Bay note 02 — Core geometry before core size
EE, ETD, PQ and RM are not interchangeable blanks. Geometry decides leakage, window and heat — with an area-product worked example for a 65 W flyback.
Two cores with the same Ae can fail the same converter for opposite reasons. One leaks flux into the sniffer probe; the other cooks the centre leg. Geometry first, millimetres second.
The shapes we actually wind
| Shape | Window | Leakage | Heat | Typical bay use |
|---|---|---|---|---|
| EE / EF | Generous | Moderate | Good side faces | General SMPS, cost-sensitive |
| ETD | Round centre | Lower than EE | Excellent | Higher power, lower AC resistance |
| PQ | Compact | Low | Hot centre if choked | Dense power, low profile |
| RM | Round, shielded | Low EMI | Fair | Noise-sensitive boards |
Worked strip — area product for 65 W / 65 kHz
Practical area-product form for a ferrite flyback:
Ap ≈ Pout / (Kt × ΔB × fsw × J)
Using bay constants for MnZn power ferrite at 65 kHz (loss-limited ΔB, not raw Bsat):
- ΔB ≈ 0.2 T (about 2000 gauss) — typical when core loss, not saturation, sets the limit
- J ≈ 4 A/mm²
- Kt ≈ 0.00025 (multi-winding flyback space-factor style constant)
- Pout = 65 W
Ap ≈ 65 / (0.00025 × 0.2 × 65000 × 4) ≈ 65 / 13 ≈ 5.0 cm⁴
That lands in EE30 / ETD29 / PQ32 territory — not EE16. An undersized core that "fits the board" will run hot or saturate when Vin dips.
Window check (the second gate)
After Ap, check copper. Window utilization Ku ≈ 0.3–0.4 for tape-heavy isolation.
Primary copper rough: Acu ≈ Irms / J. For Ipk 2.7 A and D 0.45, Irms ≈ Ipk × √(D/3) ≈ 1.0 A → Acu ≈ 0.25 mm². Multiply by turns and Ku — if the window cannot hold it with creepage tape, step up a core even if Ae looked fine.
What to put on the print
- Preferred family (EE / ETD / PQ / RM) and max height
- fsw and topology
- Isolation tape schedule or reinforced requirement
- Whether low leakage or low height wins the trade
Browse ferrite transformers for standard envelopes we already tool.