Carbon fiber for CPU air coolers

Toggle sidebar Toggle sidebar
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
I'm posting this in the CPU section in the hopes some cooler mfgr or someone thinking about getting into manufacturing them, would get wind of this post.

i'm surprised no mfgr has offered an air cooled unit fabricated from carbon fiber - n0ns3ns3 responded to this statment in another thread and somehow it got shut down as it was originally a 2 year old thread.

But CF (carbon fiber) carries heat almost as fast as copper carries electricity, literally. I worked with it on a few products and were amazed at how it performed in terms of transferring heat. We took a pc of 12 oz CF cloth (think in terms of fiberglass cloth), draped it over a vise and took a torch to it. It took forever to get a hot spot glowing red, same as you'd see with metal - and you had to hold the torch at the perfect distance so the hottest part of the flame was hitting the cloth. Once you got a hot spot glowing, about the size of a quarter (approx3.0 to 3.5CM diameter), the instant you pulled the torch away, and i mean the instant the flame lost contact with the cloth, the glow would go away, and within 1.5 seconds you could touch that very spot with your bare finger.

We manufactured what for all intents and purposes was an exhaust gas muffler - similiar to a car muffler but operating much closer to the heat source than a car's muffler. The same tube shaped "muffler" fabricated in aluminum ran 175 degrees Farenheit. Fabricating the outer tube in carbon fiber (and leaving the internal baffles fabbed in aluminum) it ran 114 degrees - and this was in an environment with the ambient temp of 94 degrees F. Additionally, the unit cooled down enough to handle it within 1 - 1.5 minutes, while the aluminum unit took 15-17 minutes. Both units contained the identical amount of interior baffles, about 15 ounces of aluminum baffles serving as heat banks.

I'm as surprised intel hasn't used CF for it's IHS shell - it would definitely do a better job of transferring the heat out - stainless steel, which appears to be what they're currently using, is actually a heat insulator - slow to accept heat and even slower to give it up to the atmosphere. That same tube muffler with only the outer body fabricated in stainless took 30 minutes to cool enough to handle.

For those curious on this subject, of the alloys, Brass, copper and aluminum have the highest rate of thermal conductivity, with Brass having the highest, copper 2nd and aluminum 3rd. But Aluminum is the cheapest so that's why you see it used so commonly. Of the elements, carbon has the highest rate of thermal conductivity.

Someone could offer replacement IHS covers, if folks will delid to enhance heat transfer, why not a replacement IHS? On the air coolers, even if the base block that holds the heat pipes pressed against the IHS, that could be fabbed easily from CF and show a marked improvement and a slight change in design to encompass the heat pipes entirely - when the CF accepts heat faster than copper, why bother putting the copper heatpipes in contact with the IHS?

CF can be purchased laid up or injected in rod or bar form, it's gotten that common and is machineable using diamond grinding cutters.

anyway, hoping someone from the air cooling industry sees this post. I'm retired so i'm out of the manufacturing field

Oh, the only negative to CF, is it is expensive, but it's cost is manageable, and if you can show a 35%+ increase in cooling efficiency, the market would accept the additonal cost
 
Sort by date Sort by votes
D

Dugimodo

Distinguished
Sep 22, 2011
1,393
11
20,115
Well I don't know why the material you experimented with acted that way but a quick google on the thermal conductivity of carbon fibre yields several results that suggest it is a poor conductor of heat and some that even describe it as an insulator. That's enough for me to decide it won't be very likely to make a good heatsink.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
not sure how they can state that - carbon, as in diamonds or carbon fiber has the absolute highest rate of thermal conductivity. Can you share some of those links?

One reason diamond grinding cutters last so long, not only because of their hardness, but they transfer the heat from the cutter out - the hotter something gets, the softer it gets and that's when you see wear on the cutter.

