Non-linearity, straightened out.

Last week, a strange WSPR anomaly was causing some headaches for me.

The problem was that G0CCL, a very useful beacon in Cambridgeshire that switches its output from 1W to 5W at intervals, was receiving strange signal reports in that the 7dB switch from 1W to 5W yielded, at various receivers I selected, a 12.5dB improvement in signal.

This seemed to be odd to me, so I asked the keeper of G0CCL to examine the setup, in case settings had drifted over time.

In the true spirit of ham radio and scientific enquiry, Jonathan Kelly, G2HFR, attached an antenna to his spectrum analyser, and found...a 12.8dB difference between the nominal settings of 1 and 5W!  A person of lesser integrity would have made excuses, failed to respond, or even denied the anomaly.  So, very much credit is due to Jonathan.

Something not quite right.  A supposed 7dB power setting change yields a 12.86dB difference in actual output.  Image courtesy Jonathan Kelly.

The agreement between analysing remote receiver reports and a local spectrum analyser assessment is remarkable - just 0.3dB difference!  I was rather happy with that, and the fact that it pointed to a problem that needed resolving with the drive amplifier of the Raspberry Pi system in use at G0CCL.  Jonathan is going to change it for a log amp with more steps next week, as a result of this finding.

As this had complicated my analysis of seaside and inland stations somewhat, I bought another WSPRlite transmitter, and set off for the beach one, last time this afternoon.  The temperature was hitting the upper 20's Celsius, which was really quite amazing for the end of September.

Warm weather won't last much longer.  At IO73rf, Traeth Llydan, west coast Anglesey.

Now I could compare two identical elevated 1/4 wave verticals with 2 radials, one at home, one at the beach, each putting out 200mW for a couple of hours.  As the tide was high when I began, and the beach slope steeper than when it's lower, I was able to keep the seaside antenna with its feet right at the water's edge throughout, occasionally being washed by waves.

The results are very interesting - and very different from fully inland stations, which are worse than my hilltop, home vertical only a few km from the coast.  My seaside antenna produced about 2dB better reports across all distances and all stations, increasing to 8dB as the distance reached long-haul DX at beyond 6000km or so.  Most eastern US stations were reporting a 5dB better signal from the seaside.

This is broadly comparable to the ~10dB improvement seen from earlier coastal work, although there were no dramatically better signals this time.  Signals to the east (landward) were a maximum of 3dB better but, generally, they were marginally (0.5-1.5dB) weaker from the seaside.

Sadly, I'm reluctant to compare with others' inland stations, because there are too many things that we can't know about them.  Some use rubbish antennas, others are surrounded by buildings, and the significance of these isn't always appreciated by the users.

As with other work, I found a significant number of stations - up to 40% - that heard me from the seaside, but not at all from home.  Those are perhaps more telling and important than the other comparisons.

Another crucial element to examine was the frequency (how often heard) of spots for each station, which is very revealing: the more inland antenna had, on average, only a quarter the number of spots from any given station that the seaside antenna did. 

All distances, all stations = 2dB improvement from the seaside.

Eastern side of US reporting 5dB improvement from the seaside.

8dB improvement from the seaside for far west US receivers.