On 17 Sep 2009 at 9:23, Roger Lapthorn wrote:

> Just had these 2 reports. I am very surprised WSPR can detect these low
> S/N levels...
> 
>  2009-09-17 08:12  G3XBM  0.503838  *-33*  1  JO02dg  0.001  G7NKS 
>  IO92ub 46  240   2009-09-17 07:56  G3XBM  0.503839 * -32*  0  JO02dg 
>  0.001 G7NKS  IO92ub  46  240
> 73s
> Roger

Still nothing at all here Roger, there were a couple of very weak 
broken lines on the waterfall but no copy.

I tend to take the dB figures in WSRP with a pinch of salt. Jim, James 
and Gus are quite strong signals and not only audible by ear all the 
time but reading well on the s meter - James 579, Gus peaking over s9 
at times but with deep QSB, and Jim - if it wasn't a row of LEDs it 
would be bending the meter. G7NKS shows between -3dB and -17dB on WSPR 
(with the input adjusted for a correct 0dB on noise), implying he is 
below the noise... I think all that can be deduced is the relative 
strengths against an unknown reference. With the signals heard here 
WSPR7 gives reliable decodes down to around -20dBm and a few gibberish 
decodes at the -30dBm level. I have not copied anything lower, and when 
I do get decodes the signal is easily copyable by ear and a very 
obvious line on the spectrogram.

One point is that with the K2 you cannot set a narrow filter at 1.5kHz 
(a limitation of the tuning range of its IF BFO). I am sure if you used 
such a narrow receive filter the results would be spectacularly better, 
but I am stuck with the SSB filter.

73 Dave G3YMC

http://www.davesergeant.com