To keep my descript as short as i did up above, i didn't mention, the tube bodies for the exhaust muffler we fabricated, were made from CF cloth saturated with epoxy resin, same as you'd do with fiberglass. I actually put off the idea of fabbing some prototypes from CF as i kept thinking the epoxy resin would insulate the CF from accepting heat., when i finally got around to it, i was surprised at the results above.

additionally, nascar race teams are using CF, not only in car frames, but in cooling elements, like radiator upper & lower radiator tanks (for additional cooling) as well as oil coolers.

we fabricated some ribbed plates to bond to cast aluminum oil pans on a race car engine, the ribbing acting as cooling fins - in 1/4 mile runs, we saw a drop in oil temps of 18 degrees. And that was just bonding the ribbed plates to the bottom of the oil pans, and iirc total area of the rib plates was 166 inches. I'd love to have had the funding to build the entire oil pan from CF, to see what additional temp drop we would see
 
Upvote 0 Downvote
T

tejayd

Prominent
Mar 11, 2018
545
0
660
I just asked my brother this. He said it's because the electrons carry heat through metal. But in (materials like diamond) the heat is moved by vibrational waves. That for some reason doesnt let the heat flow well. He said diamond is used sometimes.

He has 2 engineering degrees and actually works on the technology used to build the newest computer chips. But that's all I could get out of him. He's not an expert on everything. So maybe theres something missing. Outside of work he doesnt care to talk about computers. I was surprised he gave me that much info.

Edit: I don't understand it. But maybe that helps
 
Upvote 0 Downvote
izoli

izoli

Distinguished
Apr 29, 2011
597
0
19,210
"The short answer is NO not when regular carbon fiber is made up in regular epoxy and expected to conduct heat across the thickness. If a highly carbonized pan fiber with graphite or diamond added, is measured for heat transmission in the length of the fiber it is very good and can rival and exceed copper."

https://www.christinedemerchant.com/carbon_characteristics_heat_conductivity.html

My bet is to get it to the point of exceeding what we currently have, it's incredibly expensive?
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
the only demo i can think of, tomorrow i'll see if i can get into the shop and heat a pc of CF cloth with the torch and video record it, and let you see how quickly it cools

i did some googling and found the same results dugimodo reported but that just doesn't jibe with what we saw in the shop
 
Upvote 0 Downvote
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,453
17,911
184,590


Cooling quickly like that does not equate to efficient thermal movement.

And the raw CF cloth obviously won't work. You'd need to epoxy it into sheets.
And they need to be THIN sheets. More surface area crammed into the same cubic cm.

At the thickness required, CF+epoxy would be pretty brittle.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510


actually you have it somewhat backwards - not thin sheets but the thicker, the more mass and the more capability to absorb heat - that's the reason the bigger the heatpipes, the more efficient the cooler is

the idea behind the torch and the glow spot is showing how quickly it transfers that heat out, so much so it's cool to the touch within a second or 2

and "Cooling quickly like that does not equate to efficient thermal movement." - again, exactly backwards - the quicker the more efficient it is at transferring heat - not sure how you can argue that

and last item "At the thickness required, CF+epoxy would be pretty brittle." I'm not sure how the "brittleness" factors into anything in the way of thermal efficiency. The fact CF is extremely rigid when laid up by itself. Sandwich some kevlar between the layers of CF and it becomes extremely flexible. But none of that factors into thermal efficiency
 
Upvote 0 Downvote
D

Dugimodo

Distinguished
Sep 22, 2011
1,393
11
20,115
You need the mass to absorb heat, but you need surface area to get rid of it - hence thin fins seen on almost every heatsink ever.
The heatpipes transfer heat from one place to another using other methods, but would not work well without all those fins attached to them.
 
Upvote 0 Downvote
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,453
17,911
184,590


Thin, as in the thickness of current fins on air coolers. We need that thinness, again, for more surface area to dissipate into the box and then get sucked out via the case fan.

A solid block of iron would absorb really well. And keep it inside the case.
We don't want to absorb the heat, we want to move it outside the case.

Brittleness = be really really careful as you're handling it.
A bent alu fin can be bent back if needed. A thin CF sheet will break.

