Mounting MOSFETs on heatsink
Hi there
I want to place a MOSFET h-bridge onto a heatsink. The MOSFETs have to be monted to the heatsink, but there's a problem - the integrated heatsink on the TO220 housing is electrically conducting, should I buy two large heatsink, or do I just use grease, so only the heat is tranfered (a bit risky)?
Grease (conductive and insulating) just fills in the air gaps between the contact areas which means there is still metal-metal contact. Either way, there is still meta-metal contact. YOu can either use two heatsinks, or use some thin electrically-insulating thermally conductive pads (like thin sheets of mica). They sell these especially for the purpose.
If you want really good heatsink contact you're going to have to use seperate heatsinks, anything you put to electrically issolate the tab is going to slow down thermal transfer as well, even mica. Is there any reason you can't purchase issolated version of the FETs you're using?
"Because I be what I be. I would tell you what you want to know if I
could, mum, but I be a cat, and no cat anywhere ever gave anyone a
straight answer, har har."
But an isolated version doesn't transfer heat as well anyway - because it has the insulator built-in.Originally Posted by Sceadwian
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk
Unless the die is mounted on diamond.
Seriously I've heard that that's actually done with the more higher powered semiconductors, they use synthetic diamond of course.
I do not answer private messages asking for help because no one else can: benefit from advice I may give or correct me if I'm wrong.
Please ask on the open forum if you have a question and I'll be happy to help, if I know the answer.
The horizontal output transistor and the two transistors in the power supply of my TV are all the insulated version even though they have separate heatsinks so the insulation isn't necessary (maybe for reduced interference?).
This 32" Sony TV died a few months ago and now works fine with the 3 transistors and a fused-resistor replaced. The heatsinks are STEEL, not aluminum and the thermal compound was dry and hard as a rock.
Uncle $crooge
Nigel I think then the question is, which issolation method conducts heat better, the manufactors or the end user. That's very hard to quantify, although I would tend to think the manufactor knows what they're doing.
"Because I be what I be. I would tell you what you want to know if I
could, mum, but I be a cat, and no cat anywhere ever gave anyone a
straight answer, har har."
Mica insulators can be used. But even simpler, use an anodized heat sink. The anodized coating is non-conductive. Use screws and nylon insulators to attach the mosfets. I have also used a thermally conductive adhesive (Loctite 383) to aid the attachment.
No, it's simply because it's fairly anti-social having a large piece of bare metal with an 8000V RF pulse on it!
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk
Bad bad bad advice. Never heard anyone done that.
Don't ever rely on the anodized coating as the insulation. Nine times out of ten the insulation breakdown if sufficient pressure is applied to the fastening hardware.
L.Chung
My original comment was written late, and I was too brief. Let me clarify what I meant, which was to emphasize that the surface of anodized aluminum is non-conductive, and that property can be used to advantage. In fact, I got the idea for mounting mosfets in my application from examining some commercial power supplies used in laboratory equipment. Obviously, one needs to be careful about applying excessive pressure at point contacts and edges.
My application was a high-current, 12V supply. The mosfets were attached directly to 0.025 inch copper sheet, which in turn was bonded to the anodized heatsink with Loctite 383 adhesive. No additional insulation was used, except in areas where high pressure or point was anticipated. That adhesive is sold as a thermally conductive, electrically non-conductive, one-part material. It relies on surface activation to cure.
I have attached some photographs of the power module portion of the controller. On the left side of the assembly, where the 0.25 inch diameter screws are for connection to the cables, I used a small piece of mica under the copper and beneath the terminal block. It looks a bit like a shadow in the photos. The remainder of the copper is attached directly to the heatsink with adhesive. No other screws were used, except for those that were used to stabilize the conductors next to the terminal block.
This supply was not fully enclosed. It was simply attached by the four standoffs seen in the pictures to the winch it was used to control. It has been used for about 8 years on weekends outdoors and shows tarnish and corrosion from that service. The solder joints are really not as bad as they appear, but I would caution people who haven't tried it that soldering to heavy copper sheet requires a much larger iron than one would use for most other electronic work. John
Ok I have studied your photos.
In your case the main insulation is actually the 383 bonding adhesive and not solely rely on the anodized surface between conducting surfaces so it will be acceptable and sufficient.
I have also observed very good wiring practices, like bending of the component leads around the MOSFET pin before soldering and the use of telfon sleeves for bare conductors.
A carefully constructed project and a very nice piece of work indeed.
L.Chung
jpanhalt, the common thickness of a decorative anodizing layer on an aluminum substrate is only 5 micrometers, or about .0002 inches. I'd hardly call that insulation.
"Because I be what I be. I would tell you what you want to know if I
could, mum, but I be a cat, and no cat anywhere ever gave anyone a
straight answer, har har."
Reply With Quote
