How to Estimate Transformer Temperature Rise Before You Place the Order

Why Temperature Rise Is the Number Buyers Forget

Most transformer specifications state inductance, turns ratio, hipot voltage and dimensions. Very few state the temperature rise the part is allowed to develop in the application. That single omission is behind a large share of field failures and of “the sample worked on the bench” arguments, because a magnetic component that runs at 40 K rise in free air can run at 70 K once it is inside a sealed enclosure on a hot day.

This note gives a workable pre-order estimate, the assumptions that make it wrong, and what to put into the RFQ so the supplier is accountable for the same number you are.

Step 1: Establish the Losses

Total loss is core loss plus copper loss. Both are available before you order if you know the flux swing and the RMS currents.

Loss terms and how to obtain them
Term	Formula	Where the input comes from
Core loss	P_core = k · f^α · B_pk^β · V_e	Material datasheet (Steinmetz coefficients k, α, β) at your frequency and flux swing
Copper loss	P_cu = I_rms² · R_dc · (1 + F_r)	Winding resistance from the drawing; F_r is the AC resistance factor, typically 1.2–2.5 above 100 kHz

Two practical cautions. First, core loss at the operating flux swing is often 2–4 times the catalogue value quoted at a different B_pk; always interpolate at your actual swing. Second, copper loss at 100 kHz and above is dominated by the AC factor: a winding that measures 0.4 W at DC can dissipate 1 W in service, and that difference decides whether the part passes.

Step 2: Convert Losses to Rise

For a component in still air, a useful first estimate is thermal resistance from the hot spot to ambient of 8–15 K/W for a 30–50 mm potted unit, and 3–8 K/W once it is bolted to a metal chassis or sits in forced air at 1–2 m/s. Rise is then simply loss multiplied by that resistance.

Worked example: 120 W flyback transformer, 100 kHz
Quantity	Value	Note
Core loss	1.6 W	EE33 MnZn, 150 mT swing, 100 kHz
Copper loss (with F_r = 1.8)	1.1 W	Primary plus secondary, RMS values from the schematic
Total loss	2.7 W
R_th, free air	12 K/W	Unpotted, no airflow
Rise, free air	32 K	Acceptable for Class F (155 °C) at 40 °C ambient
R_th, bolted to chassis	5 K/W	Same part, thermal path through the mounting
Rise, chassis	14 K	Same design, four times more thermal margin

The lesson of the table is structural: identical magnetics give a 32 K or a 14 K rise depending only on how they are mounted. If the enclosure, airflow and mounting are not in the specification, the temperature rise is not specified either.

Step 3: Check the Limiting Material, Not the Average

Insulation class is set by the hottest point, which is normally the interior of the winding, not the surface. A part whose surface measures 85 °C in a 40 °C ambient is often 25–35 K hotter internally. Compare the internal hot-spot estimate against the class limit with margin: Class B (130 °C) and Class F (155 °C) are common, Class H (180 °C) costs more in materials and is worth it only where the ambient or the density demands it.

What to Put in the RFQ

Ambient temperature at the part, and whether it is still air, forced air (state the velocity), or conducted through a heatsink.
Maximum permitted temperature rise, in kelvin, measured at the winding by resistance change.
Insulation class required, and the hot-spot limit that follows.
Duty profile: continuous, or a peak rating with a stated duty cycle.
Whether the temperature rise must be verified by test on the first article, and at what load and ambient.