[Skimmertalk] SNR Measurements

[Pete Smith]

Jim, this was one of the most useful contributions in a long time on 
skimmertalk. We're all interested to know how accurate SBR comparisons are, 
though at least we have a sense of relative strengths, when two signals are 
reported within a short time interval.

It's not apparent from what I read below whether the knee in the SNR curve 
is attributable to the SDR-IQ or to the Skimmer software.  Might there be a 
way to determine that?  I'm copying Pieter at RFSpace and Alex in the hope 
they can clarify their respective positions on this.

73, Pete

At 10:57 PM 12/9/2008, Jim Baremore wrote:
>My good friend Stan, K5GO, raised the question about the SNR readings from
>CW skimmer at a remote location and asked if they were correlated to the
>power of the transmitting station.  Some readings offered by others did not
>seem to correlate well to each other.
>
>
>
>Understanding the CW Skimmer is offering a Signal to Noise ratio and not a
>signal strength as such, led me to the following quick experiment:  I used
>my old Cushcraft CE-6 service monitor as a signal source and varied the
>input to the SDR-IQ while observing the SNR ratio on the CW Skimmer display
>connected to my SDR-IQ. I used the readings from the 'calibrated attenuator'
>to record the level being inputted to the SDR-IQ.  After I was done, I
>connected the Cushman to my Anritsu spectrum analyzer and measured the
>levels at each of the points I had previously recorded from the Cushman
>dial.
>
>
>
>I entered the results in an Excel spread sheet and did a quick plot.  What I
>found was that for a SNR from 2 to about 7, there was a distinct slope of
>the SNR vs. dBm of received signal that was different from SNR readings of 8
>to 35.  I also had a slight bias in one of my two instruments as the signal
>levels were both straight lines but were not the same slope. One of the two
>is calibrated slightly differently than the other.  I think I believe the
>Anritsu more than the Cushman, but who knows.
>
>
>
>Using the Cushcraft data for the low SNR ratios, I found the slope of the
>line was about .6 SNR per dBm.
>
>
>
> >From SNR = 8 to 30, the Cushcraft showed about 1.4 SNR per dBm.  The Anritsu
>showed about 1.77 SNR per dBm.
>
>
>
>In other words, I found a distinct knee for the low SNR ratios as it took
>about 1/.6 or about 1.7 dBm to increase the SNR by one unit.
>
>
>
>Above a SNR or 7 to 8, it changed and it looks like the average of the two
>instruments would be about 1/1.55 or only about .65 dBm to increase the SNR
>by one unit.
>
>
>
>A final note is my old Cushcraft is pretty old and may not be the purest
>signal possible.  It could have some flutter which could affect the SNR
>ratios as detected by CW Skimmer.
>
>
>
>Finally, remember this was a single signal into the SDR-IQ and is in
>contrast to an environment where it is connected to antenna listening to a
>lot of signals as well as atmospheric noise. I also noticed the absolute
>values of the SNR, at a given signal input level, varied considerably if I
>changed the bandwidth of the CW filter.  This says if you want to compare
>your SNR at two different remote locations, you should be sure they are both
>using the same CW bandwidth.
>
>
>
>
>
>OK,  nuff for now
>
>
>
>73
>
>
>
>Jim K5QQ
>
>
>
>
>
>_______________________________________________
>Skimmertalk mailing list
>Skimmertalk at contesting.com
>http://dayton.contesting.com/mailman/listinfo/skimmertalk