Dale uses WSPR to test counterpoise orientation


Many thanks to Dale (N3HXZ) who shares the following guest post:

Does your antenna counterpoise orientation matter?

by Dale (N3HXZ)

I am an avid SOTA and POTA activator and love field operation. I use a portable vertical whip antenna with a single counterpoise for my antenna system and have always wondered if orienting my counterpoise would provide some signal strength gain in a particular direction.  I decided to run a series of tests using WSPR to gather field data, and use statistics to answer the question:

Does one counterpoise orientation favor another in terms of average signal strength?

WSPR is a great tool for antenna testing. You can study various antenna configurations by making some WSPR transmissions and then checking the data on the WSPRnet database to see how well the signal was received at various stations located all over the world. You have to be careful in interpreting WSPR data though as receiving stations have different antenna and radio configurations, and the band propagation can vary rapidly at times. So how do you take advantage of all the data you receive from stations and draw some meaningful conclusions? I have found that using proven statistical theory in analyzing the transmitted signal strength received from individual stations can provide you results that you can confidently trust.

So what statistical algorithm is helpful?

For antenna signal strength comparison between two configurations, you can use an independent two-sample t-test with a one-tailed t-test evaluation. It sounds like a mouth-full, but it is quite simple. For our purposes, the t-test compares the average signal strength at a given receiving station from two different antenna configurations. The one-tailed test validates or invalidates the hypothesis that one antenna configuration produces an average signal strength greater or less than the other antenna configuration.

The testing requires that you run WSPR long enough to gather multiple reports at a single receiving station for both antenna configurations. Using the signal strength reports, you compute the average signal strength and the standard deviation of the signal strength over the sampled data points. Excel can easily provide that data.  With this information and the number of sample points for each antenna configuration (they can be different), you then run a calculation by hand or in Excel to compute the ‘t’ value.

This ‘t’ value is then compared to a critical value for the number of sampling points from a ‘Students t table”. If the ‘t’ value is less than the critical value you can confidently conclude that the hypothesis is false and therefore conclude that there is no significant difference in the mean value of the signal strengths between the two.  If the ‘t’ value is greater than the critical value you can accept the hypotheses that one antenna configuration produces a greater or less average signal strength than the other configuration.

For reference there is an easy to read article introducing the t-test. Another article gives example data and results from a t-test formula calculation. Lastly, you can reference  the Students t table chart to determine the critical values to compare against you t test calculation.

Let’s now review the testing that was performed.

I used a 17’ vertical whip antenna configuration with a single counterpoise from the Chameleon MPAS Lite product. I ran two configurations with the counterpoise at 90 degrees to one-another.  Please see Figure 1.

Figure 1 Vertical Antenna Counterpoise Configurations

One set of tests was run with the “NE” counterpoise pointing north-east. The second set of tests was run using the “SE” counterpoise pointing to the south-east.

WSPR transmits for 110 seconds every 4th 2-minute cycle.  Between transmissions I switched between counterpoise configurations to minimize band propagation fluctuations between data sets. I transmitted at 5 Watts on the 20M band. The WSPR session lasted roughly 2 hours and 15 minutes. I collected data from 63 stations with 205 reports from my QTH in Pittsburgh. A view of the stations receiving my transmissions is shown in Figure 2.

Figure 2: WSPR stations receiving my transmissions (courtesy WSPRnet.org)

A sample of data received from a single station is shown in Table 1.  The receiving station was NI5F in Graceville FL.

Table 1 WSPR results for station NI5F

Six sets of signal strength (SNR in dB) data were collected for both counterpoise configurations. The average signal strength and standard deviation were calculated from Excel. The calculated t-test value is shown and the formula highlighted in the formula bar. The Students t-table is used to get the critical value. It requires the ‘degrees of freedom (df), which is the sum of the two data set counts minus 2. (6+6-2)=10.

A table look-up at a df=10 gives a critical value of 1.812 assuming a 95% confidence interval. The absolute value of the t-test (0.955) is less than the critical value (1.812). Therefore the ‘hypothesis’ that the average signal strength from the two counterpoise configurations is different is false; and the conclusion is that the mean value of the signal strengths for the two configurations are not significantly different.

An overall summary of five other stations is given in Table 2. Stations were selected from the three ‘quadrants’ pointing northeast, northwest, and southwest. (There were no reports from the southeast quadrant). These stations received signals with either the counterpoise pointing roughly in a line directed at their location, or in a line broadside to their location. In all cases, the hypothesis that the average signal strengths differed between counterpoise configurations were proven false. Hence, the overall conclusion is that the orientation of a single counterpoise for a vertical antenna does not measurably impact the average signal strength received by a given station.

Table 2 t-Test summary for five stations

Gathering signal strength data from WSPR and analyzing it with a proven statistical method provides a way to confidently assess the performance of one antenna configuration against another. The calculations are simple; you just need patience in collecting data over a couple hours!

30 thoughts on “Dale uses WSPR to test counterpoise orientation”

  1. Thanks for doing this study Dale. Earlier in the month I used my MPAS Lite and as I set out the counterpoise I had the very same thought you did – does direction matter?? Glad to know that it does not, so now I’m free to deploy it in the most convenient direction as (in fact) the MPAS Lite directions tell you to do. Did you also just use 25 feet of counterpoise, as directed?

    Makes me wonder if the 4 counterpoise/radials used with the MC-750 are actually needed, or if one is sufficient. Is there something different about this vertical that makes all 4 radials critical?

    Mark – W8EWH

    1. Mark
      Thanks for the inquiry. I used a 25’ counterpoise. Can’t answer your question on the MC750. I have tried a 4 counterpoise configuration with the Mpas light vertical but have not noticed any appreciable difference during field operation. Chameleon does state that the 4 counterpoise configuration greatly enhances overall efficiency.

      1. Dale,
        This brilliant. I am excited to use this same methodology as a framework for my own testing.

        Thank you for this post!

  2. Hi Mark
    Thanks for your question. Yes, my counterpoise was 25’. I have tried a 4 counterpoise configuration with my Mpas lite vertical,. There was no earth shattering change is signal reports but I have not yet done an extensive study on it like this one. A good test for the future!

  3. The easiest way to determine the effect of your ground or counterpoise is by tying a piece of nonconductive rope to the base of the the antenna and read Field Strength meter as you walk in a circle. Plotting your readings are quite revealing about each of your antennas.

    I love WSPR and use it each time I go portable to check propagation but a FS meter is better to use for antenna design……. IMHO.

    1. The problem with using FS readings is that near-field and far-field patterns can be quite different.
      73 de Arnie W8DU

  4. Dale, thanks for a superbly laid out article, one which easily should be in the QEX journal. I use WSPR a lot but have never done the statistical work as you have to make it truly meaningful. Thanks again and very well done!
    Vy 73, Rick -VE1RNM

    1. Rick
      Thanks for the comments. I think writing for QEX is a good idea. I will pursue this.

  5. Years ago, the HF Manpack Portable group did an “HF Antenna Shootout” testing various antenna systems. I believe it was a Yahoo User Group… In their evaluations, they compared each antenna to a vertical resonant wire, suspended by a telescopic fiberglass mast. At the feedpoint, was an elevated radial sloping and extended from the feedpoint at 6′, sloping down to 3′ off the ground. This was preported to have gain extending in the direction of the elevated radial.

    In my field experience seemed to have gain to me, but at the time WSPR beacons weren’t available to test it. Here’s my question: Would an an elevated sloping radial provide any gain?

    Great report! de W7UDT

    1. Thanks for the question Pete. I would like to do more studies with the counterpoise pointed to the receiving station., I.e., 180 degrees from my tested positions.

  6. What kind of soil was around the antenna? I think thay could be important when measuring a low ground system behaviour.
    Thanks for the analysis!

    1. Hi Mikel
      Thanks for the question. The ground consists of a few inches of topsoil over a clay base.

      1. the main point, aside from counterpoise orientation, which usually gives a slighf better signal due to radiation lobe deformation, is the effect the counterpoise(s) have, if ARRL and others aren’t wrong, adding more radials to a vertical will improve its radiation efficiency, and that would be a nice test…

  7. Hi Dale,

    I applaud your work here! It’s important for us to use databases like WSPR to further our knowledge. I’m a scientist with HamSCI, having a PhD in sociology and statistics and teaching the later for some 45 years. Let me encourage you to consider a couple of things to strengthen your results.

    A. If I am reading your Table 2 correctly, the sample sizes (NE count, SE count) are small (e.g., 2-8). If this is correct, please note that the parametric t-test needs n=/> 30 for the observed t-tests to be reliable. See https://en.wikibooks.org/wiki/Statistics/Testing_Data/t-tests. The bias in these small sample sizes is the lack of statistical power, the ability to detect a “true” difference where one actually exists. Thus, I’m not at all surprised that you fail to reject any of the null hypotheses…you don’t appear to have enough observations to detect significant differences.

    B. Consider reanalyzing your data with a non-parametric test of means: Mann-Whitney U or Wilcoxon rank-sum test. Most free statistical software will do this (e.g., PSPP, etc.). Most directly import Excel spreadsheets or CSV.

    Bear in mind that statements about the null hypothesis (H0) are not framed as true or false. The analyst either rejects the null or fails to reject the null. To claim that it is “false” is another set of probabilities involving alpha and beta error.

    If I am not fully understanding the sample sizes I’m reading from Table 2, then my suggestions might not be appropriate.

    I know that time an be challenging for such a setup. Thank you for your work on this. I’d encourage to consider collecting more data so as to have a more definitive study. If I may be of any assistance, please contact me. Im good on QRZed as K4FMH.



    1. Hi Frank
      Thank you for your comments. I was hoping that someone with expertise in statistics would read the article and comment. We should follow up off-line to discuss this further.
      My understanding is that the t-test does not have a minimum size requirement, but too small of a sample size could mask ‘meaningful’ statistical differences. But meaningful statistical differences may not translate into practical differences sufficient for an operator to worry about where to throw the counterpoise. I think the conclusion is reasonable even if the sampling size is on the small side. However I am interested in exploring the other statistical tests and welcome your help!

  8. Dale, thanks for the great WSPR analysis.

    Last year, I ran a WSPR test on the MPAS vertical with and without the Chameleon Cap Hat. In both cases, I used the supplied 25′ counterpoise.

    My conclusion at that time was that the Cap Hat improved the reported received signal strength by almost 20%. However, I now have access to another MPAS vertical and 2 WSPR transmitters, so you’ve inspired me to run a new test with 2 antennas and 2 transmitters at the same time.

    I may ask you for assistance in interpreting the results.


    1. Thanks Keith. Having multiple transmitters running simultaneously would allow you to collect twice the amount of data that I collected in the same timeframe. This would enhance the statistical accuracy as Frank (K4FMH) points out. Good luck!

  9. Thanks for your in-depth search in the counterpoise delema. I have often wondered if more make a difference. I operate QRP in the field with an end fed antenna and I thing you have saved me a lot of time.
    Thanks again and look forward to more research from you.

  10. Sorry, but you have two counterpoises, your coax shield is the second one. Also, there is a difference in gain, according my antenna models, but 2-3 dB is probably too small to be measured with your sample size.

    73 Stephan, HB9EAJ

  11. I love to see people doing actual analysis of antennas rather than just empirical reports. Here are a few comments.

    1) I don’t think you mentioned whether the counterpoise was elevated off the ground or on the ground. I would only expect to see a measurable difference for a raised counterpoise.

    2) The theoretical beam width of a vertical with an elevated counterpose is pretty wide. I think your best bet at measuring a difference is going to be at 180 degrees, not 90 degrees. A vertical with a single elevated radial at 0.05*wavelength height should yield 3-4dB of front-to -back.

    3) The picture in the article implies that you were on sloping ground. The pattern is probably stronger on the downward sloping side. Since the difference you are attempting to measure is relatively small, this could affect the results.


    1. Hi Scott,
      Thanks for the comments! You’ve given me a few good items (elevated radial and 180 degrees) to try at my next series of tests. Actually, Thomas pulled a picture of the Mpas Lite antenna from his archives! My actual terrain was level.

  12. Excellent study and report. In my experience I find 4 – 25 foot counterpoise radials (ie the Chameleon 4 wire counterpoise kit) help negate the effect of a single counterpoise and the coax shield which is part of your setup. I notice better antenna resonances, tuning and seemingly more contacts with a 4 counterpoise system. A single counterpoise is helpful but not effective. I have done several ML antenna studies and have some WSPR data where I discovered (as mentioned earlier) that my sample size was too small. I upped the study to several days and a couple thousand sorted data collections to realize other factors contribute as well, such as RF choke beads at one end of your 15 or 20 coax feed line , how the feed line is laid out on the ground, dangling in the air etc. I would recommend increasing the sample size as there should be some very small differences with your setup ie moving the wire and moving the feed line approach.
    I welcome discussing the testing and don’t mind helping and investigation of this area, Good Job thanks

    1. To all
      I want to thank everyone for your feedback and instruction. Several of you have offered to help further the cause. My main goal is to establish a sound statistical approach from which we can draw meaningful conclusions. Once that is in place it opens up a whole world of investigations as many of you have pointed out.
      The t-test provides a good framework for comparing mean responses. As Frank (K4FMH) points it, the statistical power for detecting true differences lies in the sample size. I plan to run another study to increase the sample size and see if the hypothesis test results change. Stay tuned!

      1. Hi again. Let me know if you wish to test direct wspr circuit to St Augustine FL. I would be glad to help out . Best regards K3GO. John

  13. What a great article. I’m not familiar with your antenna system they are prohibitively expensive over here in the U.K. they do seem popular with your POTA stations.

    Experimentation I’ve done was to elevate a vertical whip about a meter above ground with one then two etc elevated radials. I did notice a reasonable difference when comparing to ground mount. Directivity wise with one radial there was an improvement with SNR via WSPR. But it’s hard to replicate too many experiments with band conditions changing so quickly.

    There is much debate over here as to ground mount versus elevated verticals, it’s radio after all and with our lots so much smaller than your average we look to squeeze everything with can. However ground mount is just more practical and less visible. For POTA/SOTA radials become a magnet to attract passers by so for sheer ease I use EFHW.

    Our hobby is just magical and it’s articles such as yours that inspires people to just ‘give it a try’. Kindest regards and 73. Mike M0AZE

  14. I just today purchased the MPAS Lite and was wondering if the direction of the counterpoise would enhance performance. Thank you for your efforts and to all that responded. Very informative for a new user like me.

    Gerry W8PDS

    1. Hi Gerry
      Thanks for you question. Congratulations on your purchase, you’ll find the MPAS Lite to be a dependable and amazing antenna system. I’ve done more extensive studies since my initial report gathering data from Wspr on counterpoise orientation. The data still does not support the notion that orientating the counterpoise on the ground provides any measurable directional gain. I’m still analyzing the data and have yet to try a full front-to-back (180 Deg.) orientation case, but for now I would not be concerned about the orientation. Just find a convenient location to deploy it and have fun operating!

      1. Very interesting and thoughtful experiment, I would be surprised to see any directivity when radials are based on the ground. However a counterpoise at ¼ wave about 1 meter above ground did exhibit a difference for me. For transparency I did follow Buddipole setup arrangements. Test need to be repeated under different conditions to make my experiments valid. Enjoy your new antenna and experiment for yourself; ground mount everything and then try elevating the antenna and counterpoise. Whatever you do have fun.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.