Hey all, trying to attach an Aluminum Bar (18" long x 3.5" wide x 1/2" thick) to a set of three heatsinks in a Krell KSA-300 Chassis. Surfaces are nice and flat on both the aluminum flat bar and heatsink face. Originally, Krell had some goofy flexible metallic pad or something there, but was removed after I sandblasted the heatsink face. This interface does not need to be electrically isolated, like the connection of output device to Aluminum bar - I plan on using a the Thermon Pad to connect the output devices to the Aluminum bar. Surface of heatsink is 4" x 2.5" wide....
Question is what to use? Thermal paste, Thermon Pad, or nothing?
The thermon pads I have are around 2W/mK conductivity (I think)....I bought a bunch of these yellow pads 2.5" x 2.5" x 1/16" thick....use them where I need the electrical isolation...but don't think I should use it in this particular application...
Question is what to use? Thermal paste, Thermon Pad, or nothing?
The thermon pads I have are around 2W/mK conductivity (I think)....I bought a bunch of these yellow pads 2.5" x 2.5" x 1/16" thick....use them where I need the electrical isolation...but don't think I should use it in this particular application...
If the surfaces aren't machined, you may be better off using the thermal pads with paste.
I believe that the voids filled with thermal grease or pad is a LOT LESS conductive than the metal to metal. If I use paste, I will have entire face area conducting through the paste, limiting heat transfer, while using metal to metal (no thermal grease) will have tiny voids, but a majority of the face is in direct contact....yielding the best heat transfer...
Surfaces appear machined...
Am I wrong here??
Surfaces appear machined...
Am I wrong here??
If you think that both mating surfaces are machined and clean, try placing them together and sliding them around.
You'll be able to tell. If you fasten them together bare, use reasonable torque on the fasteners to avoid warping.
Does the surface reflect light, like a mirror if looked at down the end? Shows flatness to some degree...
What holds the two pieces together?
I'd put the grease in there, if you squish two pieces together that are flat enough, there will be no grease remaining between the two pieces anyhow.
got some pix?
_-_-
What holds the two pieces together?
I'd put the grease in there, if you squish two pieces together that are flat enough, there will be no grease remaining between the two pieces anyhow.
got some pix?
_-_-
The heatsink and bar pieces are fastened together by a bolt in middle. I have fastened the pieces tight and tried to slip a piece of paper between - all the way around - nice and tight...Since aluminum is so malleable, I am thinking a single kitchen aluminum foil rectangle sandwiched between the two faces...maybe two layers of foil...
So what?
"Voids" by definition means "NOT metal to metal contact".
No, once you apply pressure paste flows away, only stays where there's voids.
Think again.
Only way to approach full contact is to apply progressively finer abrasive paste between both surfaces and move them around, applying pressure, until you grind away differences.
You should use rouge polishing in the last step, and even so .....
Or if you think to "crush your soft aluminum foil" to make it flow and fill the gaps, remember you'll need some 40 Tons per square inch.
Compared to alternatives, thermal grease looks attractive.
I wonder what have they been using for the last 60 years.
Unless you do so, forget it.
....yielding the best heat transfer...
Surfaces appear machined...
Am I wrong here??
I'd lap the bar to the sinks then use 'enough' fasteners to hold the caboodle together..... with no paste.
If you go as far as rouge, you might, with patience, be able to get things so flat that when parts are pressed together, they 'stick' through exclusion of air - much like tool room joey blocks.
If you go as far as rouge, you might, with patience, be able to get things so flat that when parts are pressed together, they 'stick' through exclusion of air - much like tool room joey blocks.
Yeah yeah, I get it....I know that the paste flows away where there are no voids, but have you ever tried to wipe up or remove thermal grease? It remains no matter how hard you scrub it off, and I cannot believe it just flows away leaving metal to metal contact, thus lowering the conductivity at all applied locations with remaining film... that's all I am saying.....
In the performance PC world , they "buff" the HS so the compound is as
thin as possible. Where you assume (or think you see) metal to metal contact,
there is just a several molecules thick layer.
This seems to be the ideal that the watercooled , monster-fan PC
crowd aspires to.
Removing compound is easy with acetone , or even alcohol.
OS
thin as possible. Where you assume (or think you see) metal to metal contact,
there is just a several molecules thick layer.
This seems to be the ideal that the watercooled , monster-fan PC
crowd aspires to.
Removing compound is easy with acetone , or even alcohol.
