Why Toroid Choice Is More Than Just Permeability
When you hear “Choose a toroid by its permeability (µ),” it sounds simple. Too simple. In real HF transformer design for baluns, ununs, and matching networks, permeability is just one piece of a larger puzzle.
Frequency-Dependent Losses
Ferrite mixes don’t have a single µ value — they have complex permeability: - µ′ gives usable inductance - µ″ represents losses If µ″ dominates, the core turns RF into heat instead of efficiently transferring energy. Mix 77 works fine below 5 MHz but runs hot on 20 m.
Resonance and Usable Range
Every mix has a frequency range where it performs best: - High-µ mixes (75, 77): excellent for 160/80 m, lossy above ~5 MHz. - Mid-µ mixes (43): the true TX workhorse for 4:1, 9:1, and 49:1 builds, efficient across 3–20 MHz. On 49:1 builds, Mix 43 is excellent up to 20 MHz, but above that it behaves more like a capacitive filter than a true broadband transformer — still “working” on paper but with reduced efficiency. - Low-µ mixes (52, 61): suited for VHF, but misleading when applied to broadband HF.
Mix 52 looks attractive because datasheets show low loss above 20 MHz. In practice:
- Its lower µ means you need more turns for enough inductance at 80/40 m → more capacitance and leakage inductance.
- The “flat” sweep into a dummy load doesn’t translate to efficiency on a real EFHW — the high-Z antenna load dominates.
- At QRO, Mix 52 saturates earlier than expected with multi-turn windings, while capacitance pushes the effective SRF down.
Geometry and Power Handling
A core’s physical size directly affects performance: - Larger cross-section → lower flux density → better QRO handling - Window size → how many turns can fit without excess leakage - Mean path length → effective inductance Stacking cores is often more effective than using a single higher-µ core.
Saturation and AL Value
Once flux density exceeds Bsat, the core saturates and performance collapses. Lower-µ mixes usually allow higher current before saturation. The AL value (inductance per turn²) is useful, but only in low-power conditions — it doesn’t tell the whole story under real RF drive.
Thermal Limits
As ferrite warms up, losses climb. Near the Curie temperature, a core rapidly loses effectiveness. Heat sinking, stacking, and mounting all matter when designing for kilowatt levels.
Winding Layout
The winding itself often determines bandwidth more than the core: - Interwinding capacitance limits high-frequency performance - Leakage inductance causes imbalance at band edges - Bifilar/trifilar winding helps broadbanding — but adds capacitance
Don’t Add Parallel Capacitors
Some builders try to “flatten” response by adding capacitors across windings. In reality, this beats the purpose of using a wideband ferrite transformer. Instead of a true broadband match, you get resonant peaking and narrow-band artifacts.
Measuring Primary Inductance Properly
Primary inductance (Lm) should be measured at 100 kHz, not at HF:
- At 100 kHz, you’re below winding self-resonance, so you measure true magnetizing inductance.
- An “unloaded” transformer is never really unloaded: leakage and parasitics distort high-frequency readings.
- Usable frequency range depends more on the ferrite mix than on the apparent HF inductance without load.
The Takeaway
Permeability is necessary — but never sufficient. Real design balances: - Core mix choice (loss vs. frequency range) - Geometry (flux handling vs. winding space) - Winding technique (capacitance vs. coupling) - Thermal limits (QRO survivability) - Proper 100 kHz inductance checks
Mix-at-a-Glance: Typical Uses
| Mix | µ (approx) | Usable Range | Typical HF Use | Notes |
|---|---|---|---|---|
| 31 | 150 | 1–30 MHz | Line isolators, RX baluns/ununs | Too lossy for TX transformers; excellent for common-mode suppression |
| 43 | 850 | 3–30 MHz | TX baluns/ununs (4:1, 9:1, 49:1) | The workhorse for HF TX; on 49:1 builds reliable up to 20 MHz, then acts more like a capacitive filter |
| 52 | 250 | 10–100 MHz | VHF transformers | Not a good choice for HF EFHW; more turns needed, capacitance rises, efficiency drops |
| 61 | 125 | 30–200 MHz | VHF/UHF chokes, narrow-band transformers | Low µ; efficient at VHF/UHF but poor at HF |
| 75 | 4700 | 0.1–5 MHz | Low-frequency RX loops | RX only; too lossy for TX, low Curie temp |
| 77 | 2000 | 0.1–5 MHz | 160 m and 80 m TX baluns/ununs | Suited for low-HF TX when stacked; rolls off above ~5 MHz |
Mix 43 vs Mix 52 for 49:1 EFHW (Reality Check)
| Property | Mix 43 | Mix 52 |
|---|---|---|
| Permeability (µ) | ~850 | ~250 |
| Inductance per turn² (AL) | High → fewer turns needed | Low → many turns needed |
| Capacitance impact | Moderate (single-layer windings feasible) | High (extra turns add parasitic capacitance) |
| Efficiency up to 20 MHz | Excellent (workhorse bands: 40/30/20 m) | Compromised by excess turns |
| Behavior above 20 MHz | Acts more like a capacitive filter than a true transformer | Looks flat on a VNA, but not efficient with real EFHW load |
| Thermal margin at QRO | Proven when cores are stacked (2–3× 240-size) | Earlier saturation and heating under multi-turn load |
| Verdict | Best choice for HF 49:1 (up to 20 MHz) | A trap: marketing flatness, poor real efficiency |
Recommended Mix by Transformer Type
| Transformer Type | Recommended Mix | Notes |
|---|---|---|
| 4:1 current balun (TX) | 43 | Broadband 3–30 MHz, stack for QRO |
| 9:1 unun (RX/TX) | 31 or 43 | 31 for low-noise RX, 43 for TX |
| 49:1 EFHW (TX, 40–20 m) | 43 | Excellent up to 20 MHz, efficient on 40/30/20 m, not useful for 17/15/12 and 10m |
| 49:1 EFHW (TX, 160/80 m) | 77 (stacked) | Band-optimized, lossy above 5 MHz |
| Chokes / Line Isolators | 31 or 43 or 52 | Frequency dependant CM suppression |
Looking for the right ferrite cores? Explore our selection here: Ferrite materials and toroids at RF.Guru.
Mini-FAQ
- Is permeability the most important factor? — No. Losses, winding layout, and saturation limits are often more critical.
- Which mixes are best for wideband HF? — For TX: Mix 43. For RX or chokes: Mix 31 or 43. Avoid Mix 52 for HF EFHW 49:1 builds — it looks good on paper but adds turns, capacitance, and loss.
- Why not add capacitors to flatten response? — Because it ruins broadband behavior, replacing it with resonant peaking.
- Why measure inductance at 100 kHz? — At HF the transformer is never truly unloaded. 100 kHz gives the real magnetizing inductance without parasitic capacitance effects.
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.