I've had the opportunity to measure NR performance on several firmware loads
on a v1 Orion and the Orion II (2.032).
The results are both not surprising and surprising at the same time. All of
these measurements are with a constant carrier in the presense of a roughly
S-5 noise level (Elecraft n-Gen and HP 8640 through a T-T 651 combiner).
This table may not be easy to read unless you are using a fixed width font
on your mail reader.
All values are the SNNR improvement (in dB) from NR=0ff to NR=9. This is a
static measurement, and may not reflect perceived performance against a
dynamic signal. For example, I happen to think NR on the II works
reasonably well on SSB even given the apparently poor relative numbers. Not
so on CW. But if you consider that even the v1 NR doesn't help SNNR at
narrow bandwidths, perhaps it falls out in the wash. Perhaps. Values are
in the range of +/-1 dB.
1.371 1.372 1.373b5 2.056 2.032
-----------------------------------------------------
3 Khz +6.7dB +6.5dB +6.5dB +0.4dB +1.0dB
2 Khz +5.4dB +5.5dB +5.4dB -0.7dB no data
1 Khz +3.5dB +3.7dB +3.6dB -0.8dB -0.4dB
500 Hz +1.0dB +0.5dB +0.5dB -1.2dB -1.0dB
100 Hz -0- -0- -0- -0.9dB -2.0dB
Clearly, by this metric, there is a substantial difference between 1.3x and
2.0x versions of NR (stating the obvious). I had expected to see some
difference between 1.371 and 1.372, given the view held by some that 1.371
was the last "good" version of NR. There are other factors of course, and I
haven't done much in the way of subjective comparisons. From the point of
view of SNNR improvement (something that might actually help you copy a weak
signal in noise), however, the currently popular 1.37x versions are
identical in performance. None of the versions do much once the bandwidth
gets down in the 500Hz range. I'm not surprised by the results on v1 at 500
or 100 Hz. Something isn't quite cricket with v2/II -- either SNNR really
is negatively affected by NR or there are distortion products disturbing the
measurement that aren't obvious on the FFT spectrum plots.
One thing substantially better about v1 is that NR is clearly INSIDE the AGC
loop, whereas v2/II does not appear to be. Hopefully that will change. On
the downside, if you are using NR on the v1, AN doesn't work - but the
manual notch does.
Unfortunately, I won't have access to this v1 long enough to really run the
two radios side-by-side for the extended period of time I had hoped. But
I'll probably go on the hunt for another one in the not too distant future.
Grant/NQ5T
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|