If CF were a better material, we'd see some boutique maker selling one.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
"Thin, as in the thickness of current fins on air coolers. We need that thinness, again, for more surface area to dissipate into the box and then get sucked out via the case fan."
The reason the fins are so thin is to get more fins into a given area, as you stated to get maximum surface area and to limit total weight of the cooler

"A solid block of iron would absorb really well. And keep it inside the case.
We don't want to absorb the heat, we want to move it outside the case. "
Negative, a solid block of iron would NOT absorb heat as readily. Accepting the heat though, is only part of the equation - it has to be able to transfer or move that heat, ie have a high transfer rate - that's one reason i question those websites - i can't reconcile how they rate cast iron so high relative to CF - you heat cast iron, you do not touch it for a considerable amount of time because it's so slow in cooling down, ie in releasing that heat into the atmosphere, same as stainless steel. That is what i am referring to as "transfer rate" - accepting it and then releasing it, whether into air, water etc. Talk to any welder about either SS or cast iron

"Brittleness = be really really careful as you're handling it.
A bent alu fin can be bent back if needed. A thin CF sheet will break."
I'm talking about CF for use in the square block that mounts to the IHS and in the heat pipes. As far as the fins go, they matrix other fibers into the CF weave, depending on the application - 20% kevlar fiber woven in would give the CF incredible flex, but most folks learn to handle the fin area carefully

"If CF were a better material, we'd see some boutique maker selling one."
The fact there isn't one already offering it is not proof against CF - as i stated, Nascar race teams are now using CF in cooling applications - it took them a long time, and i know. The same reason we didn't do a complete oil pan - the development cost, mainly the tooling, would have been excessively high, and there had to be a large enough market to amortize those development cost over a large enough number of units. We didn't see it, and had other more promising products in our sights.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
USAFRet - But i'm talking not only the IHS but the mounting block and the heat pipes. The fins wouldn't hurt either.

That example that i'm going to shoot the video of, getting a spot glowing cherry red (same as molten metal - you can do the same with any metal, get it hot enough to glow red where the torch is heating it), then pulling the torch away and being able to touch that spot with you finger and not burning any flesh. To be able to do that, means the heat is being conducted away, some of the heat energy into the air, and the rest along the CF fibers to the rest of the cloth. But remember i said it took effort (and time) with the torch finding the exact distance to hold the torch from the cloth to get it hot enough - during that time, a considerable amount of heat is already being introduced into the cloth.

So for it to cool down fast enough to be able to touch, means one thing - it's transferred that heat into the atmosphere. Try that trick with any thin sheet of any metal, whether brass, copper or aluminum. Aluminum would have the highest melt point, 1600-1700F (iirc), so you'd never get it to glow, instead the spot where you're focusing the torch flame would simply burn thru. Even so, once it burns thru, pull the torch away and you still can't touch the area surrounding the hole for a good bit - i'm guessing here but 30-60 seconds, definitely not 1 to 2 seconds the CF demonstrates.

 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
still haven't gotten around to shooting that video, but in the meantime came across this review here on TOm's referencing thermal pastes that employing different alloy particles (silver, aluminum, copper) and graphite & carbon particles. Graphite is simply a softer form of carbon - here's the Tom's review statement

"Some pastes, like Arctic Silver 5, even contain silver. Other pastes are based on graphite, like the professional-grade WLPG 10 by Fischer Elektronik. It foregoes the silicone and claims very high thermal conductivity (10.5 W/m·K), but is more difficult to apply and typically electrically conductive. There are also pastes that employ carbon nanoparticles, though they're not suitable for most enthusiasts due to their electrical conductivity and price. The number of copper-based pastes on the market has shrunk, but if you search, you can still find a few.
and the link to the page https://www.tomshardware.com/reviews/thermal-paste-comparison,5108.html

 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
carbon is carbon - carbon fiber is simply acrytlic that has been vitrifried to 5000F+ and pulled out in fiber form - it's still carbon, the clue is it's called carbon fiber - all the carbon nanoparticles are are dust made from carbon, carbon fiber, graphite- i could make some by simply sanding the carbon fiber cloth

and again as to the epoxy, i'll repeat what i said earlier, i initially had delayed or put off building a prototype thinking the epoxy would insulate the carbon fiber from the heat. When i finally got around to it, in a 94F environment, we put 450 pulses (1 second spurts of 1650F exhaust gas) thru "mufflers" - one fabricated entirely from aluminum, including it's baffles, the 2nd with a carbon fiber tube body (replacing the aluminum body).

