aluminum heatsink anodizing
in [Design]
|
Prev: Automotive Protection Circuit
Next: uav matlab model
From: John Larkin on 18 Dec 2008 20:42 On Fri, 19 Dec 2008 01:08:45 +0000, Eeyore <rabbitsfriendsandrelations(a)hotmail.com> wrote: > > >John Larkin wrote: > >> Jamie Morken <jmorken(a)shaw.ca> wrote: >> >> >Does an anodized or painted aluminum heatsink perform better thermally >> >than a bare aluminum heatsink for the situation of to-220 components >> >attached to the heatsink with sil-pads? >> >> Sil-pads are so bad, it doesn't matter how you finish the heat sink. > >Try some of the Warth and Bergquist pads and take your words back ! NOT the >grey ones. Look at the white, green and 'rust' coloured ones. They are >STUNNING and conform *perfectly* to imperfect surface finishes, thus >ensuring even better heat conductivitity over the likes of Thermapath Al2O3 >grease. Not sure about how they compare to the silver particle grease, >never done a comparison. No way. Sil-pads are organics, rotten thermal conductors, and are mils thick, some as thick as 10 mils. That's a huge penalty. Dow filled silicone grease squishes down well below 100 microinches with modest pressure, 100x thinner than a sil-pad, and theta *is* proportional to thickness, you know. I've measured a lot of sil-pads. Even at insane mounting pressures, they are never as good as their datasheets claim. TO-220 plus sil-pad is an especially nasty combination... the mechanics are all wrong. I haven't tried silver particle grease, although I have tried diamond-filled grease. It's not any better than the regular stuff, because theta is dominated by the grease between the particles, not by the theta of the particles themselves. The main thing is to have small particles so they squish down thin. I suspect that unfilled grease may actually be the best, if things are really flat. The phase-change things are bad too, because they don't actually flow out from under the can. Thickness again. John
From: NoSPAM on 18 Dec 2008 21:18 "Eeyore" <rabbitsfriendsandrelations(a)hotmail.com> wrote in message news:494AF258.F974A11(a)hotmail.com... > If the heatsink is fan blown, then the main heat transfer mechanism is > conduction, not radiation or convection anyway. I beg to differ here, You have forced convection with a fan rather than natural convection without one, leading to improved heat transfer. If you really need to dissipate an enormous amount of heat from a small area, look at the Thompson CSF hypervapotron technology, a specialized form of nucleate boiling, used to cool extremely large transmitting tubes. They have achieved a heat transfer of 3000 watts per square centimeter surface using water as the boiling liquid. To go back to the original question, anodizing the surface of the heatsink will have little effect on the heat transfer while providing a scratch resistant, electrically insulating surface. Painting the heatsink will only lessen the heat transfer. At the temperatures normally found with semiconductor electronics, radiation heat transfer is usually negligible. For example a perfectly emitting (black body) object at 150 C radiating to a perfectly absorbing evacuated room at 20 C, the heat flux will be approximately 0.14 watts per square centimeter. If you have to insulate your TO-220 device from the heatsink, beryllium oxide provides an excellent electrical insulator which has a thermal conductivity nearly that of aluminum. Of course, BeO is extremely toxic if you should break one. 73, Dr. Barry L. Ornitz WA4VZQ
From: Archimedes' Lever on 18 Dec 2008 21:25 On Thu, 18 Dec 2008 15:38:23 -0800, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >On Thu, 18 Dec 2008 14:24:25 -0800, Jamie Morken <jmorken(a)shaw.ca> >wrote: > >>Hi, >> >>Does an anodized or painted aluminum heatsink perform better thermally >>than a bare aluminum heatsink for the situation of to-220 components >>attached to the heatsink with sil-pads? >> > >Sil-pads are so bad, it doesn't matter how you finish the heat sink. > >John I would use a level 3 hard anodize for electrical resistance, and use silver filled IC chip bonding epoxy to nail the part to it. Less serviceable, but passes the thermals as best as can be done while maintaining no electrical contact.
From: Archimedes' Lever on 18 Dec 2008 21:28 On Fri, 19 Dec 2008 01:08:45 +0000, Eeyore <rabbitsfriendsandrelations(a)hotmail.com> wrote: >Having said that I'd rather have a live heatsink with direct collector to >heatsink contact any day. Even so, you have to fill the micro-gaps. >Insulating a heatsink is easy. Use silver filled hard setting chip bonding epoxy for the absolute best result. That will fill all the gaps with a highly thermally conductive metallic media.
From: Phil Hobbs on 18 Dec 2008 21:45
John Larkin wrote: > On Fri, 19 Dec 2008 01:08:45 +0000, Eeyore > <rabbitsfriendsandrelations(a)hotmail.com> wrote: > >> >> John Larkin wrote: >> >>> Jamie Morken <jmorken(a)shaw.ca> wrote: >>> >>>> Does an anodized or painted aluminum heatsink perform better thermally >>>> than a bare aluminum heatsink for the situation of to-220 components >>>> attached to the heatsink with sil-pads? >>> Sil-pads are so bad, it doesn't matter how you finish the heat sink. >> Try some of the Warth and Bergquist pads and take your words back ! NOT the >> grey ones. Look at the white, green and 'rust' coloured ones. They are >> STUNNING and conform *perfectly* to imperfect surface finishes, thus >> ensuring even better heat conductivitity over the likes of Thermapath Al2O3 >> grease. Not sure about how they compare to the silver particle grease, >> never done a comparison. > > No way. Sil-pads are organics, rotten thermal conductors, and are mils > thick, some as thick as 10 mils. That's a huge penalty. Dow filled > silicone grease squishes down well below 100 microinches with modest > pressure, 100x thinner than a sil-pad, and theta *is* proportional to > thickness, you know. > > I've measured a lot of sil-pads. Even at insane mounting pressures, > they are never as good as their datasheets claim. TO-220 plus sil-pad > is an especially nasty combination... the mechanics are all wrong. > > I haven't tried silver particle grease, although I have tried > diamond-filled grease. It's not any better than the regular stuff, > because theta is dominated by the grease between the particles, not by > the theta of the particles themselves. The main thing is to have small > particles so they squish down thin. I suspect that unfilled grease may > actually be the best, if things are really flat. > > The phase-change things are bad too, because they don't actually flow > out from under the can. Thickness again. > > John > Textured indium foil is the best solid thermal interface material. Solder is better, but isn't reworkable, and has other drawbacks such as ripping the chip off the board as it cools. Liquid metals (gallium alloys) are better still but have serious corrosion issues if you aren't very careful. (My next-but-one office neighbour invented this idea--currently used by Apple in their highest performance machines, and by IBM here and there.) Cheers, Phil Hobbs |