A/B comparison test on 40m

An extended test comparing signals received in Florida from three nearby stations in Australia was conducted 21/12/13, a low elevation path of more than 14,000km.

The A/B comparisons were performed on statistics gathered from the W4HBK grabber, in this case measurements of the noise and each of two transmissions of 1min of steady carrier every 5min over a half hour.

The two systems were:

  • VK2DVK using a quarter wave ground plane at about 3m above ground; and
  • VK1OD using a G5RV using tuned feeders, in inverted V configuration at about 11m at the apex.

Observations were excluded where S/N of both signals was not greater than 10dB to reduce statistical noise, 293 paired observations remained.

Clip 115

The scatter chart above shows the difference in the signals S/N of both were greater than 10dB (click on the chart for a larger view). The large variation in the data is probably mainly due to ionospheric fading, but is there an underlying difference in the performance of the two stations?

In analysing experiments of this type, it is important to determine the probability that the result could be due to chance alone, that there is no underlying effect and that repeated experiments could go either way.

I have previously shown that received signal power is approximately log-normally distributed, and therefore power expressed in dB is approximately normally distributed and can be analysed using parametric statistics.

So, the question we will ask is “can we be 90% sure that there is no difference between the systems based on the measurements”. A two tailed paired t-test was used to test the null hypothesis Ho that there was no difference between the systems at confidence level of 90%.

The result is that there is no significant difference in the systems, the probability that the difference observed (0.3dB) was due to chance alone is 49.6%.