The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: email@example.com (Bob Wilson)
Subject: Re: Question about linear derating factor for MOSFETs
Date: Wed, 13 Nov 2002 03:56:32 -0000
Organization: Your Organization
X-Newsreader: WinVN 0.99.9 (Released Version) (x86 32bit)
References: <firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org>
In article <email@example.com>,
>>>>The figure of 0.5C/W for a greased (and non-insulated) interface between
>>>>and heatsink is pretty optimistic. You'd be lucky to get much less than
>>>>4 times that, and if you are planning on insulating the interface, the
>>>>you can do is several times that. Thermal resistance from cast to
>>>>going to be a problem with the hich power density you are looking at.
>For the TO-220/247 devices, I agree. But, the SOT-227 cased device I
>used in the example has a very low thermal resistance to the heat sink
>(0.20-degrees C/W). This, along with the 0.34-degrees C/W thermal
>resistance of the heat sink I was considering allows me to come in at
>about 0.7 (with the more realistic number of 0.2 for the thermal
>grease, not 0.05). I've since switched to a heat sink that has a
>0.18-degrees C/W thermal resistance to air. This keeps the total
>resistance to about 0.5-0.6. The SOT-227 case is already isolated so
>all I need is some high performance thermal grease between it and the
>>>>By the way, the very best insulating interface is a thick aluminum
>>>>(a rectangular one in this case), that has been heavily anodized (which
>>>>what provides the electrical insulation), then a thin layer of thermal
>>>>compound is applied to both sides. These things are available
>>>>but hard to find.
>I've seen them mentioned in Wakefield and Thermalloy app notes and
>tech papers. I'm lucky enough to have an isolated case with the
>>>>Another thing. You MUST forget about attempting to mount the TO247 using
>>>>normal single-screw mount. At this power level mounting it with a single
>>>>screw to the heatsink will cause your thermal resistance to go right
>Thanks for the tip. I've read that overtightening can cause the case
>to tip up and significantly increase the thermal resistance. It's
>nice to have confirmation that the single screw mount hole isn't the
>best thing for high-power applications.
>>>>Standard procedure for mounting TO220 and TO247 devices that are
>>>>high power, is to clamp them using a bar that presses in the center of
>>>>device (i.e. on the plastic case). Usually this intails a bar that has
>>>>at each end, laying sideways across the center of the device. These
>>>>are hopeless for high dissipation when mounted with a single screw.
>I've seen the clips that go across the TO-220/247 devices to hold them
>against a heat sink (Thermalloy has an huge line of clip-only heat
>sinks for high-power apps) but haven't seen screw mount bars. Seems
>like that would be a great idea to help pull heat from the top of the
>tab too (if machined accurately enough). Are these bars commercially
>available as a standard product?
Typically not. But few people who design volume-produced products (e.g.
inverters) would be crazy enough to use them anyway, since the prices
Wakefield, Thermalloy and others charge for their mounting hardware are
absurdly high. Typically these things are custom made. In their simplest
form, they are just a small sheet metal "U"-shape, or even a small bar, with
holes at each end.
Hell, even heatsinks from these guys are ripoffs. A typical price for
custom extruded aluminum shapes (directly from the extruder) is $2.50 per
pound. If you calculate the cost per pound for a commercially available
heatsink, it is 10 times that. It pays to make the tool (extrusion tools are
cheap) and squirt out your own if there is even a moderate volume involved.
I can't remember when any company I worked for actually BOUGHT a commercial
heatsink. The joke is that none of the commercial heatsink companies
actually extrude anything. They contract it out to a real extruder (the same
people one can go to direct).
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup