You can also sand the transistor package itself if you want to get down a couple °C:
Sanding transistor packages
The pictures are misleading though as he uses lots of paste (too much as it seems).
Sanding transistor packages
The pictures are misleading though as he uses lots of paste (too much as it seems).
That's a perfect description of solder surely? Just be sure to use copper heatsink and grind surfaces properly flat.
In my work, we use phase change stuff all over, but the advantage of that stuff is repeatability and consistency only. We're not setting the material up in factory, it is thought to be done at customers usage.
In diy one can achieve much better even using the oldschool white grease by careful application.
Also, why care if it dries out if in the first place its needed to fill the gaps only? There is no need to have that, except where micro-movements are to be expected, but then, one should use pads anyway.
Solder 247/264 of yours to a decently sized (like 2"x2") 1/4...3/8 inch piece of copper with properly machined top and bottom mating surfaces. Use phase change material if you do not need isolation or boron nitride filled silicone pad if you do. Use plurality of evenly distributed screws to fasten. The whole deal will be like 20...30 time the price of the silicon to be cooled. Yet another thing would be thick base heatsink optimized for a particular coolling but it should of been like that to begin with. Some people, like the one and the only, use plurality of transistors to spread the heat.
It happens that i'm using the exact same transistors 🙂 .
I have 4 pnp +4 npn, each will have its own heatsink, obviously copper interface.The first trz and heatsink were polished and I applied the cheappest graphite grease I found in a car shop just to see in a couple of days if it reacts in any way with the copper.Some more copper fine powder will be added at the end in the grease before sticking the trz to the heatsink.Of course the existent one will be removed snd replaced.
As I can't drill the heatsink I'll need to press the trz against the heatsink helped by some aditional aluminium profiles if everything goes right.
Instead of going with the best compound I chose the cheapest one...
I'm also considering for tests some very heavy duty grease for industrial ball bearings, the blue zinc sulphonate variety as I was told it's very resilient to the harshest environments and I might find some use for my bike hubs if it fails to show any significant performance.For now i need a lot of zen attitude cause one heatsink and one trz took me more than half an hour to polish.
Attachments
Won’t rigidly bonding the transistor to the heat sink cause stress fractures due to thermal expansion if the glue and the transistor have slightly different thermal expansion coefficients? Thermal grease is there to enhance contact while allowing expansion.
I gave up that ideea the minute I wrote it down...some sort of exorcism was needed 🙂 Once all the copper parts will be fairly flat more fluidity and and less thermal properties will be needed for the grease.
The point's: properly machined/polished flat surface along with paraffin (phase change material) will do.
That' said I would not try to squeeze 200W out of 300W rated 247 package.
Simple clip will do miracles against "standard" trough hole attachment.
There are MMM SiC filled paraffin interfaces. The interface itself will improve "typical" 0.05 deg C/W resistance Case-Heatsink, let's say down to 0.03, having like 0.5 deg C/W j-c one of the 247 itself.
Do not forget AlN or alumina pads, which could be as thin as .25mm 10 mil.
Diamond grit is not expensive. Used in ton quantities in industry. https://www.amazon.co.uk/Diamond-Powder-000-Grit-0-0-5microns/dp/B009NA0A5S/ref=sr_1_4
Al = aluminium
N = nitrogen
Aluminium nitride, when pure, has extremely good thermal conductivity. https://en.wikipedia.org/wiki/Aluminium_nitride
Probably a lot more expensive than diamond grit though, as its newer.
It took me a full day to polish 3 more heatsinks and one more day to polish 8 transistors...It was a learning curve to make it better anyway...The most uneven proved to be the transistors...if all transistors are that bad then maybe optimizing that contact surface can be a really big deal...And the most weird thing is that the graphite grease has a very low conductance so I'm not sure how much graphite actually is in that...Most probably the final grease will actually be some very thin mineral oil for sawing machines or maybe Mos2 grease if it's going to be too thin.For the moment I used Cam2 Ultra580EP#2 universal grease to cover and protect the transistor's copper surface from oxidation.I used that grease for years not just in my bike rear hubs but also for every thermal contact without surface conditioning and I know it works really well.Used it also in some 2kWatt frequency converters for motor speed control and they are working 24/7 for about 3 years now...Some of my friends are using car motor oils for decades ...cheap country, cheap products...Thinking of buying some dense buttery blue mos2 grease, but I heard it's not that good for bike's hubs being too dense 🙂 Yet it offers a good thickener for any weird combination I might wanna try.
Attachments
After making some digging as I'm not a computer guy, I found that Mx4( based on graphite) is much easier to use than mx5 arctic (99% purity silver) while providing just 0.5 degree celsius less performance and both of them are blown away by Grizzly Kryonaut based on...aluminium and silicones which is almost double the price of mx4.
