Heatsinks

Looking at Pass' figures on the dissipation of an Aleph 5 I figure I would need a heat sink or mulitple heat sinks equaling .083 C/W to run at 50 degrees celcius. I found some big surplus listed at $42.50 but they're really $35.00. How many of these would I need (power consumtion is 300 watts)? If I buy new sinks how much should I be prepared to spend on them. Thanks for the imput.
 

paulb

Member
2001-06-01 4:53 pm
Calgary
Keep in mind you have to add the thermal resistance of the heatsink (Theta_sa, sink-to-ambient) to the resistance of the junction to the case (Theta_jc, shown on datasheets) and the case to the heat sink (Theta_cs: depends on how you mount it, won't be much less than 0.5 C/W). Even an infinite heatsink would not effectively give you .083 C/W.

When using parallel transistors, calculate for one individual transistor. Divide the temperature rise (25 C?) by (300 watts divided by the number of devices): this will give the TOTAL thermal resistance (C/W). Subtract the Theta_jc and the Theta_cs. If you're left with a ridiculously small or negative number, you need more devices.
Other folks browsing this site will undoubtedly have some experience to say what they get for Theta_cs.

More info is at http://sound.westhost.com/heatsinks.htm.
 
I've looked at that same heatsink, and it seems that each one may be sufficient to dissipate somewhere near 125 watts. A rough eyeball indicates the heatsink is #61570 from R-Theta and is rated at 0.8 c/w per 3" length.

Get yourself some SilPad 2000 or SilPad k-10 from Bergquist to insulate your mosfets from the heatsink. Its rated at about 0.2 c/w which is excellent. Search the Digikey web site for SilPad.
 

djk

R.I.P
2001-02-04 4:23 am
USA
'promitheus' is in error.Kapton by itself makes a very poor isolating pad."The Bergquist silpads are the best rated that I've seen so far at 0.20 c/w... " I think you are correct here.These are a boron-nitride filled elastomeric. The only thing better would be BeO2 ceramic.But one thing to remember is that the 0.20*C/W is based on a TO-3 case with high mounting pressure.The highest performing of the Bergquist sil-pad series, the 2000, at 100PSI for a TO-220 case ranges from 1.65*C/W ~ 2.0*C/W depending on how thick it is.This kind of clamping pressure should not be applied with a single mounting screw but rather with a bar and spring clamp.The larger TO-247 packages can get by with a large bell washer. http://www.bergquistcompany.com/thermal/pdf_2000/SP_2000_2001.pdf
 
I agree that Kapton is one of the best isulating material for power transistors. But use of very thin films needs perfectly machined, and perfectly cleaned heat sinks. Because of this, mass production uses elastomeric films, the elasticity of which is supposed to cancel surface irregularities. But thermal resistance increase dramatically.

Regards, P.Lacombe.
 

djk

R.I.P
2001-02-04 4:23 am
USA
"hmm, i tend to disagree on that. " Gee, you must have a really good lab.I mean better than the people that produce these things.Note that I gave a direct link to the spec sheet.Kapton film insulators by themselves, without the filled elastomer, have 0.3*C/W for a TO-3 package.The BQ sil-pad 2000 is 0.2*C/W and BeO2 is 0.17*C/W .Thermal compound with no insulator is 0.09*C/W ."I think the numbers speak for themselves..." They certainly do.promitheus said: "The best are kapton and have 0,07 thermal resistance." This is clearly incorrect for any case sized transistor used in audio.While kapton has good high voltage insulation properties it has poor conformal or gap filling properties, the secret to good thermal performance.Some of the filled elastomer insulators use a very thin piece of kapton film for voltage isolation but it is the binder/filler that makes them conduct heat well, not the kapton.The search for something to replace grease type thermal compounds was driven by two main items: 1)eliminating the grease itself because it boils off in space applications and contaminates wash solutions 2)it is expensive to re-finish the heatsink every time it is dis-assembled.The ZnO2 in filled heatsink compound is abrasive.In high performance situations wih hocky-puck devices, SCR driven inverters and energy management, two approaches are taken: 1)Gap filling compound, the white stuff 2)clear grease, It costs more to use clear grease because the heatsink must be milled to a TIR of 0.001" before mounting the device.ZnO2 filled compound fills the gaps well enough to use the the extrusion as received from the mill.It is abrasive enough that just assembling, dis-assembling once, and subsequent re-assembly degrades the thermal performance enough to justify re-finishing the assembly from a mean time before failure standpoint.I like the Grafoil washers from Avid where voltage isolation is not required.I like the Sil-Pad products from Bergquist where insulation is required.I use BeO2 where the circuit demands low capacitance to the sink.
 

djk

R.I.P
2001-02-04 4:23 am
USA
Catalogs frequently present incomplete or wrong data.The people doing the layout usually know nothing about the items they are cataloging.That is why I gave a direct link to the data sheet for the part in question.In the data sheet there are charts showing *C/W at different case areas and mounting pressures.After you examine this data you will realize that the 0.07*C/W number you quoted is simply not possible for any transistor case syle used for audio.
 
So since the Aleph 5 draws 150 watts per channel I would need at least 3 of theese. I figured that I needed to mount the transistors together in the groups of 3 that are indicated on the schematic. Since there are 4 sets I figured I needed an even number of sinks. So I need 4 sinks one for each group. Is that correct? I know that may be overdoing it a bit, but I think it's still cheaper than buying new isn't it? If I did get 4 what chance would there be that it wouldn't be enough? Thanks for all of the imput. I'm looking at the esp page right now.

Rob
 

djk

R.I.P
2001-02-04 4:23 am
USA
("High Performance Phase Change Thermal Compounds" with a low thermal impedance of 0.008 °C-in²/W like POWERFILM
I can't seem to find if they are insulators or not?) First off it's non-insulating.Second is that it's for hockey-puck and energy management packages at very high clamping pressure on a sink that is polished to 64 µ inches.At typical thicknesses and pressures used for audio semiconductors it is comparable to regular ZnO2 heatsink compound.The heatsink must also be pre-heated to apply this stuff(it is a little like the Avid Kon-Ducs that must be vulcanized to the sink).Lots of neat stuff out there.I listed my preferences earlier.
 
AOS Thermal Compounds has these, they appear more suitable.
<i><a href="http://www.aosco.com/microfaze.shtml">AOS MICRO-FAZE® K</a> Dry Film Interface Material
It was developed by AOS to offer the lowest thermal resistance in a thermal interface without the mess of grease.

MICRO-FAZE K is a die-cut polymide insulating substrate coated on both sides with specially formulated thermal grease (non-silicone, non-wax-based) that is naturally tacky but dry to the touch. It offers unique heat transfer and high insulating capabilities, and has a high cut-through resistance.

<li>Unlike phase change materials, MICRO-FAZE K requires minimum force to achieve total interface contact and heat transfer starts at any temperature.

<li>A positive coefficient of thermal expansion increases the wetting action for total interface contact.

<li>MICRO-FAZE K microscopically changes to fill all microscopic voids on part surfaces.

<li>Thermal Resistance: 0.03°C in<sup>2</sup>/W</i>

Regards
James
 

djk

R.I.P
2001-02-04 4:23 am
USA
And we're back to Kapton with a coating again.I wonder what they are calling a low clamping pressure? And I wonder how much this stuff costs? The better Berquist products cost more than the 250V 250W 25A outputs I am using.Actually I am using forced convection with electrically 'hot' heatsinks isolated from the chassis.No insulators.