Power supply efficiency
Work out the efficiency of a supply from its input voltage and current and its output voltage and current.
History
Formula
Loss: Ploss = Pin − Pout [W]
Design notes
The efficiency of a supply is simply the ratio of output power to input power. The lower it is, the more of the difference comes out as heat and the more thermal design you need.
Rough figures:
• A good DC-DC converter: 85–95%
• A linear regulator (LDO): the theoretical ceiling is Vout/Vin
• So a 5 V → 3.3 V LDO cannot do better than 66%
The heat to get rid of is input power minus output power, and that is the number your thermal design has to work from.
When you need this
For bench evaluation of DC/DC converters and linear regulators — measure input and output, get efficiency, and see how far the real board departs from the datasheet efficiency curve.
From efficiency to loss
Efficiency η = Pout / Pin, and the loss is Pin − Pout, which becomes heat.
For design purposes the absolute loss matters more than the percentage. 90% efficiency at 50W output still dissipates 5.6W, and finding somewhere for that heat is the real problem. At 1W output, even 80% efficiency is only 0.25W and thermal design barely enters into it.
Typical efficiency by topology
| Topology | Efficiency | Notes |
|---|---|---|
| Linear regulator | Vout/Vin | Fixed by the voltage ratio. 5V→3.3V is 66% |
| Non-synchronous buck | 80 – 90 % | Diode forward drop dominates; poor at low Vout |
| Synchronous buck | 90 – 96 % | The mainstream choice; many hold efficiency at light load |
| Boost | 85 – 93 % | Input current is higher, so input-side losses matter |
| Isolated flyback | 75 – 88 % | Transformer and snubber losses add |
The key point is that linear regulator efficiency is Vout/Vin and nothing else. Dropping 12V to 3.3V is 27.5% efficient — the other 72.5% is heat. Where the differential is large, use a switching converter.
Where efficiency is lost
- Light load. The controller's own quiescent current becomes proportionally significant. If standby efficiency matters, choose an IC that skips pulses at light load (PFM or burst mode).
- Large input-to-output ratio. A buck runs at very small duty and switching losses grow as a fraction. Stepping 24V to 1.2V can be more efficient in two stages.
- High switching frequency. Raising it shrinks the inductor and capacitors, but switching and gate-drive losses grow proportionally.
- High temperature. MOSFET Rds(on) rises with temperature — commonly 1.5–1.8 times at 125°C — and conduction loss rises with it.
Measuring properly
- Measure voltage at the converter terminals. Including the drop along wiring and probes makes efficiency look worse than it is. Four-wire measurement is ideal.
- Input current carries ripple. Use a meter that reports a true average; peak or RMS readings understate efficiency.
- Let it reach thermal equilibrium. Efficiency can differ by a few percent between switch-on and ten minutes later.
Frequently asked questions
Can I improve linear regulator efficiency?
How do I get temperature rise from loss?
My measurement came out above 100%.
Standards and references
- IEC 62301 — Standby power and efficiency measurement.
- IEC 61204-3 — Low-voltage DC power supplies: EMC and performance.
Last updated: 2026-08-29