I am analyzing the thermal dissipation of the FDB33N25 MOSFET - datasheet

The procedure I usually follow is as stated:

$$I_{d} = 15~A (requirement)$$ $$R_{ds} = 92~m\Omega$$ $$\Theta_{ja} = 62,5~K/W$$ $$P = I^2 \cdot R_{ds} = 20,7~W$$ $$T = P \cdot \Theta_{ja} = 20,7 \cdot 62,5 = 1293~K$$

The MOSFET can sustain a max of 20 A. So 15 A should be fine. But the temperature generated is mind blowing.

Is this analysis correct? If yes, then I can never achive 15 A without extravagant cooling. Also, will this temp of 1293 K be in the ambient space (theoretically of course)?

  • 1
    \$\begingroup\$ Rthja = 62.5 Celsius/Watt tell us that for every 1 watt of power dissipation in the MOSFET the junction temperature will rise 62.5 celsius above the ambient temperature. And welcome in the real world. mcmanis.com/chuck/robotics/projects/esc2/FET-power.html And Rthja is a thermal resistance between junction and ambient without any additional heat sink \$\endgroup\$ – G36 Mar 12 '17 at 14:00
  • \$\begingroup\$ You used the Rth junction-ambient figure, which is applicable when you don't use any heatsinking at all. The max figures, especially the marketing-department-written ones on the first page of a datasheet, assume the most favourable situation, in this case an unlimited heatsink. \$\endgroup\$ – Wouter van Ooijen Mar 12 '17 at 14:05
  • \$\begingroup\$ @ThePhoton it's actually (i)x(i)xRds. Sorry about the typo \$\endgroup\$ – Board-Man Mar 12 '17 at 16:00

Thermal Resistance, Junction-to-Case, Max. 0.53 °C/W

So, if you could stick a 2°C/W heat sink on your D2PAK, it would dissipate 20W without problems. I suspect you will find this rather difficult though, so here are other solutions:

  • FET with lower RdsON
  • FET with a more heat-sink friendly package
  • Several FETs in parallel

The last one allows both to lower the dissipation, and to spread it between several packages, where it will be easier to dissipate.

Important Note: Since you selected a 250V MOSFET, I'm assuming you need the voltage rating. If this is a low voltage application, you will get much better RdsON with an appropriate choice of lower-voltage MOSFET.


What is your applied gate-drain voltage.

RdsON is a fickle number that has a number of dependent variables. Do not just rely on the front sheet marketing number.. Check the graphs.

Your gate voltage may be insufficient.

And yes.. 20W = LARGE HEATSINK. 62.5 Celcius/Watt is for the device package to air on it's own.

You should also consider looking for a better MOSFET. 92mOhs is not that great these days. It also increases with temperature, so dumping heat is very important.

  • \$\begingroup\$ BTW I was going to show you a better MOSFET example, but you did not list your REAL voltage requirement. \$\endgroup\$ – Trevor_G Mar 12 '17 at 17:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.