Three bands. A garden full of wire. Kevin McKeown, 09/10/202609/10/2026 M7XOM field notes · 9 October 2026Two Flowerpots, a vertical EFHW and the trusty 80 m reference. More reports for the challengers—but what do the signal strengths say?15 m · 17 m · 20 m / FT8 / PSK Reporter The resultsThe test15 m17 m20 mDirection & distanceMethod Yesterday’s antenna comparisons left me with a useful question: would the next round tell the same story? Today I put the 15 m Flowerpot, 17 m Flowerpot and 20 m vertical EFHW up against my 80 m EFHW reference, each on its respective band.The routine was the same: an odd/even pair of FT8 CQs, switch antennas, and repeat at the same power and frequency settings within each band. Then let the receiving stations around the world do the reporting.The short version: all three challengers appeared at more reporters. Only the 15 m Flowerpot showed a positive overall median SNR difference at receivers that heard both antennas. On 17 m and 20 m, the median was a draw. 15 m Flowerpot+2 dBMedian paired advantage · 208 pairs309 vs 244 reporters · +26.6%17 m Flowerpot0 dBMedian paired difference · 170 pairs247 vs 225 reporters · +9.8%20 m vertical EFHW0 dBMedian paired difference · 320 pairs462 vs 377 reporters · +22.5%The afternoon at a glance. Teal is the antenna under test; gold is the EFHW80 reference. The SNR panel shows the median and middle 50% of paired differences. Coverage percentages compare receiver counts, not antenna gain. Click the chart to view it full size.Within-band comparisons; positive SNR differences favour the antenna under test.Band / antenna under testReportersTest / referenceShared SNR pairsMedian / mean differenceStronger / equal / weaker15 m Flowerpot309 / 244208+2 / +2.39 dB143 / 18 / 4717 m Flowerpot247 / 2251700 / +0.10 dB79 / 18 / 7320 m vertical EFHW462 / 3773200 / +0.39 dB146 / 36 / 138 Six bursts, three comparisonsThe main activity is packed into roughly eight minutes. The test antenna went first on each band, followed about a minute later by the reference. These are sequential comparisons, not simultaneous measurements.Follow the clock. Bubble area shows the number of reports timestamped in each 15-second bin, not how long I transmitted. Two late 15 m records are visible alongside the later-band tests. Times are UTC; add one hour for British Summer Time. Click the chart to view it full size.That order matters. The exports do not establish an A–B–B–A sequence on any of these bands. Changing propagation can therefore contribute to a difference between antennas. Most reporters appear only once per antenna, so the files also do not preserve a complete odd/even signal-strength history for each receiver.I have kept the bands separate throughout. More receivers on 20 m than 17 m does not tell me that the 20 m antenna is better: band conditions, listening populations and the antenna under test all change. 15 m: the Flowerpot makes its caseThis is the clearest positive paired-SNR result of the afternoon. The Flowerpot appears at 309 reporters against 244 for the reference. Of the 208 receivers with usable SNR for both, 143 favour the Flowerpot, 18 are equal and 47 favour the reference.The median advantage is +2 dB; the middle half of differences falls between 0 and +5 dB. That is a useful improvement spread across many receivers, although it is still one short, ordered test.15 m in detail. There are 225 shared reporter identifiers, of which 208 have SNR for both. “Test” means the 15 m Flowerpot. The diagonal in the scatter plot marks equal SNR. Click the chart to view it full size.The farthest Flowerpot report is VK3GLN at 16,823 km, with −14 dB. The reference’s farthest is VK5ARG-2 at 16,238 km. At that shared receiver, the Flowerpot reports +2 dB and the reference −10 dB: an eye-catching +12 dB difference, rather than the typical result.Sensitivity check: one late report per antenna comes from IU8CKP, around 15:19–15:20 UTC. Removing those gives 308 vs 243 reporters, 207 SNR pairs, median +2 dB and mean +2.43 dB. The main interpretation is unchanged. 17 m: more reporters, a signal-strength drawThe 17 m Flowerpot appears at 247 reporters against 225. That is 22 extra identifiers, or 9.8%. The paired SNR result is much more balanced: 79 receivers favour the Flowerpot, 73 favour the reference and 18 report equal levels.Across 170 pairs, the median difference is 0 dB and the mean is +0.10 dB. The middle half spans −2 to +3 dB. I would call that broadly comparable signal strength in this sample.17 m in detail. Of 193 shared reporter identifiers, 170 have SNR for both antennas. One exact duplicate in the Flowerpot export was removed before counting. Click the chart to view it full size.Both antennas reach VK3GLN at 16,823 km: −11 dB on the Flowerpot and −10 dB on the reference. At VK5ARG-2, the reference is 5 dB stronger. Australia clearly does not award every round to the Flowerpot. 20 m: wider reporting coverage, similar signalsThe vertical EFHW appears at 462 reporters versus 377, a 22.5% increase. There are 131 vertical-only reporters and 46 reference-only reporters. That is a substantial difference in who appears in the exports.At the 320 shared receivers with SNR, however, the median difference is 0 dB. The mean is +0.39 dB, with 146 receivers favouring the vertical, 138 favouring the reference and 36 equal. The middle half spans −2 to +3 dB.20 m in detail. “Test” is now the 20 m vertical EFHW. There are 331 shared identifiers and 320 complete SNR pairs. The one repeated vertical reporter is counted once. Click the chart to view it full size.Both antennas reach VK3BVP at 17,056 km, the farthest report in either 20 m file. The reference is 3 dB stronger there. Elsewhere, VK2IO favours the vertical by 2 dB, and VK3BDX by 16 dB. Same broad destination, very different individual results. Where did each antenna do better?The polar plots place shared receivers by distance and initial geographical bearing from IO91WI. Teal means the antenna under test had the stronger report; brown means the reference did. All three use the same distance and colour scales.Geography, not a radiation pattern. North is at the top; radial labels are kilometres. The plots show the available paired reporters, not every direction in which the antenna radiated. Colour saturates beyond ±12 dB. Short-path bearing does not establish the actual ionospheric route. Click the chart to view it full size.Read the sample sizes with the medians. Positive values favour the antenna under test. Compass sectors span 90° centred on north, east, south and west. These groups describe different sets of receivers across bands. Click the chart to view it full size.15 m: the western sector has a +3 dB median advantage for the Flowerpot; east is +2 dB. North contains just eight pairs, so I would put little weight on its +0.5 dB median.17 m: north and east favour the reference at their medians, while south and west favour the Flowerpot. The overall draw conceals some geographical variation.20 m: west favours the vertical by 2 dB at the median; north favours the reference by 2 dB. East is a draw and contains more than half of the paired receivers.These are useful patterns to revisit. They are not yet permanent characteristics I can confidently assign to either installation.Is there a consistent DX advantage?The strongest distance-group result belongs to 15 m: both groups beyond 3,000 km have a +3 dB median Flowerpot advantage. On 17 m the longer-distance medians are smaller, at +1 and +0.5 dB. On 20 m they are +1 and 0 dB.The spread matters. Boxes contain the middle half of paired differences; the line marks the median. Whiskers extend to the most extreme observations within 1.5 interquartile ranges; circles show values beyond them. All panels share the same vertical scale. Click the chart to view it full size.The long-haul detail. The six farthest shared receivers with SNR on each band. A more negative SNR is weaker, even though its bar extends farther left. These selected examples illustrate variation; they are not a separate performance score. Click the chart to view it full size. Next: change the order, keep the wire stillFor the next round I want a logged A–B–B–A sequence on each band, then another starting with the reference. Same power, same odd/even CQ pair, same geometry, with UTC switch times written down. Preserving individual block reports, where the export permits it, is the other priority.For now, the useful conclusion is not that one antenna wins everywhere. It is that I have a promising 15 m result, two competitive alternatives on 17 m and 20 m, and specific paths to investigate. Quite a lot to get out of a few CQs and some wire in the garden. Method, filters and data auditAll results come from six PSK Reporter ADIF exports supplied on 9 October 2026. Each band uses its explicitly named pair: 15M FlowerPot / 15M EFHW80, 17M FlowerPot / 17M EFHW80, and 20M EFHW20 / 20M EFHW80. The exports overlap, so they are not concatenated. Antenna attribution follows the filenames and operator’s description for the relevant band.The common selection window is 14:43:52–15:43:52 UTC, the hour preceding this combined-report request. This retains the same relevant records as the slightly earlier hour windows used in the individual reports. Only FT8 transmissions by M7XOM or M7XOM/P are retained, with frequencies inside 21.000–21.450, 18.068–18.168 or 14.000–14.350 MHz, as appropriate. Receive-direction reports are excluded.CALL identifies the transmitter and OPERATOR the receiving reporter. Receiver identities are uppercase OPERATOR values, retaining suffixes such as -2. They identify reporters or receiver instances, not guaranteed independent physical sites. MY_GRIDSQUARE supplies the receiver locator; COUNTRY and DXCC describe the transmitting station in these exports and are not used to count receiving countries.Audit15 m test / ref.17 m test / ref.20 m test / ref.All input records2,950 / 8012,999 / 8472,999 / 847Selected raw reports309 / 244248 / 225463 / 377Exact duplicates removed0 / 01 / 00 / 0Unique reporting identifiers309 / 244247 / 225462 / 377Shared identifiers225193331Complete paired SNR208170320Exact duplicates are removed, then multiple reports for one receiver and antenna are reduced to median SNR. W3USR supplies two vertical reports on 20 m (+18 and +21 dB), represented by +19.5 dB; it has no reference report, so it does not affect paired statistics. Missing SNR remains missing and never becomes zero. Coverage includes reporters without SNR.Directions are initial great-circle bearings from the centre of IO91WI to each receiver’s Maidenhead locator centre. Coordinate conversion and bearings were checked against an independent geodesic calculation. Distances are the values supplied by PSK Reporter. Corrected bearings are used throughout this article. No antenna-height changes, unstated power values or external propagation conditions are assumed.The six files were downloaded at different times. This article uses the dedicated file pair supplied for each band, rather than silently substituting additional reports from later exports. Receiver counts are observations in those exports, not a guaranteed complete census of everyone who decoded the transmissions.Shared receivers experienced common propagation and transmission windows; they are not hundreds of independent antenna trials. No significance claim or confidence interval is used to imply repeatability from this single session. All comparisons describe the complete installed systems, including feed arrangements and surroundings. Antenna
M7XOM field notes · 9 October 2026Two Flowerpots, a vertical EFHW and the trusty 80 m reference. More reports for the challengers—but what do the signal strengths say?15 m · 17 m · 20 m / FT8 / PSK Reporter The resultsThe test15 m17 m20 mDirection & distanceMethod Yesterday’s antenna comparisons left me with a useful question: would the next round tell the same story? Today I put the 15 m Flowerpot, 17 m Flowerpot and 20 m vertical EFHW up against my 80 m EFHW reference, each on its respective band.The routine was the same: an odd/even pair of FT8 CQs, switch antennas, and repeat at the same power and frequency settings within each band. Then let the receiving stations around the world do the reporting.The short version: all three challengers appeared at more reporters. Only the 15 m Flowerpot showed a positive overall median SNR difference at receivers that heard both antennas. On 17 m and 20 m, the median was a draw. 15 m Flowerpot+2 dBMedian paired advantage · 208 pairs309 vs 244 reporters · +26.6%17 m Flowerpot0 dBMedian paired difference · 170 pairs247 vs 225 reporters · +9.8%20 m vertical EFHW0 dBMedian paired difference · 320 pairs462 vs 377 reporters · +22.5%The afternoon at a glance. Teal is the antenna under test; gold is the EFHW80 reference. The SNR panel shows the median and middle 50% of paired differences. Coverage percentages compare receiver counts, not antenna gain. Click the chart to view it full size.Within-band comparisons; positive SNR differences favour the antenna under test.Band / antenna under testReportersTest / referenceShared SNR pairsMedian / mean differenceStronger / equal / weaker15 m Flowerpot309 / 244208+2 / +2.39 dB143 / 18 / 4717 m Flowerpot247 / 2251700 / +0.10 dB79 / 18 / 7320 m vertical EFHW462 / 3773200 / +0.39 dB146 / 36 / 138 Six bursts, three comparisonsThe main activity is packed into roughly eight minutes. The test antenna went first on each band, followed about a minute later by the reference. These are sequential comparisons, not simultaneous measurements.Follow the clock. Bubble area shows the number of reports timestamped in each 15-second bin, not how long I transmitted. Two late 15 m records are visible alongside the later-band tests. Times are UTC; add one hour for British Summer Time. Click the chart to view it full size.That order matters. The exports do not establish an A–B–B–A sequence on any of these bands. Changing propagation can therefore contribute to a difference between antennas. Most reporters appear only once per antenna, so the files also do not preserve a complete odd/even signal-strength history for each receiver.I have kept the bands separate throughout. More receivers on 20 m than 17 m does not tell me that the 20 m antenna is better: band conditions, listening populations and the antenna under test all change. 15 m: the Flowerpot makes its caseThis is the clearest positive paired-SNR result of the afternoon. The Flowerpot appears at 309 reporters against 244 for the reference. Of the 208 receivers with usable SNR for both, 143 favour the Flowerpot, 18 are equal and 47 favour the reference.The median advantage is +2 dB; the middle half of differences falls between 0 and +5 dB. That is a useful improvement spread across many receivers, although it is still one short, ordered test.15 m in detail. There are 225 shared reporter identifiers, of which 208 have SNR for both. “Test” means the 15 m Flowerpot. The diagonal in the scatter plot marks equal SNR. Click the chart to view it full size.The farthest Flowerpot report is VK3GLN at 16,823 km, with −14 dB. The reference’s farthest is VK5ARG-2 at 16,238 km. At that shared receiver, the Flowerpot reports +2 dB and the reference −10 dB: an eye-catching +12 dB difference, rather than the typical result.Sensitivity check: one late report per antenna comes from IU8CKP, around 15:19–15:20 UTC. Removing those gives 308 vs 243 reporters, 207 SNR pairs, median +2 dB and mean +2.43 dB. The main interpretation is unchanged. 17 m: more reporters, a signal-strength drawThe 17 m Flowerpot appears at 247 reporters against 225. That is 22 extra identifiers, or 9.8%. The paired SNR result is much more balanced: 79 receivers favour the Flowerpot, 73 favour the reference and 18 report equal levels.Across 170 pairs, the median difference is 0 dB and the mean is +0.10 dB. The middle half spans −2 to +3 dB. I would call that broadly comparable signal strength in this sample.17 m in detail. Of 193 shared reporter identifiers, 170 have SNR for both antennas. One exact duplicate in the Flowerpot export was removed before counting. Click the chart to view it full size.Both antennas reach VK3GLN at 16,823 km: −11 dB on the Flowerpot and −10 dB on the reference. At VK5ARG-2, the reference is 5 dB stronger. Australia clearly does not award every round to the Flowerpot. 20 m: wider reporting coverage, similar signalsThe vertical EFHW appears at 462 reporters versus 377, a 22.5% increase. There are 131 vertical-only reporters and 46 reference-only reporters. That is a substantial difference in who appears in the exports.At the 320 shared receivers with SNR, however, the median difference is 0 dB. The mean is +0.39 dB, with 146 receivers favouring the vertical, 138 favouring the reference and 36 equal. The middle half spans −2 to +3 dB.20 m in detail. “Test” is now the 20 m vertical EFHW. There are 331 shared identifiers and 320 complete SNR pairs. The one repeated vertical reporter is counted once. Click the chart to view it full size.Both antennas reach VK3BVP at 17,056 km, the farthest report in either 20 m file. The reference is 3 dB stronger there. Elsewhere, VK2IO favours the vertical by 2 dB, and VK3BDX by 16 dB. Same broad destination, very different individual results. Where did each antenna do better?The polar plots place shared receivers by distance and initial geographical bearing from IO91WI. Teal means the antenna under test had the stronger report; brown means the reference did. All three use the same distance and colour scales.Geography, not a radiation pattern. North is at the top; radial labels are kilometres. The plots show the available paired reporters, not every direction in which the antenna radiated. Colour saturates beyond ±12 dB. Short-path bearing does not establish the actual ionospheric route. Click the chart to view it full size.Read the sample sizes with the medians. Positive values favour the antenna under test. Compass sectors span 90° centred on north, east, south and west. These groups describe different sets of receivers across bands. Click the chart to view it full size.15 m: the western sector has a +3 dB median advantage for the Flowerpot; east is +2 dB. North contains just eight pairs, so I would put little weight on its +0.5 dB median.17 m: north and east favour the reference at their medians, while south and west favour the Flowerpot. The overall draw conceals some geographical variation.20 m: west favours the vertical by 2 dB at the median; north favours the reference by 2 dB. East is a draw and contains more than half of the paired receivers.These are useful patterns to revisit. They are not yet permanent characteristics I can confidently assign to either installation.Is there a consistent DX advantage?The strongest distance-group result belongs to 15 m: both groups beyond 3,000 km have a +3 dB median Flowerpot advantage. On 17 m the longer-distance medians are smaller, at +1 and +0.5 dB. On 20 m they are +1 and 0 dB.The spread matters. Boxes contain the middle half of paired differences; the line marks the median. Whiskers extend to the most extreme observations within 1.5 interquartile ranges; circles show values beyond them. All panels share the same vertical scale. Click the chart to view it full size.The long-haul detail. The six farthest shared receivers with SNR on each band. A more negative SNR is weaker, even though its bar extends farther left. These selected examples illustrate variation; they are not a separate performance score. Click the chart to view it full size. Next: change the order, keep the wire stillFor the next round I want a logged A–B–B–A sequence on each band, then another starting with the reference. Same power, same odd/even CQ pair, same geometry, with UTC switch times written down. Preserving individual block reports, where the export permits it, is the other priority.For now, the useful conclusion is not that one antenna wins everywhere. It is that I have a promising 15 m result, two competitive alternatives on 17 m and 20 m, and specific paths to investigate. Quite a lot to get out of a few CQs and some wire in the garden. Method, filters and data auditAll results come from six PSK Reporter ADIF exports supplied on 9 October 2026. Each band uses its explicitly named pair: 15M FlowerPot / 15M EFHW80, 17M FlowerPot / 17M EFHW80, and 20M EFHW20 / 20M EFHW80. The exports overlap, so they are not concatenated. Antenna attribution follows the filenames and operator’s description for the relevant band.The common selection window is 14:43:52–15:43:52 UTC, the hour preceding this combined-report request. This retains the same relevant records as the slightly earlier hour windows used in the individual reports. Only FT8 transmissions by M7XOM or M7XOM/P are retained, with frequencies inside 21.000–21.450, 18.068–18.168 or 14.000–14.350 MHz, as appropriate. Receive-direction reports are excluded.CALL identifies the transmitter and OPERATOR the receiving reporter. Receiver identities are uppercase OPERATOR values, retaining suffixes such as -2. They identify reporters or receiver instances, not guaranteed independent physical sites. MY_GRIDSQUARE supplies the receiver locator; COUNTRY and DXCC describe the transmitting station in these exports and are not used to count receiving countries.Audit15 m test / ref.17 m test / ref.20 m test / ref.All input records2,950 / 8012,999 / 8472,999 / 847Selected raw reports309 / 244248 / 225463 / 377Exact duplicates removed0 / 01 / 00 / 0Unique reporting identifiers309 / 244247 / 225462 / 377Shared identifiers225193331Complete paired SNR208170320Exact duplicates are removed, then multiple reports for one receiver and antenna are reduced to median SNR. W3USR supplies two vertical reports on 20 m (+18 and +21 dB), represented by +19.5 dB; it has no reference report, so it does not affect paired statistics. Missing SNR remains missing and never becomes zero. Coverage includes reporters without SNR.Directions are initial great-circle bearings from the centre of IO91WI to each receiver’s Maidenhead locator centre. Coordinate conversion and bearings were checked against an independent geodesic calculation. Distances are the values supplied by PSK Reporter. Corrected bearings are used throughout this article. No antenna-height changes, unstated power values or external propagation conditions are assumed.The six files were downloaded at different times. This article uses the dedicated file pair supplied for each band, rather than silently substituting additional reports from later exports. Receiver counts are observations in those exports, not a guaranteed complete census of everyone who decoded the transmissions.Shared receivers experienced common propagation and transmission windows; they are not hundreds of independent antenna trials. No significance claim or confidence interval is used to imply repeatability from this single session. All comparisons describe the complete installed systems, including feed arrangements and surroundings.