Alright, continuing with the debugging of my new board tonight I found another issue worse than the one before: My switching MOSFET is getting so hot that I'm getting molten solder.

I replaced the MOSFET with a short circuit just to see if the components sorounding the MOS were the cause of the problem (mostly a big Inductor I have right next to it) but the copper reaches a temperature of 110°F, not enough to melt solder.

This is my schematic:

enter image description here

As you can see, my VGS is around -11V and I'm running 3A through it using an Electronic Load. Based on the datasheet of my NTD20P06L-D this thing can do up to 15.5A.

Any ideas of what might be going on?

The only thing I could think of was that the heat on the copper could be increasing RDS(on) but at that particular low current it shouldn't be so serious.

I am using the TO-252-3, DPak (2 Leads + Tab) package.

  • \$\begingroup\$ I separated the PAD from the board and connected it via 18 AWG wire and the melting occurred almost immediately, so it is definitely the part getting super hot. \$\endgroup\$ Commented Oct 20, 2012 at 2:01
  • \$\begingroup\$ Have a look at figure 11 (pg 5 in that datasheet). For an 11 V Vgs, the maximum dc current is more like 5 A than 15. You're still under that, but it's not surprising you need a heat sink. Also notice notes 1 and 2 on pg 1 of the datasheet. The drain current ratings assume a 1" square pad on your pcb. If you are just using the minimum pad, you should de-rate by about 40% according to the thermal resistance values. Also, the 15 A rating is an absolute maximum, not a recommended operating condition. If you want your circuit to last, you should de-rate from this to begin with. \$\endgroup\$
    – The Photon
    Commented Oct 20, 2012 at 3:38
  • \$\begingroup\$ Is this PWM switched or essentially low speed on/off? What switching speed. Gate is driven off by 100l pullup and will a long time to turn off. If occasional that's fine enough. If PWM then you need a gate driver. Very simple driver cct available if needed. \$\endgroup\$
    – Russell McMahon
    Commented Oct 20, 2012 at 4:12
  • \$\begingroup\$ This is steady state. It remains always on. \$\endgroup\$ Commented Oct 20, 2012 at 4:32
  • \$\begingroup\$ Short the collector and emitter of your transistor, and see what happens. Also (in that state) measure the gate-drain voltage and the source-drain voltage of your FET. \$\endgroup\$ Commented Oct 21, 2012 at 19:31

1 Answer 1


It sounds like you need a heatsink.

Another thing to check is your gate voltage when you have the FET switched on. If your Gate-Drain voltage is too high, the MOSFET may not be properly biased on, a situation which would generate a lot of heat.

Also, how often is this switching? Is it a steady-state thing, or is this part of a switching supply? If it's a switching supply, you will also need to look at the rate at which the system switches.

Anyways, assuming you have everything biased properly (probably a safe assumption, but measure it anyways):

RDS(on) = 130mΩ @ G-D voltage of 5V

So, with 130 mΩ in series with 3A:

$$V = 0.130 * 3$$ $$V = 0.39$$ $$Power = V * A$$ $$Power = 0.39 * 3$$ $$Power = 1.17W$$

So you're going to be dissipating 1.17W of power in the MOSFET, in the best-case situation.
That will get very toasty without a heatsink. If you're just running this as a bare TO-220 device, it getting extremely hot isn't too suprising.

So, assuming we have a TO-220 in free-air:

TO-220 junction-to-air thermal to ambient equals 62.5 degree per watt.

(From here)


$$Δ°C = 62.5 * 1.17$$

$$Δ°C = 73.125$$ $$Device Temperature °C = 73.125 + Ambient$$ $$Device Temperature °C = 98.125$$

So assuming ideal thermal dissipation on a bare TO-220, it's still going to easily reach ~100°C.
Any environmental factors that further reduce the device's cooling will make it worse.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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