IRF part specs confusion.

IR is dishonest and insane. Their current ratings are ludicrous, and they often add a footnote specifying a lower rating that is "packaged limited."

They apparently measure power dissipation capability with the entire device in some unspecified boiling liquid, no doubt being pulled down with a vacuum pump. That lets them push current through a source lead that would melt in real life.

John

ductID=3DAUIRF3805

=2E

=20

lf

Reminds me of 'Sears' horsepower ratings. They had ridiculously high horsepower ratings on a lot of their electrical powered devices.

It turns out they were fudging the numbers in a very misleading manner.

They'd feed the poor motor with humongous, low resistance cables. The motor shaft would be locked solid. There would be no overcurrent/short circuit protection applied.

Then they'd use the resulting inrush amperage and the PRE-short circuit, line voltage to get input watts. This is the number used to derive the advertised horsepower.

mike

The Id rating for your device is 160A if you can hold the case temperature to 25C. The Id rating is the "Case Limited" value shown in the Continuous Drain current under Absolute Maximum Ratings.

I don't know why they give the "Silicon Limited" Id values.

John S

..

Useful for running 'short term'?

To expand on what everyone else is saying: you should always look at the pertinent parameters (in this case R_DS_ON, maximum junction temperature, the various thermal resistances) and do a thumbnail sketch of your thermal design using _your anticipated_ maximum ambient temperature.

John Larkin got part way there when he said that IR lies: _everyone_ tells a misleading story. _Everyone_ tries to get a design-in from some thoughtless engineer or manager. _Everyone_ has at least one current specification that assumes a case temperature of 25C -- which no real application (unless it's designed to work in a freezer) is going to attain. So you _always_ need to do that step.

..

e.

to

yself

they probably use something like R-11, at atmospheric pressure it boils at 24C, just dump the whole thing in

has all the explanations for how they spec/measure their fets, so at least they are honest about what the numbers mean

-Lasse

Don't ever expect to run a part at the "absolute maximum" rating point, or even at the one-time pulse limit. The full data sheet (not the one-page summary) will have far more details on the real-world operating regime for the part.

Jon

d...

age.

ut

me

d to

myself

Interesting, Thanks. I had no idea about what is nucleate boiling.

Wiki to the rescue.

George H.

Disappointing that the article gets the delta-T values so wrong. A temperature difference of, from the article, 4, 10, or 30 degrees C is a

I'd rather use the SOA and transient thermal curves.

We've been here before. All the FET companies that I know of spec dissipation this way. It's just junction to case thermal resistance and holding the case at that temperature so that at that dissipation, if you could hold it at 25 C, the die would be at it's limit.

We've had several FETs above their rated die temperature limits and seen them live. Can't usually do that for very long. Not that we were trying to do that, except for destructive tests which are also fun.

Reality usually has us have maximum dissipation below 10 watts or so.

Higher max die temperature and lower J-C resistance and you get to play the higher dissipation game. Unrealistic, but a calculated number that should be correct. IR's max die temp is 175 C so it's better than a FET with 150 C max temperature.

As for the the TO220 Leads vs. current rating ? Over 10 years ago we ran 150 Amps through a regular TO220 lead and it was fine. Anything above that and it turns to slag. That's just the lead. Some newer FETs have larger leads and some even extra source leads for lower RDSon.

boB

I solved my problem, tonight. I decided to order 50 of these HEXFETS at the $2.01 marked price. I Didn't need all that ID amp handling, I just wanted something that was useable in many different things how ever, when I hit the "Add to cart" and went to the next step to complete my order, the price then jump to $3.65 for each, I found that distressing and gave me a bad taste. I went over to mouser and decided on this

for $1.65 each.. Mouser didn't magically up the price in the middle of the transaction. These will do fine, even though the Vgs(th) is a little higher than the HEXFET is, around 4.0. BUt that shouldn't pose a problem since the circuit these are going in are driving the gate at 12.

Thanks..

Jamie

Did you look to the SOA? It looks too good to be true. If I didn't misread it, it can be nice for some linear applications.

It's all about keeping it cool.

Jamie

As usual Maynard doesn't give us a part number, or where he was trying to buy this otherwise imaginary part. I'll bet he was looking at the price per 100 or 1K, instead of the quantity he wanted and had a hissy fit when the site gave him the right price.

Jamie, look to various other mosfets (with similar power rating). It is not just keeping it cool. You can destroy other mosfets well below their power rating, even when you do the derating on power dissipation for case temperature correctly.

Yes I am aware of that, I've worked with many of these things over the years, especially HEXFET line which is why I was trying to get those instead. I've been very successful in implementing them into designs with out any issues. I've run into a bad one now and then from bulk packages. Most likely floor sweepings, who knows! :)

I do love it how ever when you don't take care of the ringing and induction in the circuit. I normally place a lot of precautions in the app to avoid such possible issues. I've seen lots of exploded packages due to ignoring such safeguards under heavy loads in various equipment.

But most issues people have with them are just careless design problems.

Jamie