Choosing a Ferrite Toroid: Permeability Is Only the Beginning
Choosing a Ferrite Toroid: Permeability Is Only the Beginning
The familiar permeability figure is useful, but it cannot tell you whether a ferrite transformer or choke will cover the required band, remain efficient, preserve balance or survive the intended power and duty cycle.
“Which mix?” is not the first design question. Start with the job: energy transfer, impedance transformation, common-mode suppression, pulse coupling or energy storage. Then define frequency, impedance, waveform, voltage, current, duty cycle, temperature and acceptable loss. Only then does a material and core geometry become a defensible choice.
Joeri’s short version: I do not choose a toroid from one permeability number or a colour chart. I choose a material and geometry for a declared circuit, wind it as a real component, measure it across the intended loads and frequencies, and then drive it until its electrical and thermal limits are visible.
Permeability Is a Family of Conditions
Initial permeability is a small-signal property measured under stated conditions. It helps estimate inductance, but a working ferrite sees frequency, magnetic-field amplitude, temperature and sometimes DC bias. Those conditions change the effective behaviour. Manufacturer data therefore need to be read with their test frequency, flux level, core shape and temperature attached.
At RF, permeability is complex. Its real component, commonly written μ′, describes the field-storage contribution; the imaginary component, μ″, represents magnetic loss in the selected equivalent model. Both vary with frequency. Winding resistance and dielectric, proximity and radiation losses add to the complete component, so even the material curves are not a transformer-efficiency plot.
μ = μ′ − jμ″
Z = R + jX
The sign convention can vary with the assumed time dependence, but the engineering point does not: impedance has both resistance and reactance, and the balance between them changes across frequency. A single scalar permeability cannot describe that curve.
A Choke and a Transformer Reward Different Behaviour
A common-mode choke is intended to oppose unwanted current. A substantial resistive component of common-mode impedance can be useful because it damps resonances and dissipates unwanted common-mode energy. That is not a licence to ignore heating, voltage or differential insertion loss, but loss in the unwanted-current path can serve the objective.
A transmission-line transformer or impedance transformer has another job: transfer wanted energy between ports. Magnetic and winding loss then reduce efficiency and create heat. A material that produces an excellent suppression impedance in a choke fixture is not automatically the best material for a power-transfer transformer, even when both devices occupy the same frequency range.
This distinction is why universal mix tables are dangerous. “Good at HF” is incomplete until the table says for which topology, impedance, core size, winding, power, waveform, temperature and acceptance limit.
Flux, Frequency and Core Area Travel Together
Voltage applied to a winding produces changing magnetic flux. For a given waveform, fewer turns, lower frequency or smaller effective core area raises peak flux density. Increasing turns can reduce flux density, but it also increases conductor length, leakage inductance and parasitic capacitance. There is no free turn.
Saturation is an important boundary, but it is not the only one and often is not the first one reached in a broadband RF transformer. Core loss can raise temperature long before a dramatic saturation event. Winding loss, high electric field between turns, insulation breakdown or an unwanted resonance may set the practical limit first.
Core geometry matters as much as the material name. Effective area influences flux density; magnetic path length and permeability influence inductance; core volume and surface area affect loss density and cooling; the window limits winding construction. Stacking cores changes several of those quantities, but it does not supply a universal power multiplier. The finished assembly must still be tested.
The Inductance Factor Is a Starting Point
The manufacturer’s AL value relates inductance to turns squared for a stated core and test condition:
L ≈ ALN²
It is valuable for a first winding estimate. It also carries production tolerance and does not include every high-frequency parasitic or powered condition. There is no universally correct measurement frequency such as 100 kHz for every ferrite transformer. Use a frequency low enough that the selected equivalent circuit is meaningful, but within the instrument, fixture and material data’s useful range—and record test level and temperature.
Then sweep the actual wound component across its working band. A low-frequency inductance result cannot reveal high-frequency leakage, distributed capacitance, fixture coupling, winding imbalance or transmission loss.
The Winding Can Set the Band Edges
At the low-frequency edge, insufficient magnetising impedance can load the circuit and increase current. At the high-frequency edge, leakage inductance, interwinding capacitance, turn-to-turn capacitance, conductor length and physical coupling can dominate. Fair-Rite’s broadband-transformer note explicitly separates these low-, mid- and high-frequency regions in the transformer model.
Bifilar, trifilar, coaxial and separated windings create different coupling, impedance and electric-field distributions. Tight coupling may reduce leakage while increasing capacitance. Greater spacing may reduce capacitance while increasing leakage. The best layout depends on the port impedances, transformation ratio, balance requirement and band.
