Why Broadband RF Transformers Aren’t “Copy-and-Paste”

And why high-ratio EFHW / F(H)W transformers like 68:1 and 70:1 are especially unforgiving on the top bands.

It’s tempting to treat a broadband RF transformer like a cookbook recipe: pick a core, wind a fixed number of turns, and you’re done. The turns-ratio formula looks clean and reassuring:

The myth: “Same turns = same performance”

A broadband transformer does not behave like an ideal transformer across HF. In practice, it behaves like a frequency-dependent network:

• On the low end, performance depends heavily on having enough magnetizing inductance, so the primary doesn’t look like a reactive load (which shows up as rising SWR and loss).
• On the high end, performance is dominated by parasitics: distributed inter-winding capacitance, leakage inductance, imperfect coupling, lead length/layout effects, and SRF (where it stops being “transformer-like” and starts behaving “LC-like”).

That’s why two transformers that look identical on the bench can behave differently on the air.

Why high-ratio EFHW transformers are harder than the usual ones

High impedance ratios (like 68:1 and 70:1) typically require more turns and more demanding winding geometry. That creates a tradeoff that becomes brutal on the top bands:

• More turns increases inductance (good for low frequencies).
• More turns also increases distributed capacitance and leakage inductance (bad for high frequencies).
• More turns often pulls SRF downward… sometimes into bands you actually want to use (20 m through 10 m).

EFHW feedpoint impedance is not a fixed number. The “right” ratio depends on the wire length, installation height, environment, and where the voltage/current maxima end up along the radiator. A higher ratio isn’t automatically “better.”

Why copy-paste fails in real production

Ferrite cores aren’t perfectly identical

Even from reputable manufacturers, ferrite varies across tolerances and production lots. That includes effective permeability / A_L, loss behavior versus frequency, and dimensional variations that change how the winding actually sits on the core.

One overlooked consequence is that inductance follows:

L = AL × N²

If A_L shifts, inductance shifts even when the turn count is “identical.” That directly affects low-band behavior, current, heating, and where “non-ideal” behavior starts to creep in.

Wire diameter and insulation thickness change the RF outcome

Switch to a new spool or a new batch, and you often get small changes in copper diameter, enamel thickness, and insulation properties. Those differences change:

• how tightly the turns pack,
• turn-to-turn spacing,
• inter-winding capacitance,
• and therefore SRF.

On the top HF bands, SRF movement is one of the fastest ways to turn a “good” transformer into a touchy one.

Winding geometry is part of the circuit

Even with the same core and the same wire, process variation changes the RF “shape”:

• winding tension and consistency,
• whether turns overlap or stay evenly spaced,
• how the winding is distributed around the core,
• how start/end leads are routed and how long they are,
• how tightly the winding couples primary to secondary.

Those are not cosmetic details. They directly change leakage inductance and distributed capacitance, which is exactly what dominates on the top bands.

“One more or one less turn” is sometimes the correct fix

With high-ratio transformers, insisting on a fixed turn count for every unit can create inconsistent outcomes when tolerances stack up. If A_L is slightly higher, one unit may need a small adjustment to land in the same performance window. If A_L is slightly lower (or capacitance ends up higher), another unit may need a different adjustment.

That’s not sloppy work. That’s what repeatable RF behavior looks like when you stop pretending all parts are ideal and identical.

Tolerance stacking shows up first on the top bands

One variation might be small. But production rarely gives you just one:

• core tolerance,
• wire tolerance,
• winding geometry tolerance,
• assembly/layout tolerance.

Stack them and the top bands reveal it immediately: SRF shifts, coupling shifts, return loss changes, and suddenly 10 m / 12 m / 15 m becomes “touchy” while the lower bands look fine.

What consistent transformers require

Repeatable performance comes from verification, not assumptions. That’s why serious broadband transformer work looks like this:

• Controlled build method (repeatable winding pattern, spacing, lead routing, and mechanical assembly).
• Post-build measurement to catch inductance shifts, SRF movement, and high-band return-loss changes before the part ships.
• Reject/adjust policy when the RF behavior lands outside the intended window.

Bottom line

A broadband RF transformer is not a copy-and-paste component. RF performance is shaped by the interaction of core properties, winding geometry, and parasitics. With high-ratio EFHW transformers like 68:1 and 70:1, those interactions become even more critical, and small production variations are enough to shift SRF and change behavior on the top bands.

Turns are a starting point. Repeatable performance requires measurement and quality control.

Interested in more technical content? Subscribe to our updates for deep-dive RF articles and lab notes.

Questions or experiences to share? Feel free to contact RF.Guru via our RF.Guru contact page.