# Inconsistent diode and MOSFET power rating

I’m trying to design a power supply for an electromagnet. For this, I’m using a Buck topologie in which the electromagnet is the coil. The voltage comes from a 400V PFC. Required current is 8 Amps (mean and rms).

So, I’m looking for a freewheeling diode rated 600V, 10A. I found such as VS-15EWX06FNTR-M3 (a bit overkill but OK).

Nonetheless, if I do the math about conduction losses :

Pcond = Vforw * Iaver = 1.5 * 8 = 12 W

Now, I’m not familiar with thermal problems but I’m not sure a DPAK package can endure that. So I checked the conductivity :

And it's about .... 70°C/W for junction to ambient ! That would raise the temperature over 700°C ..

So my question is : how can it be rated 15A ?! I understand that those figures are given for particular conditions like a case maintained at 25°C. But even with a perfect cooler, it seems impossible to me to dissipate like 22W in a diode with this package (or even more for higher ratings)

On the same hand, MOSFET rated 100Amps can have a RDSon about 0.1 Ohm which would lead to 1kW of dissipated power !!

Bob

• case directly to ambient is not the only cooling mechanism. Useful search term : heatsinks.
– user16324
Apr 4, 2022 at 15:53
• That was what I meant with "cooler". But you might be right. It's just that the amp rating is so useless then. With the appropriate heatsink you are just limited by the junction to case ...
– Bob
Apr 4, 2022 at 16:10

Your flyback diode will not be dissipating 12 watts the entire time it is on - consider the duty cycle of your buck converter. Ignoring that:

At 70 degrees C per watt, the diode will certainly not survive a 12-watt dissipation on its own. However, note the junction-to-case resistance: 1.8 degrees per watt. That is about 21 degrees at 12 watts. That is quite manageable. This is what you should get with a perfect heatsink attached to the case, which reduces the case-to-ambient thermal resistance to zero.

Noting that the maximum junction temperature is 175 degrees and that the constant-current power dissipation decreases with temperature, you have a lot of room for your heatsink to not be perfect.

Real heatsinks consist of large areas of copper attached to the diode on the PCB, or large finned blocks of aluminium attached on top. When those are not enough, a fan may also be added. You can probably get by without a fan, but test it and see.

Also consider investigating synchronous buck converter designs - using an actively controlled MOSFET, you should be able to achieve lower power dissipation than by using a diode.

This question may already be answered elsewhere, but Absolute Maximum refers to critical damage like fusing open or short or catastrophic voltage breakdown. So you don't design to just meet this steady-state but for any state including a 1us pulse. You always choose the Recommended maximum for any design and factor with some margin.

Consider a spec for your environmental stress from room ambient, box ambient and other parts self-heating and this part self-heating with heatsink design and validation with margins.

note: D is for dutycycle

https://www.vishay.com/docs/93239/vs-15ewx06fn-m3.pdf

• Ok, but then why do they even bother to say something on the datasheet ? I'm sure those devices can handle like 1 terra amps for 1 femtosecond... The number they give is of no use doesn't it ? But thank you for your quick (!) answer !
– Bob
Apr 4, 2022 at 16:08
• If you understand the voltage and Rs needed to conduct that in a few nH , you would know that is not even possible or true @Bob with the shortest time constant possible. The stored energy can also be destructive. Apr 4, 2022 at 16:59