How much can TO-220 handle?

[Mercury]

to220 thermal model

Hello,
I'm designing a buck converter at 250kHz. The N-FET needs to shitch at 300V and about 6A at 50% duty cycle. Let's say I take a good N-FET in TO-220 package; 0.2 Rdson. The conduction losses are about 8W. I don't know what the switching losses are, but I can provide 100ns rise/fall times.

My question is, can a transistor in TO-220 handle such power? Anyone has any experiences designing at such power?

Best regards,
George

 

[max0412]

to220 power handle without cooling

It can handle as much power as you want as long as you keep the junction temp below the rated maximum at worse case extremes. This means you need a suitable sized heatsink and maybe forced air (fan) for cooling to insure the junction doesn’t exceed 125 deg C.

If you go through a thermal calculation and find that your heat sink is too large and expensive you have to select another FET with lower RDSON to reduce conduction losses or parallel FETS.

The package cannot dissipate more then about a watt with no heat sink.

Heres some links with details.

http://robots.freehostia.com/Heatsinks/HeatsinksBody.html

http://elearning.algonquincollege.com/coursemat/saurioc/ELE-3/ELE-3-NOTES/HEAT-NOTE.pdf

http://www.fairchildsemi.com/an/AN/AN-1030.pdf

 

[Mercury]

Thanks you for the links! I do understand the thermal model; but I'm having problems with estimating the switching loses. In the datasheet of SPP20N60C3, that I'm planning to use, this chart is given:

What is that E ? The unit is mWs; I've never encountered this before. Any ideas?

Thanks again!

 

[max0412]

I not sure, I’ve never run across that before or never noticed, If I had to guess if you stick with there previous definition of ”E” as energy and units are milliwatt (mWs) you shouldn’t pluralize abbreviation SI, at least that is what I was taught.

Equations for estimating switching losses are in several pdf’s here is one.

**broken link removed**

Sorry if I’m stating the obvious but you know that your mosfets rdson increases with junction temperature? It won’t be 0.19 oHms in your buck more closer to 0.3 ohms.

All your switching loss calculation will be an estimate only but they should only be a small percentage of your conduction losses or you have bigger fish to fry.
You can use a thermocouple to measure the case temp at worst case ambient under your worst case conditions from that you can get the junction temperature, then you can get the actual power dissipation.