Being that they get way more expensive than the norm...I started to like my car graphite grease even more.It's conductive but i can deal with that for sure while I just found that the usual molibdenum sulfide found in Mo2 grease for heavy duty ball bearing has about 45W/mk which is about 8 times higher conductivity than mx4...Not sure how much molibdenum disulphide actually is in that car grease but the rest of it seems to be lithium, zinc oxide and some other petrol based organic compounds ... Being for heavy duty ballbearings these greases are not allowed to keep heat, they actually have pretty good conductivity, otherwise the balbrarings nondissipated heat would destroy the grease... An interesting study about two phase combinations of lithium grease and Fe, Cu or Al gave me sone hints to somehing that was probably best demonstrated by the Grizzly Kryonaut paste which is that Alumium can be superior to Silver (obviously to Copper too) when combined with something else in the same mass percentage due to volume/mass ratio , but I can't tell for the moment if It's the same thinking process that should be applied to Gryzzly grease. https://www.researchgate.net/public...ltipurpose_grease_filled_with_metal_particles
All in all it makes me feel hopefull about trying some moly blue grease too in between two copper surfaces.The best silicon thermal grease I found was almost always blue-ish and I wonder where it took its colour from ...Titanium oxides can be blue too...
But wait...after all this empty speach there are two guys who debunked the uneasiness of using graphene-diamond and copper paste in real world cpu cooling but also made these two videos which show some really professional options:
And see it at work where you really need it:
And...have some fun too:
Being that they get way more expensive than the norm...I started to like my car graphite grease even more.It's conductive but i can deal with that for sure while I just found that the usual molibdenum sulfide found in Mo2 grease for heavy duty ball bearing has about 45W/mk which is about 8 times higher conductivity than mx4...Not sure how much molibdenum disulphide actually is in that car grease but the rest of it seems to be lithium, zinc oxide and some other petrol based organic compounds ... Being for heavy duty ballbearings these greases are not allowed to keep heat, they actually have pretty good conductivity, otherwise the balbrarings nondissipated heat would destroy the grease... An interesting study about two phase combinations of lithium grease and Fe, Cu or Al gave me sone hints to somehing that was probably best demonstrated by the Grizzly Kryonaut paste which is that Alumium can be superior to Silver (obviously to Copper too) when combined with something else in the same mass percentage due to volume/mass ratio , but I can't tell for the moment if It's the same thinking process that should be applied to Gryzzly grease. https://www.researchgate.net/public...ltipurpose_grease_filled_with_metal_particles
All in all it makes me feel hopefull about trying some moly blue grease too in between two copper surfaces.The best silicon thermal grease I found was almost always blue-ish and I wonder where it took its colour from ...Titanium oxides can be blue too...
But wait...after all this empty speach there are two guys who debunked the uneasiness of using graphene-diamond and copper paste in real world cpu cooling but also made these two videos which show some really professional options:
Some more digging shows that the molybdenum disulfide content goes down to the aimed speed use...as high speed compounds don't really have molybdenum disulphide in it but molybdenum oxides and sulphures that reacts after mixing at high speed and make the moly...thus the thicker low speed the grease is the more molybdenum disulfide already formed it might have in it...
On the other hand it seems that generally the same substances usually used in thermal greases are also used for thermal coupling
The real question I never asked myself is what's that grease going to do with copper...As it can be in contact with brass in ordinary mechanics it should be stable enough at ambient temperatures , but when you add watter to it...this happens: https://www.nature.com/articles/203750a0
In some conditions that I don't know though it ca form a cathalitic substance:
https://pubs.rsc.org/en/content/articlelanding/2012/ee/c2ee22611a
Or supercapacitors... https://www.sciencedirect.com/science/article/abs/pii/S2352152X21008434
But one thing is certain: Copper oxides are Audiophobe diodes so your beloved sound will be rejected by them, not allowing for audiophile leaks to the ventilation system and enhace your musical experience 🙂
In some conditions that I don't know though it ca form a cathalitic substance:
https://pubs.rsc.org/en/content/articlelanding/2012/ee/c2ee22611a
Or supercapacitors... https://www.sciencedirect.com/science/article/abs/pii/S2352152X21008434
But one thing is certain: Copper oxides are Audiophobe diodes so your beloved sound will be rejected by them, not allowing for audiophile leaks to the ventilation system and enhace your musical experience 🙂
How is that i could forget this? Just some magnesium and I won't need a rectifier bridge anymore 🙂
Some heatsinks are varnished or anodized. Then try without any plates (multimeter)
And: We do hear the materials of heatsink, plates, screws, transistors...! A soundtuning does regard the materials and current...
And: We do hear the materials of heatsink, plates, screws, transistors...! A soundtuning does regard the materials and current...