OS
At the micron level, all metal is irregularly shaped. The compound is designed to have molecules smaller than the voids in the metal, so it cannot be cleaned off with a dry wipe and needs a solvent.
Paste is the ideal solution if you are confident the surfaces are perfectly flat and properly mated. Just the outside edge will not confirm that, you need a shadow print using a thin layer of paste (toothpaste works fine, we use it when checking out our lapping progress).
If you have a slightly uneven fit, it is better to use a Sil-pad or phase change material between the two for a little better filling of gaps. It is not possible to eliminate all gaps but we can get close.
Paste is the ideal solution if you are confident the surfaces are perfectly flat and properly mated. Just the outside edge will not confirm that, you need a shadow print using a thin layer of paste (toothpaste works fine, we use it when checking out our lapping progress).
If you have a slightly uneven fit, it is better to use a Sil-pad or phase change material between the two for a little better filling of gaps. It is not possible to eliminate all gaps but we can get close.
.....Or save yourself the headache of trying to make a fully worked engineering solution or rely on special materials. You don't need insulation, so why introduce it in the form of non-conductive rubbers or grease? Simply make larger, over-sized joints that reduce the need for precision, high surface quality or any interface materials. After all, you are only trying to achieve a sufficiently low thermal resistance, not make an exemplary showpiece.
In any case though, multiple fasteners apply more even force, which is what you really need to couple 2 flat surfaces. Few people have any skill in lapping surfaces flat, as opposed to just polishing them. Don't waste your time trying this without the appropriate equipment and materials either.
Unless the extrusions are not able to be clamped flat or of poor surface quality, you shouldn't need to further machine or lap them if you simply clamp them over an area of at least 10 x the thinnest material's cross-sectional area. Obviously, that area ratio is the main factor of the conductance you can achieve, so the more area, the better.
In any case though, multiple fasteners apply more even force, which is what you really need to couple 2 flat surfaces. Few people have any skill in lapping surfaces flat, as opposed to just polishing them. Don't waste your time trying this without the appropriate equipment and materials either.
Unless the extrusions are not able to be clamped flat or of poor surface quality, you shouldn't need to further machine or lap them if you simply clamp them over an area of at least 10 x the thinnest material's cross-sectional area. Obviously, that area ratio is the main factor of the conductance you can achieve, so the more area, the better.
A picture is worth a thousand words.
IF this is those Krells with the multiple heatsinks, each with splayed out fins, and they held it on with a single bolt in the middle of flat area (where normally it is drilled for TO-3 footprint and mounting), then that's a BAD way to do it.
Once you put torque on the single point, the metal will try to dish, with the points closest to the bolt making contact. One would want at least two bolts, and best would be some means of distributed pressure, which amounts to multiple holes and bolts...
I'm a bit confused, is this a Krell clone or an original Krell. If it is an original Krell, then why the need to mess about with it??
And the sandblasting - where was that done? On the mounting face or the fin side, or both?
IF you did it on the mounting face, then you have a problem. You'll want to get that side/face flat and smooth before attempting to mount anything to transfer heat to it.
Please do let us know more about what the project consists of, how much heat it needs to transfer and pix of the thing, if possible??
IF this is those Krells with the multiple heatsinks, each with splayed out fins, and they held it on with a single bolt in the middle of flat area (where normally it is drilled for TO-3 footprint and mounting), then that's a BAD way to do it.
Once you put torque on the single point, the metal will try to dish, with the points closest to the bolt making contact. One would want at least two bolts, and best would be some means of distributed pressure, which amounts to multiple holes and bolts...
I'm a bit confused, is this a Krell clone or an original Krell. If it is an original Krell, then why the need to mess about with it??
And the sandblasting - where was that done? On the mounting face or the fin side, or both?
IF you did it on the mounting face, then you have a problem. You'll want to get that side/face flat and smooth before attempting to mount anything to transfer heat to it.
Please do let us know more about what the project consists of, how much heat it needs to transfer and pix of the thing, if possible??
You can't just increase the surface area like that. If you made the joint big enough to have low thermal resistance even without thermal paste, most of it would be so far away from the heat source that the total thermal resistance would not be lower. This is the reason why water cooling works - there's a limit to how much surface area you can cram in around the heat source, so you need the lowest possible thermal resistance to move the heat away to where you can fit in more.
Also, since I haven't seen it mentioned yet: Use the highest clamping pressure you can, as this makes a big difference to the thermal resistance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.