Everyone publishes 4:1, 9:1 and 49:1. But the tap on an autotransformer can sit on any turn, so the ratios you can actually build are far denser than that — 9:22 turns lands on 299 ohms, which no reference tabulates. Tell it what you are matching and it works out the turns, whether the core has enough reactance on your lowest band, and what to wind a transmission-line version with.
Turns to wind for the ratio you actually need.
Everyone publishes 4:1, 9:1 and 49:1. But the tap on an autotransformer can sit on any turn, so the ratios you can actually build are far denser than the standard list — 9:22 turns lands on 299 ohms from 50, which no reference tabulates.
A transformer only transforms while its own winding reactance is high compared to the source. The published rule is at least 5×, ideally 10×, at the lowest frequency you use it on.
This is why the classic 2:14 on an FT240-43 disappoints down low. Run the numbers and it comes to 1.0× on 160m, 2.0× on 80m, 3.9× on 40m — only reaching a comfortable 7.7× by 20m. It is not that the ratio is wrong; it is that two primary turns cannot hold off 50 ohms at 3.5 MHz. Stack a second core, or use more turns at the same ratio (4:28 rather than 2:14).
Everything this tool winds is a flux-coupled autotransformer — impedance ratio is the turns ratio squared.
A Guanella or Ruthroff transformer, which is what most 1:1 current chokes and commercial 4:1 baluns are, works by transmission-line action instead. Turns barely describe it. What matters is the characteristic impedance of the line you wind with, which has to be the geometric mean of the two ends — 100 Ω for a 4:1 from 50, 150 Ω for a 9:1, 350 Ω for a 49:1. Wind one with whatever bifilar wire is to hand and it will be wrong however carefully you count the turns. The tool prints the figure you need.
For a balanced load a Guanella current balun is usually the better answer, because it forces equal and opposite currents rather than merely presenting a balanced voltage.
A transformer changes the resistive ratio and nothing else. If your analyzer reads 450 + j80, a 9:1 gets the 450 right and leaves the reactance exactly where it was. For that you still need the Matching Network or a tuner.
Count turns as passes through the middle of the core, not as loops visible from outside. It is the most common way a wound ratio comes out wrong.
Published ratings for the same core disagree by a factor of four — the same FT240-43 gets quoted anywhere from 250 W to 1 kW — because “rating” hides three separate questions: at what duty cycle, into how good a match, and how hot are you willing to let it get.
So the figure here is built from one documented product spec and scaled by core volume, which is what sets both the thermal mass and the surface it sheds heat from. The anchor is 200 W SSB on a single FT240-43, deliberately below the ARRL End-Fed Half-Wave kit that uses that core and is rated 250 W PEP. Scaling that by volume then independently lands on the commonly published 60 W SSB for an FT140-43 — two unrelated numbers agreeing is the reason to trust the shape of it.
All figures are SSB voice. Speech spends most of its time near zero, so the average heating is a fraction of the peak. Anything that keys a constant carrier does not:
These are duty-cycle ratios, not safety margins on top of one. And all of them assume the transformer is running into something close to the match it was wound for — one absorbing a bad mismatch gets hot regardless of what any label claims.
