of listening to a weak signal.

..