Saturday 5 August 2017

WSPRlite - What's Half a Decibel Between Friends?

Power measurement has been the flavour of the week here at Copper Mountain HQ.

Inevitably, attention has turned to the extraordinarily-successful WSPRlite unit, which is marketed as an "antenna performance analysis system" (WSPRlite web page accessed 21:25UT 05/08/2017).

Although the output of the WSPRlite is tiny - just 200mW at most - it is, nevertheless, readily measurable with an oscilloscope.

Now, power measurement that is traceable to a known, stable standard is something we would all like to have, but few ever get a chance (or the means) of owning.  So my own measurements, even with a calibrated oscilloscope, have to be taken with a considerable degree of caution.

My scope showed a somewhat lower output than 200mW, and I know of others who have measured around 170mW at 200mW set.  If - and I stress if - there is a significant difference between set and actual output, then that is not unusual for any instrument, no matter how expensive, and we do always have to keep in mind this particular unit sells for quite a low price.

The problem for the end-user of WSPRlite is that he/she has no idea, other than relying on the selected set value, what that particular unit is putting out on any given transmission. I certainly wasn't provided with a calibration certificate or statement of output accuracy when I bought mine, nor when I received a replacement when that failed.  I couldn't find any reference in current documents to power accuracy in the instructions or FAQs published online (accessed 07/08/2017).

This is odd, because SOTABEAMS appear to be aware of the need for accuracy. This is what they say on their web site (accessed 07/08/2017):

'WSPRlite has accurate digital power control over a 16 dB range to facilitate precise measurements of antenna performance differences.'

Note what appears to be a conflation of accuracy and precision in that sentence.  They are not the same thing. It is possible to be accurate with low precision.  The graphic below illustrates:




Contact with SOTABEAMS today gave a very brief response that the "spec is +/-0.5dB". Without any elaboration from the company, I am taking that to mean that each unit yields that degree of power output accuracy at any one time, relative to the set value.  An alternative possible interpretation could be that each unit, once assembled, always transmits with a fixed deviation within the stated range from the set value.

I was also told that SOTABEAMS "measure" the ouput (presumably on an ongoing sample of manufactured units), using an "in-spec Agilent power meter".  I wasn't told which Agilent unit (the company is now known as Keysight), nor was any information on how that unit, in turn, is kept calibrated on an ongoing basis against a known, independent reference standard.

The trouble with trying to be scientifically-accurate is that it is hard

The stated accuracy is very good for such an affordable and widely-available unit.  But we must always keep in mind that declared purpose: to analyse antenna performance (with both accuracy and precision).  That often means (and the associated DXPlorer webware features heavily leans towards) comparisons with other antenna systems using the same kind of WSPRlite unit.  Accuracy of actual vs. set power output is therefore undeniably important to the end-user where any comparison with another user is being undertaken.

If one unit is running at +0.5dB somewhere, and another happens to be running at -0.5dB, there is obviously a maximum potential inter-station output difference of 1dB.

At a set value of 200mW, this equates to outputs of 224mW at +0.5dB, and 178mW at -0.5dB.  Or, a difference of 46mW between highest and lowest potential outputs.  And that is assuming the power output accuracy is always within the stated range.

Whether this difference of +/- 11% or so in output is of significance in the real world of reception reports is up for debate, and comments of a reasonable nature are always welcome. My own view is that this - being almost 1/20th of a Watt -is sufficient for a clear difference to appear in received signal strength and reach.

Indeed, because there is a 50mW and 5mW setting (or 50mw and 100mW) on the WSPRlite unit, it is quite easy to run a real comparison of the difference 46mW makes, provided the antennas are exactly the same systems and located in the same environment (and these aspects are absolutely crucial), and run at the same time.

I suppose it does act, at least, as a starting point for getting the inter-unit differences down even lower sometime in the future.


No comments: