Two wires, one planet: putting my EFHWs to the test

An afternoon of FT8 antenna comparisons, a few hundred receiving stations, and some rather pretty graphs.

I’ve spent quite a lot of time swapping antennas and deciding that one feels better than another. More stations on the waterfall, more spots on PSK Reporter, the occasional pleasing bit of DX. Useful observations, but I wanted to put some numbers behind them.

So I compared my 20 m vertical EFHW with my 80 m EFHW, using both on 20 metres. The longer antenna is the reference. Transmit power and the other settings stayed the same.

The short version? Almost identical signal strength at receivers that heard both. The vertical appeared at more receiving stations, but there’s a wrinkle before I award it a medal.

The experiment

The idea was simple: send an odd/even pair of FT8 CQs on one antenna, switch to the other, then reverse the order. The intended sequence was A–B–B–A, to help balance out propagation improving or declining during the test.

I used M7XOM for the vertical and M7XOM/P for the reference to distinguish the reports. The suffix was a bookkeeping device here, not a change of location. I exported the PSK Reporter data and used ChatGPT to match receiving stations and analyse the reports of my transmissions.

The relevant activity was on 8 October 2026, around 13:38–13:42 UTC. Reports from other bands and earlier operating were left out.

A surprisingly even contest

404 distinct receivers across the two antennas; 278 had usable SNR reports for both.

Measure	20 m vertical EFHW	80 m reference EFHW
Unique reporting receivers	358	325
Receivers reporting only this antenna	79	46
Farthest reporting receiver	ZL3DMH, 19,017 km	ZL3DMH, 19,017 km

At the 278 common receivers with usable SNR, the vertical’s median advantage was 0 dB. Its mean advantage was +0.01 dB—which is about as close to a draw as I could ask for, and certainly not meaningful to two decimal places!

The vertical was stronger at 119 receivers, equal at 29, and weaker at 130.

Positive differences favour the vertical. In the scatter plot, points above the diagonal favour it.

The middle half of the differences sat between −3 and +3 dB. That spread is interesting: a near-zero average doesn’t mean the antennas were identical on every path. Some stations preferred one, some the other. Fading and changes in receiver noise will also contribute.

Both antennas reached New Zealand’s ZL3DMH: −13 dB on the vertical and −12 dB on the reference. Not a bad afternoon for some wire in the garden.

Where did the vertical do better?

North is at the top; radius is distance. Colour shows the vertical’s SNR advantage or disadvantage.

There was a hint of a southern advantage. The median difference was +3 dB to the south, compared with +1 dB west and −1 dB north and east. EA8-DF4UE was a striking example: +2 dB on the vertical against −7 dB on the reference.

Before anyone mistakes that polar plot for a radiation pattern: it isn’t one. It shows where the reporting receivers were, and what they reported. The sample was heavily weighted towards the east, with 175 paired receivers there and only 23 in the southern sector. These are geographical short-path bearings, not proof of the route the signals actually took.

Was there a DX advantage?

Paired signal differences on the left; unique reporting receivers on the right.

There was no consistent increase in the vertical’s advantage with distance. The median differences were −1 dB below 1,000 km, zero from 1,000–3,000 km, +1 dB from 3,000–6,000 km and −1 dB beyond 6,000 km.

That doesn’t rule out an advantage on particular DX paths. It simply wasn’t a general feature of this short sample.

The reporting wrinkle

The vertical appeared at 33 more receivers, roughly 10% more than the reference. Tempting as it is, I can’t translate that directly into “10% better antenna”.

Almost every receiver appeared only once per antenna in the exports. Although the timestamps show an early and a late vertical window around the reference test, they don’t give us a complete before-and-after SNR series at each receiver. Those separated windows could contribute to the extra coverage.

There’s another limitation: paired SNR comparisons necessarily leave out stations that heard only one antenna. The signal-strength and coverage figures therefore tell different parts of the story.

Bonus round: the 15 m flowerpot

Separate test, separate band: the 15 m flowerpot versus the 80 m EFHW reference, both transmitting on 15 m.

Earlier that afternoon, the flowerpot appeared at 131 receivers, against 157 for the reference. Across 100 common receivers with SNR, its median difference was −1 dB and its mean was effectively zero.

Again, broadly similar signal strength where both were heard. This time, the reference had the higher reporting count.

What have I learned?

My reference antenna is putting up a respectable fight. The vertical and flowerpot are both competitive, but these tests haven’t produced a universal winner.

The next step is repetition: different times, more comparisons, and looking for advantages on particular paths that turn up consistently. I also want to preserve separate test-block reports where possible, so the before-and-after comparison survives the export.

For now, I have some evidence to accompany my impressions—and some lovely graphs. Not a bad return on a few CQs.

---

Data: my PSK Reporter ADIF exports from 8 October 2026. Antennas were assigned from the filenames. Missing SNR values were excluded; the one repeated receiver in the 20 m vertical data was represented by its median SNR. These results describe the complete installed antenna systems during this test, not calibrated antenna gain. Analysis and charts prepared by M7XOM.