Calculator KB-34
MOSFET power loss calculator
Where power is lost in a switch and what to do about it: use a different transistor or lower the frequency.
Result
In brief: Calculate MOSFET losses: conduction and switching, power dissipation and average gate-driver current, and junction temperature based on thermal resistance. Example: Transistor losses: 1.6 W.
Calculation example
| Drain current | 10 A |
|---|---|
| Rds(on) | 20 mΩ |
| Duty cycle | 50 % |
| Voltage | 24 V |
| tr + tf | 50 ns |
| Frequency | 100 kHz |
| Gate charge Qg | 50 nC |
| Gate voltage | 10 V |
| Junction → ambient | 40 K/W |
| Ambient temperature | 25 °C |
| Transistor losses | 1.6 W |
|---|---|
| Conduction | 1 W |
| Switching | 600 mW |
| Gate driver | 50 mW |
| Average gate current | 5 mA |
| Junction temperature | 89 °C |
How it is calculated
Ps = ½ · U · I · (tr + tf) · f
Pg = Qg · Ug · f
When on, the switch heats up like a resistor. At every transition, it also briefly carries both current and voltage. The first part depends on Rds; the second increases with frequency. The switching formula is simplified and suitable for an estimate.
What to look for
- At high temperatures, Rds is significantly higher than at 25 °C—often 1.5–2×. Check the graph in the datasheet.
- The lowest Rds usually comes with a higher gate charge. For high-frequency switching, the best choice is a compromise.
Frequently asked questions
How do I calculate MOSFET losses?
Conduction: I² · Rds · D. Switching: ½ · U · I · (tr + tf) · f. At 10 A, 20 mΩ, 50% duty cycle, 24 V, 50 ns, and 100 kHz, that works out to 1 W and 0.6 W.
How much current does a gate driver need?
The average current is Qg · f: 50 nC at 100 kHz gives 5 mA. But peak current during switching reaches amps: without a driver, the transistor switches slowly and heats up.
Can I drive a MOSFET directly from an Arduino?
Only use a logic-level transistor with Rds specified at 4.5 V, or at 2.5 V for 3.3 V operation, and at a low switching frequency. A standard 10 V switch won't turn on fully and will overheat.
