Bay note 07 — Copper loss, core loss and the temperature you actually get
A thermal strip that adds winding loss and core loss, then translates watts into rise — so sample approval is not a surprise on the hot plate.
Electrical designs fail quietly as heat. This plate turns two loss numbers into a temperature story the bay can test.
Copper loss
Pcu = Irms² × Rac — at SMPS frequency Rac > Rdc because of skin and proximity.
Rule of thumb for layered windings above about 100 kHz: budget 1.5–3× Rdc unless you have Litz or foil. Measure Rac on the sample when it matters.
Core loss (Steinmetz-shaped intuition)
Pcore ≈ Ve × ke × f^α × B^β — use the material sheet. For quick bay checks at 65–100 kHz and B ≈ 0.2 T on MnZn power grades, expect on the order of 100–400 mW/cm³. Multiply by effective volume.
Worked strip — EE30 flyback sample
- Pcu ≈ 0.8 W (measured Rac at fsw)
- Ve ≈ 4 cm³, pcore ≈ 200 mW/cm³ → Pcore ≈ 0.8 W
- Ptotal ≈ 1.6 W
Natural convection rough: ΔT ≈ P × Rth. For a small transformer on a busy PCB, Rth might be 30–50 K/W without airflow:
ΔT ≈ 1.6 × 40 ≈ 64 K → 25 °C ambient becomes about 90 °C hotspot — inside many Class B budgets, outside if ambient is 60 °C.
What to put on the print
- Max ambient and allowed hotspot / insulation class
- Airflow or sealed enclosure note
- Prefer Litz / foil if fsw and layers demand it
We can sample with a thermocouple map when the print asks for it — contact.