A compensation capacitor is therefore neither magic nor automatically wrong. It may intentionally compensate a known parasitic or shape the response of a complete network. It may also create resonant peaking, higher circulating current or voltage, and poor behaviour on an unexpected load. Accept it only after measuring the whole device over frequency, load, power, temperature and component tolerance.
Curie Temperature Is Not the Operating Rating
Permeability changes with temperature and eventually collapses around the Curie region. That material transition is not a recommended enclosure temperature. Insulation, wire coating, capacitors, potting, connectors, plastics and solder may have lower limits, while repeated thermal cycling can create damage well below Curie temperature.
A useful thermal limit states ambient temperature, airflow, enclosure, mounting, duty cycle, waveform, mismatch and the allowed hotspot temperature. Temperature should be followed until equilibrium—or until a conservative stop condition is reached—and the device should be remeasured afterwards. A short carrier test does not establish a continuous-duty rating.
Read the Datasheet in the Right Direction
| Datasheet item | What it helps answer | What it cannot prove alone |
|---|---|---|
| Initial or effective permeability | A starting estimate for small-signal inductance under stated conditions | Broadband efficiency, common-mode suppression or power rating |
| Complex permeability versus frequency | How magnetic storage and loss evolve with frequency | The complete wound component’s impedance or insertion loss |
| AL value and tolerance | Approximate turns needed for a target low-level inductance | High-frequency parasitics or behaviour under RF drive |
| Flux and loss curves | Material behaviour at the specified waveform, frequency, flux and temperature | A different waveform, geometry, cooling arrangement or load |
| Curie temperature | The material’s magnetic transition region | A safe component or enclosure operating temperature |
| Core dimensions | Area, path, volume and winding-window inputs | Finished winding balance, insulation, cooling or connector limits |
Measure the Job You Expect the Core to Do
- Declare the circuit. Record topology, turns, winding placement, port impedances, intended balance and the reference planes.
- Start at low level. Measure complex impedance or S-parameters over a range wider than the intended band, with the fixture calibrated or compensated at the appropriate plane.
- Use representative loads. A nominal resistor is useful, but it does not replace the range of complex impedances the device will encounter.
- Separate wanted and unwanted modes. Measure transformer transfer and balance independently from common-mode impedance and current suppression.
- Close the loss budget. Back-to-back testing can help estimate insertion loss, provided two devices, fixture loss and mismatch are treated correctly. Calorimetry or careful power accounting can provide another check.
- Increase stress deliberately. Test power, waveform, duty cycle, mismatch and ambient condition in controlled steps while observing voltage, current and temperature.
- Repeat after heating. Recheck electrical behaviour after thermal equilibrium and after cooling to expose reversible drift and permanent damage.
Keysight’s impedance guidance emphasizes that the selected technique, fixture, cabling, signal level, frequency and error compensation all affect the result. Ferroxcube’s handbook likewise shows that permeability and loss depend on field, frequency, temperature and effective core dimensions. The data are not obstacles; they are the map.
Primary and Authoritative References
- Fair-Rite — Use of Ferrites in Broadband Transformers
- Fair-Rite — Technical flyers, material papers and application guidance
- Ferroxcube — Soft Ferrites and Accessories Data Handbook
- Keysight — Materials Testing: Magnetic Material Measurement
- Keysight — Impedance Measurement Handbook
Joeri’s Bottom Line
A mix number is not a design and a clean VNA trace into one resistor is not a power rating. Start with the current path and the job, use the manufacturer’s conditioned material data, design the winding and core together, and test the finished device in the circuit it must survive.
Permeability can help you choose the first winding. Loss, flux, parasitics, temperature and measurement decide whether it becomes the right transformer.
Mini-FAQ
- Is the highest-permeability ferrite the best choice? No. Permeability is frequency-, level- and temperature-dependent, while loss, flux, geometry, winding parasitics and the circuit’s job also matter.
- Can one mix table specify the best HF transformer? No. A useful recommendation must include topology, frequency, impedances, core size, winding, waveform, power, duty cycle, temperature and acceptance limits.
- Is ferrite loss always undesirable? It is usually unwanted in a power-transfer transformer. A resistive common-mode impedance can help a choke damp unwanted current, provided heating and differential performance remain safe.
- Should primary inductance always be measured at 100 kHz? No. Select a test frequency, level and equivalent model appropriate to the material, winding and instrument, then sweep the finished component across its operating band.
- Do compensation capacitors ruin a broadband transformer? Not automatically. They can compensate known parasitics, but the complete network must be checked for peaking, loss, stress, load sensitivity and tolerance.
- Does staying below Curie temperature make a transformer safe? No. Core loss, winding insulation, capacitors, connectors, plastics and enclosure temperature can impose much lower limits.