Heatsinking high power diodes

Fun! Back in the 1970s they could have been a very bad influence on me. ;)

Cheers

Phil Hobbs

Ohhhhh. That's an adorable pup!

It's never too late.

My first wife, the cajun girl, had a killer Kawasaki 500, 0-60 in under 4 seconds, which actually almost killed her. It was fun to ride behind her and hang on for dear life.

It's too late when the illusion of immortality wears off. I have a Mustang convertible with a stick shift, which is about my speed for excitement at this point. ;)

You must have been close to her size. I tried riding behind a friend once long ago on his Yamaha RD400, but the bike went where I leaned instead of where he did. We never got out of the parking lot. He was about 80 pounds lighter than I was. ;)

Cheers

Phil Hobbs

Not sure how you get Dropbox is less embarrassing... ;-)

Imgur is the best free host out there. It's drag-and-drop, registration by traditional means (email + PW), or social media if you like, and you can share your images with the community (not that you'd ever want to do that).

Tim

Of course, I meant that it's never too late to get another motorcycle. I'd like a dirt bike, but you need a real death wish to ride on the street here.

...

One of the diodes shorted out (mysteriously, sigh), so I cut it in half to see what the heat path from the anode might be. Clearly inadequate to carry away much.

lots of nice toxic oil with a long history of killing people.

NT

Hey, that's better than I thought. I wonder if the brass-copper joint is soldered or brazed, and likewise for copper-silicon. And how wide the copper piece is (from a top-down view), if it's spoked or what.

Also neat to know it's two (hopefully, matched!) dies internally. Or maybe more.

Hmm, wonder what that wide die bond part is. Molybdenum? Or maybe the smaller part isn't actually silicon, and is actually a bond pad?

Tim

be

The usual use of an module is to rectify the output of a center tapped tran sformer. For that it makes no difference if the are matched or not. So t here is probably no effort done in matching the dies. But they are most li kely to be from the same wafer and at the same temperature.

Dan

Those two diode dies are hard in parallel. That's fairly normal in a big power diode. It spreads the heat out. The chips are no doubt well matched, and at high currents are ohmic with positive TCs, so will share current well.

But most of the heat will come out the bottom, the common cathode. The anode connections should be fat enough that they don't generate ohmic heat. Steel screws have a lot of resistance.

I'd use more diodes.

torsdag den 2. august 2018 kl. 02.19.08 UTC+2 skrev John Larkin:

maybe

ransformer. For that it makes no difference if the are matched or not. S o there is probably no effort done in matching the dies. But they are most likely to be from the same wafer and at the same temperature.

if for nothing else, just to get good clamping to a heatsink with three scr ews

On 8/1/2018 12:59 PM, Tim Williams wrote: ...

Well, when I was sawing it (from the side), there were 2 hard spots where the saw slowed significantly.

No, my Oxford has a notice on the nameplate saying to fill it with Shell Diala B which is pretty harmless (not polychlorinated biphenol). It is much smaller, I think 110 or 120 A max.

I can't help but imagine that there would be large inductive voltage spikes in an arc welder?

Do you need a snubber across the diodes?

that's good news. I trust it was correctly filled :)

NT

I had not considered that, being unaware of it. Thanks for the idea. However the LT4320 has an 80v max & that is the welder's OCV.

I Googled high power MOSFET's and they are remarkable - milliohm Rds means fractional watt power dissipation at 100A. What puzzles me is their packaging: TO-220AB has leads about the size of #18 wire, but carries 140A! I use #2 cable for welding & it would be bizarre to have that #2 cable connected to a TO-220 package. What am I missing?

Possibly an asterisk on the data sheet that says "silicon capacity" or something in tiny print.

IR invented preposterous current and power specs for mosfets, and everyone else had to go along.

The metallization of power diodes is intentionally resistive; you want that to prevent hot spots. But, that's an added layer over the aluminum (ohmic alloy) connection layer that all silicon diodes have...