aluminum heatsink anodizing
in [Design]
|
Prev: Automotive Protection Circuit
Next: uav matlab model
From: Michael A. Terrell on 20 Dec 2008 17:43 Eeyore wrote: > > "Michael A. Terrell" wrote: > > > Eeyore wrote: > > > > > This is 'state of the art' equipment. > > > > The state of what art? > > But one and find out. But one, and find out what? > Who else here can say that ? Affordable (from your pocket money second hand), good > performance, ultra reliable and possibly as much as a 25+ year lifespan. I don't need used British Craptronics. -- http://improve-usenet.org/index.html aioe.org, Goggle Groups, and Web TV users must request to be white listed, or I will not see your messages. If you have broadband, your ISP may have a NNTP news server included in your account: http://www.usenettools.net/ISP.htm There are two kinds of people on this earth: The crazy, and the insane. The first sign of insanity is denying that you're crazy.
From: Eeyore on 20 Dec 2008 17:43 krw wrote: GARBAGE
From: Nico Coesel on 20 Dec 2008 17:44 John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >On Fri, 19 Dec 2008 22:26:49 +0000, Eeyore ><rabbitsfriendsandrelations(a)hotmail.com> wrote: > >> >> >>John Larkin wrote: >> >>> No insulators at all of course... the heatsink is hot to the load. >> >>Isn't that just what I recommended ? > >Was it? > >> >> >>> This is in a 17KW peak-output MRI gradient driver. >> >>And the dissipation in the devices is ? >> > >About 300 watts peak each. We tested lots of fets to to destruction, >find the ones with adequate margins. > >ftp://jjlarkin.lmi.net/ExFets.jpg > >Each fet has its own local opamp-closed-loop driver, to ensure precise >current sharing regardless of gate thresholds. A microprocessor >measures everything and runs a realtime dynamic simulation of junction >temperatures, so we can push them precisely. The uP can also test each >fet individually, at low available power supply energy, which helps a >lot in production test. Ever considered going for a ceramic substrate glued to a heatsink? It allows to use SMD parts and would save a lot of manual labor. I've seen this for high power battery chargers. The only problem is that it is very difficult to do rework once the substrate is glued to the heatsink. -- Failure does not prove something is impossible, failure simply indicates you are not using the right tools... "If it doesn't fit, use a bigger hammer!" --------------------------------------------------------------
From: Michael A. Terrell on 20 Dec 2008 17:58 Eeyore wrote: > > krw wrote: > > GARBAGE Oh, my. Now he's posting his lunch orders. -- http://improve-usenet.org/index.html aioe.org, Goggle Groups, and Web TV users must request to be white listed, or I will not see your messages. If you have broadband, your ISP may have a NNTP news server included in your account: http://www.usenettools.net/ISP.htm There are two kinds of people on this earth: The crazy, and the insane. The first sign of insanity is denying that you're crazy.
From: John Larkin on 20 Dec 2008 18:53
On Sat, 20 Dec 2008 22:44:30 GMT, nico(a)puntnl.niks (Nico Coesel) wrote: >John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > >>On Fri, 19 Dec 2008 22:26:49 +0000, Eeyore >><rabbitsfriendsandrelations(a)hotmail.com> wrote: >> >>> >>> >>>John Larkin wrote: >>> >>>> No insulators at all of course... the heatsink is hot to the load. >>> >>>Isn't that just what I recommended ? >> >>Was it? >> >>> >>> >>>> This is in a 17KW peak-output MRI gradient driver. >>> >>>And the dissipation in the devices is ? >>> >> >>About 300 watts peak each. We tested lots of fets to to destruction, >>find the ones with adequate margins. >> >>ftp://jjlarkin.lmi.net/ExFets.jpg >> >>Each fet has its own local opamp-closed-loop driver, to ensure precise >>current sharing regardless of gate thresholds. A microprocessor >>measures everything and runs a realtime dynamic simulation of junction >>temperatures, so we can push them precisely. The uP can also test each >>fet individually, at low available power supply energy, which helps a >>lot in production test. > >Ever considered going for a ceramic substrate glued to a heatsink? It >allows to use SMD parts and would save a lot of manual labor. I've >seen this for high power battery chargers. The only problem is that it >is very difficult to do rework once the substrate is glued to the >heatsink. We needed maximum thermal mass in intimate contact with the 300 watt fets (32 of them!) so we clamped the big fets directly to very flat copper heat-spreader blocks. At higher powers, thermal spreading resistance gets to be important, so the first inch or so radius from the fet should have the lowest theta and highest thermal mass possible. Once the heat is spread out some, we can transition to aluminum. Thin heatsinks can be a problem: ftp://jjlarkin.lmi.net/Infinite_Sheet.jpg At higher power densities, you just need a thick heatsink baseplate to spread the heat away from the fet, and not get a local hot spot. The "theta" specs of extruded heat sinks don't account for this. Mismanaging this can mean under-using expensive fets and drivers and such by 2:1 at least. Maybe 5:1 if you use sil-pads and sheet metal heat sinks. We can't afford to under-use fets, one reason being that the drain capacitance matters, too. High-speed stuff often gets to be a thermal problem... small parts running at high power dissipations are difficuly to cool without adding lots of capacitance. Except for diamond, BeO has by far the highest ratio of thermal conductivity over dielectric constant. AlN is next. John |