At the end of the 450 pulses spaced about 5 seconds apart to build up heat soak, we then ran 50 pulses thru in rapid fire fashion, about 1 second apart

Measured with a calibrated non-contact thermometer (laser) the aluminum muffler showed 175F and the carbon fiber one 114F. But that was not just the surface temp of the muffler - as the aluminum bodied muffler took (and still takes) 12-14 minutes to cool down enough to put away in a nylon pouch, while the CF one is really ready the instant we were finished with the test, but to bring it down to near ambient temp didn't take a 1 to 1.5 minutes, which meant it was draining the heat from the internal aluminum baffles as fast as it was giving it off into the environment. And the carbon fiber tube was laid up in epoxy

now, either i'm lying, or imagining what that prototype demonstrated, or it's actually factual - but if you want to refute me, try not to just tell me i'm wrong

how those websites showed such poor thermal conductivity for CF i can't explain nor understand - i offered the same from toms thermal paste review as some proof that some ;companies knew that carbon has a high thermal conductivity
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510
never bothered making that video, too little time for stuff i need to do

but just came across this youtube vid (the poster is someone i've been exchanging with on another forum)
https://youtu.be/HfswixhCTMM

this was an experiment with CF that he did for his Masters degree - took an aluminum computer case (empty of components) and rigged it to pump heated air into, so he had a controlled btu feed into the case - after 45 seconds, temps inside hit 160F
identical case, all he did was stretch CF over the outside surfaces ( i suspect he bonded the CF to the case panels with some sort of spray adhesive, all i know is he said the CF had not been wetted with any resin)- same case, same heat source, after 45 seconds 91F - nearly 70 degrees cooler from just letting the CF take the heat from the aluminum

this should give some credence to the fact whoever posted that chart comparing thermal conductivity had no idea what they were doing
 
Upvote 0 Downvote
izoli

izoli

Distinguished
Apr 29, 2011
597
0
19,210


1. What type of aluminum is it, hell I don't even have a reason to believe it is aluminum.
2. He showed the inside of aluminum case, but never showed the inside of carbon fiber.
3. He never gave a reference temp for the supposed aluminum case.
4. He never showed when test began(first I saw was at 1:30 and it was already on) first time he pans over to temperature it has already been almost 20 seconds and it is at 90f.
5. Even if we assume it started right at 1:30, he still ran it for nearly 2 and a half minutes compared to the 57 seconds that he ran the carbon fiber case then compared those results.

The main thing being, how do I know that this isn't a steel case that he is calling aluminum? And considering the other factors how can it be trusted without having it tested again, properly?

Also even in 2009, 158F/70C was not a dangerous temperature for *most* CPU or Motherboards unlike he is saying in the video that people would have to unplug their system because it's so hot, which is sort of unrelated, but still affects my judgement.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510


No one has stated 158F was a dangerous temp - I can't tell if that criticism shows you were in a rush to criticize and distracted yourself, or trying to distract me, i'LL SAY THIS REAL SLOW, SO TRY TO READ IT SLOWLY -again he was demonstrating CF's ability to draw heat out of another material whethr air or another substrate such as aluminum, which was the experience I had when we fabricated CF ribbed plates for aluminum oil pans, as i described way up above, and the difference in firearm suppressors, with alum tube body showing 175F, same suppressor with aluminum baffles & spacer rings in CF gave us 114F

Izoli - like i stated above, you're only going toi know what you think you already know - the experiences i described happened, but if I were to make a video of a sunrise, I'm confident you wouldn't know it wasn't a sunset i was playing in reverse.

and here's something else you don't know -

oNx9the.jpg


that's one of my early CF suppressors, but you don't know i didn't just photoshop that or even stole from somewhere on the web

and here's another you don't know is real
G9ATRxX.jpg


you don't know that's me showing Mikhail Kalashnikov (designer/inventor of the AK-47) an Erma SR-100 precision rifle - he actually sent his secretary to our booth in UAE 1999 to make an appointment for someone to show him the rifle. As he left the booth after inspecting it, he said something in russian that sure sounded like "damn that capitalism driven technology". But you don't know that

Point izoli, what you don't know doesn't matter to me
 
Upvote 0 Downvote
izoli

izoli

Distinguished
Apr 29, 2011
597
0
19,210
He states twice in that video that at the temps seen in the aluminum case the motherboard would have shut down from heat and the user would have probably unplugged the power.

There could be dozens of different cooling solutions put in the CF case that would dilute measurements, to show 1 and not the other just makes it worse.

Im not convinced of anything, its just the opposite really. I am waiting to be convinced. This video in all honesty makes me less convinced.

No sir, the first temperature shown is 90f which jumps to 101f. Also on that note does it only take readings every several seconds or is it a low quality sensor?

Even if it isnt a big deal from your point of view, it is a big deal in terms of credibility. Steel cases are magnetic, could slap a fridge magnet on it and if it doesnt stick thatd be convincing enough for me to take at face value. Cheap, fast, painless. Not asking for much.

You could clearly indicate the direction you are facing with a form of compass while taking video of sunrise and I would certainly know if it was reversed or not.

I dont know anything about gun suppressors. Carbon fiber being an insulator rather than a conductor though, is it possible that the carbon fiber spacer rings are just insulating and the heat is trapped somewhere else? Have you looked with a thermal camera at the whole firearm or just measure the same 2 specific areas?
 
Upvote 0 Downvote
D

Dugimodo

Distinguished
Sep 22, 2011
1,393
11
20,115
It occurs to me you might need a bit more thermal mass but I'll admit that's a total guess, but maybe a copper block with carbon fibre fins attached could be worth investigating. A certain amount of mass helps stabilise things so it doesn't heat up or cool down too rapidly and the system has time to find it's equilibrium. Even with a really good heat conductor transfer is not instantaneous.

Maybe it's a great Idea and nobody has thought to try it yet, or mabe it's too expensive and/or difficult to manufacture in a suitable form, or maybe it just doesn't work. I find it interesting but not enough to be thinking how to build a prototype.

I think too some of your related experience tells us the carbon fibre (just realised I'm not spelling it the american way but I'm not going back to change it :)) cools down very quickly itself, but not how well it conducts heat away from other materials. A thermal insulator could exhibit similar traits to some of what you describe while still not being any good as a heatsink. I couldn't watch the video link though, it's blocked from where I currently am (I have restricted internet access)

If you heat one end of a piece of carbon fibre does the whole piece get hot quickly? A very basic test to demonstrate conductivity easily carried out if you have access to some.
 
Upvote 0 Downvote
The Original Ralph

The Original Ralph

Distinguished
Sep 14, 2011
2,222
1
20,510


buy yourself some carbon fiber and wrap yourself in it next blizzard comes thru - if you're down south then where it in the freezer section of an ice cream factory and you'll understand why i choose to just ignore you

i had a dozen engineers work for me over the years, most were young but what i noticed most had in common, like you they always knew why something wouldn't work when you'd ask them for a solution. My response was always, i didn't ask you why or what wouldn't work, we need to get to point x, find the solution.

I finally found an older engineer, that understood his role wasn't to tell me we couldn't get there from here.

Telling me CF is an insulator tells me all i need to know
 
Upvote 0 Downvote
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,453
17,911
184,590
Just to be clear...

We're no longer talking about the whole "CPU air cooler" being CF, as stated originally.
But rather just the block that contacts the CPU, and then to the actual dissipation material, the fins or pump and liquid.

Correct?
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom