A few questions Merle:
1. If the starting point is using the DSP's
100 Hz bandwidth, will enabling NR improve S/N?
I don't see that with 1.373b5 and Spectran.
2. If enabling NR makes no improvement in S/N,
why bother since NR takes more time to respond
than directly setting DSP BW to 100 Hz?
Still looking for that "DSP Magic". So
far only Icom's advertising agency seems to
know where it is hidden.
73, Bill W4ZV
........................................................................................................
Hi Bill,
1. No, if you start with a 100Hz bandwidth you have the narrowest bandpass that
the Orion can create(Check my comments on this in the earlier post). The
narrowest filter I have seen NR build, on a S3 CW signal, was about 500Hz. That
really was the reason for my comments about being able to do better on S+N/N
then the NR code for CW by manually adjusting the BW. In terms of the NR
algorithm, it is "starved for noise" when you start with the 100Hz bandpass in
the Orion and the LMS error signal goes immediately to zero. Part of the issue
of using various parts of the Orion "noise tool kit (Notch filters, NR, PBT,
BW,Sidetone Freq, NB(SW), AGC settings and NB(HW)) is making the signal easier
for an individual to copy. I'm pretty sure that different people "hear better"
with different combinations of these settings. I know from seeing others put
their settings on the list that I often use different settings. For instance, I
always set the AGC to its lowest threshold except for AGC slow - which I use
for easy SSB QSO's. I personally find the sound of the CW signal at BW=100Hz
more difficult for me to copy then with a wider bandwidth. I guess my "brain
computer" rationalizes the additional noise better then the "mushy" - my term,
not science - signal I hear with the BW at 100Hz. So sometimes I find that a
wider BW with NR gives my brain a signal that it can decode better then an
extremely narrow BW. I find it depends on the signal and the noise - including
interferring signals(See my comments in the earlier post). However, I do think
the V1.373b5 NR provides a tool set that really can improve the S+N/N on both
CW and SSB - depending on where you start with the Orion BW setting. I suspect
there are other settings of other noise tools that I use differently then
others as well.
2. Bill, I use NR occasionally on SSB as well. I certainly could narrow the SSB
bandpass with the BW and cut controls to better match the voice on the other
end, but I find that sometimes NR makes a nice improvement in the readability
and reduces the listening fatigue without me doing a lot of knob twisting.
Bill, keep watching for the DSP magic. I was recently talking to someone at
TenTec about DSP technology for a possible Orion II follow-on, whenever that
happens in the future. The current Orion DSP does, I think, about 60 million
multiply-adds per second (With two that gives you about 120 million multiply
adds per second.) The same family of DSP today has units that do 800+ million
multiply-adds per second per DSP. Two of those would increase the processing
power of an Orion by 10X+. I have found over the years - in technology - that
10X+ changes can be discontinuous. Some of the really complex algorithms that
people are studying today could then become possible to use. Of course the DSP
does much nore then noise reduction! Actually I kind of think of some of the
DSP functions in the Orion today as "like magic," because the mathematical
versions of many receiver functions work so much better - when correctly
implemented - then the "physical circuit" versions of the functions.
I hope this helps with what I was trying to say about NR. I certainly was not
trying to rewrite the laws of physics.
73,
Merle - W0EWM
|