Calculation of FET Temp/Power Maximums

The Temperature Ratings Of Electronic Parts « Electronics ...

" if mounted to a piece of FR4 its only 3W". That's in air. In a vacuum it would be much less.

Hi,

I guess I'm still missing something. The data sheet lists two different PD numbers for the FET, 125W with no footnote - 3W with a footnote that says if mounted to a piece of FR4. I don't understand how these two numbers relate to each other, there is a huge disparity between them. I'm assuming the fact that it is mounted to a piece of FR4 means that it's in air. The case is not soldered to a pad on the board and there would be little to no heat sunk from the part to the board, correct? If I have that much right under what circumstance would the 125W parameter apply?

The link you sent me has lots of good info. However I don't see anything that takes the parameters I mentioned in my original question and uses them to calculate the temp/power maximums. Do you know of any text that would speak to this.

Thanks for the response and the link.

Not sure exactly what you need.

Power Transistors and Heat sinks - Learn About Electronics

Basics of cooling transistors, IGBTs, and power FETs with ...

Many Thanks Lyn,

These links answered all questions.

Welcome.

It's all about providing a low thermal resistance path for the heat to escape. It's analogous to an electrical circuit where temperature relates to voltage, heat flow to current, and thermal resistance to electrical resistance.

125W assumes a heatsink at the specified temperature with enough heat transfer to prevent any rise in temperature. Most times, the 'high' power dissipation is measured at 25C which is crazy low for most applications except active liquid cooling. You have to know your delta-T (or be able to calculate it) across each of the interfaces to be able to calculate the flow of heat (thermal power) with the given thermal impedances.

You can work it from the environment back to device (start with a known temperature) or make some assumptions about power dissipation at the device and calculate the delta-T's out to the environment to find the junction temperature.

You can usually ignore the radiated heat loss as that is typically in the noise unless you design for space, and the convective losses are usually ignored if you are using a heat sink. The FR-4 case (3W) is the maximum convection power dissipation before the part lets out the magic smoke in a still air, 25 C environment.

Radiant is not negligible, even at 10'C differential for building heat loss calculation, it is a little more than convection.

However, since most electrical equipment is enclosed, with ventilating grills, the casing temperature is as hot as the air in the case - so radiant transfer is uncertain.

Makes me wonder, never thought before, how much "heat sink" makers might massage the figures upwards by including radiant losses.

I'm sure heat sink makers do take radiation into account.

For whatever worth this may be, even a properly heat sinked big power semi-conductors, as exprienced can easily convert sprinkled or sprayed water into steam and evaporation during its normal operation! To alleviate such conditions additional exhaust or cooling fans were added to improve air movement inside its enclosure for a more stable operation!

If there is a heatsink in use, the radiated heat loss (of the device) becomes a diminishing small percentage unless there is a very large delta-T. I ignore it for most of the work I do. At worst, it makes the other heat sinks slightly over-sized which in most cases is not always a bad thing.

I'm not ignoring the radiated heat loss of the heat sink, that's a different issue. This question was about calculating the junction temperature and the thermal impedance of the heatsink includes the convection, radiation, and any conducted losses.

"High power dissipation is measured at 25C which is crazy low"

I do agree it is crazy and I fail to understand the rationale behind it.

Why is it crazy?

Room temperature is where most humans are designed to operate.

The equipment would be there too.

The heatsink will never be at that temperature even with fan cooling. Usually they will run too hot to touch. Unless of course you're one of those blessed (or cursed) with 'asbestos skin'.

Another way to look at it is, if the heatsink is at 25C, how do we reject any heat into a room at 25C?

Lyn,

Almost all of the data sheets of devices provide ratings for Tj being at 25DegC which is absurd. Even if you consider Ta (ambient) @ 25DegC, Tj will always be higher for a regular heat sink except in case of liquid cooled plates as mentioned by Sir Robin.

So, shoot me!

Most semiconductor factory tests are done, for speed, with pulses so short there is no temperature rise. Presumably, the factory test bay is at 25'C, since this avoids putting devices in ovens (or fitting them on heatsinks) to do a test.

Like the QWERTY keyboard, it may have been settled by history - maybe the standard USA air-conditioner setting for human in T-shirt & jeans.

Is there a supplier and a telephone number on the datasheet? If so, ringing the supplier will be the quickest route to finding answers to these questions. The technique is immediate, interactive, and one forms working relationships with real people - highly recommended.

The 125W is on an 'infinite' heatsink which is never attainable so you need to know the actual thermal resistance of your heat path to the environment to work back to junction temperature. This assumes you know what power is dissipated in the device which can be very difficult to gauge if current/voltage is switched. A solution if the device has a diode like the source drain parasitic diode in a mosfet is to measure the volt drop across the diode with the FET unpowered at different temperatures. The value will vary according to the 'diode law'. This gives you a calibration curve. Then power the system, let stabilise, switch off and immediately measure the volt drop again. The value can be compared with the calibration curve to get actual temperature. In practice this must be done by ATE to be quick enough. Residual system voltages must be allowed to decay. This dynamic method is how semi manufacturers characterise their devices.