[TenTec] ARRL Phase Noise Data Plotted to 1 MHz

[Jim Brown]

Off list.

I've been thinking about this.

First, I think it's non-correlated noise, so the wideband noise would 
add by 10 log BW, where BW is the RX bandwidth that would need a 
correction factor for filter shape. A WAG for that correction factor 
might be 2-3 dB. So for a 400 Hz (nominal) filter, 10 log 400 = 26 - 3. 
So we would add 23 dB to the phase noise. The corresponding computation 
for a 2.8 kHz SSB RX would be 34.5 dB - 2 dB (another WAG) for the 
skirts, so add 32.5 dB.

Now, we want to add these random noise voltages from clicks to the 
random phase noise voltage, so we must convert both to volts and compute 
an RSS (root of the sum of the squares) sum. Many years ago, I did that 
when designing sound systems before there were computer programs to do 
it. My systems had multiple loudspeakers at different locations 
contributing to many seats, so I computed field strength  for each 
speaker and then did the RSS sum. My "computer" was an HP41 -- PCs 
weren't around yet. :)

And then we get to add the IMD.

Another question, of course, is, what are the units of the keying 
display? Is that also dBC/Hz? If so, the numbers for clicks really get 
scary!

What I do know is that P3 spectrum displays correlate quite well with 
what I hear.  Many of the top (and most active) contesters have gone to 
K3s, mostly because of RX performance, which has been widely publisized, 
but these data show them to be quite good on TX as well.

On the air, I can almost always identify a K3 on CW by its very narrow 
keying signature, simply because none of the other rigs are even close.  
And this -- K6XX is three miles from me with K3s and tube amps. I also 
run K3s and Titans. Both legal limit, we're both running (calling CQ) 
most of the time, but also "search and pounce" on a second radio.  His 
sidebands are 60 dB down at 350 Hz from his carrier, and we can work 500 
Hz apart and often not know the other is there.

73, Jim K9YC

On 7/27/2014 2:32 PM, Kim Elmore wrote:
> How might we look at a total noise figure? Phase noise is in dBC/Hz, 
> so we can't simply sum it with things like IMD. Should we integrate 
> the phase noise power as a function of bandwidth, so that we can look 
> at the phase noise power for CW and SSB (pick a defining bandwidths)? 
> How, then, do we handle the other noise sources that don't integrate 
> as nicely across the pass band?
>
> Perhaps there's no single number that we can generate -- each 
> represents a different "dimension" of the total problem.