How come no CPU coolers w/heatpipes and vapor chamber?

[claes]

Definitely not always

 

[Mussels]

Heatpipes (and vapor chambers) have higher thermal conductivity than solid copper block of the same size.

In specific circumstances only. They need to be in set size ranges compared to what they're attached to for example - and good luck with that and CPU's with chiplets, you'd need a different heatsink for every model of CPU out there.

It could be implemented at an IHS level, but the costs would be prohibitive - imagine if the IHS vapour chamber cracked on otherwise good CPU

 

[claes]

You’d also have to make a giant IHS in proportion to the doe for there to be any advantage over metal

 

[Mussels]

You’d also have to make a giant IHS in proportion to the doe for there to be any advantage over metal

Zen3/4 dies are pretty small, but yeah - the size issue is a big one.

 

[rockit00]

The Asetek Vapochill Micro CPU cooler from 2005 had vapor chamber and heatpipes IIRC. It got a decent review on TPU back in the day!

 

[GreiverBlade]

@LabRat 891
The N8 Dynatron sounds impressive: 270Watts dissipated in such a small package (113 x 78 x 64 mm) -- passively! The NH-D15s is only rated to dissipate 167 watts.

interesting my Enermax ETS-T50 is rated for 230w and it's a 5 heatpipe direct contact single tower with one fan, i suspect they are a bit liberal with the TDP but it's techncially "only" 3°c hotter than the NH-D15s
while considerably cheaper 120mm and single tower

 

[freeagent]

That Dynatron would be in an environment with many, many cfms blowing through it.

 

[80251]

That Dynatron would be in an environment with many, many cfms blowing through it.

You're not kidding either, I've seen and heard the fans that run in 2U enclosures, dual, contra-rotating blades, monstrous RPM's and mucho noise. The ones I saw were only 92mm wide, but really long to accommodate the contra-rotating blades. There are wind baffles that channel the airflow directly across the heatsink(s).

 

[LabRat 891]

In specific circumstances only. They need to be in set size ranges compared to what they're attached to for example - and good luck with that and CPU's with chiplets, you'd need a different heatsink for every model of CPU out there.

It could be implemented at an IHS level, but the costs would be prohibitive - imagine if the IHS vapour chamber cracked on otherwise good CPU

IIRC, (ambient-terrestrial) Heat Pipes/Vapour Chambers drastically lose performance in "Winter Temperatures" and can 'heat soak'.
Basically, if you use a heatpipe outside of its expected operating temperature range, it works much worse than solid metal. Older ones also had issues with orientation, and I think even modern-manufacture examples often have minute but repeatable differences in performance (based on orientation).

AFAIK, allotropic carbon has neither of those issues but, it's (usually) a mediocre conductor in the Z-axis (vertical, thickness)
-been wondering if Synthetic/Pyrolytic Graphite / Graphene could be used in lieu of a vapour chamber or heatpipes?
The material itself is already fairly common in(on) small-devices like Phones, Laptops/Notebooks, SSDs, RAM, etc.

 

[freeagent]

IIRC, (ambient-terrestrial) Heat Pipes/Vapour Chambers drastically lose performance in "Winter Temperatures"

About -15-20 or so..

 

[80251]

IIRC, (ambient-terrestrial) Heat Pipes/Vapour Chambers drastically lose performance in "Winter Temperatures" and can 'heat soak'.
Basically, if you use a heatpipe outside of its expected operating temperature range, it works much worse than solid metal. Older ones also had issues with orientation, and I think even modern-manufacture examples often have minute but repeatable differences in performance (based on orientation).

AFAIK, allotropic carbon has neither of those issues but, it's (usually) a mediocre conductor in the Z-axis (vertical, thickness)
-been wondering if Synthetic/Pyrolytic Graphite / Graphene could be used in lieu of a vapour chamber or heatpipes?
The material itself is already fairly common in(on) small-devices like Phones, Laptops/Notebooks, SSDs, RAM, etc.

But doesn't Newton's law of cooling state that the greater the temperature differential the better the cooling? Sure the heatpipe may lose some efficiency, but who's to say it isn't offset by the increased cooling capacity of cooler air?

 

[Count von Schwalbe]

But doesn't Newton's law of cooling state that the greater the temperature differential the better the cooling? Sure the heatpipe may lose some efficiency, but who's to say it isn't offset by the increased cooling capacity of cooler air?

Not if the fluid inside cannot vaporize and condense properly.

 

[Bones]

Heatpipes (and vapor chambers) have higher thermal conductivity than solid copper block of the same size.

Not really because it also takes a given amount of mass to absorb and move the heat along.

A copper block can and will move more thermal energy because of it's extra mass, meaning it takes more heat to thermally saturate the mass of material before it stops wanting to absorb more heat to pass along.
Another thing related to how quickly it can release the heat to atmosphere so more heat can move up and be dissapated too - Every link in the chain from actual source (Silicon) to destination (Atmosphere) matters and it's only going to work as well as the weakest link in the chain.

Not to mention the more links you have the less efficiency you'll tend to have too.

That's why for example you don't see pots for Ln2 made with pipes - EVERY bit of cooling efficency matters and best results are obtained from a solid piece instead of things like pipes because there is more mass present for it to do as described.

But doesn't Newton's law of cooling state that the greater the temperature differential the better the cooling? Sure the heatpipe may lose some efficiency, but who's to say it isn't offset by the increased cooling capacity of cooler air?

Yes that's more or less "It" pertaining to cooling efficency.

When it's fully saturated it's less efficient because at that point, the material is "Overwhelmed" or "Overloaded" in how much heat energy it's handling at that time. If the material is already close to the same temp as the CPU itself, it's not going to attract much heat to move along but if the material is noteably cooler than what it's trying to cool then it just works better.

I have some coolers that have both, the baseplate/block with pipes and just the pipes themselves at the point of contact to the CPU. The ones with a plate in my experience can handle a heavier heatload and that's partially due to the surface it has making better contact overall.

I've had a couple of the direct contact heatpipe coolers not do well and one had a pipe that was "Sticking Out" below the others causing very poor contact across the chip's surface.
It's easier and cheaper to make a direct contact cooler but it's also not quite as good overall - I do not prefer them at all and based on my own results and issues I've ran into, I'll keep using the ones with a plate/base made to them.

 

[joemama]

IIRC, (ambient-terrestrial) Heat Pipes/Vapour Chambers drastically lose performance in "Winter Temperatures" and can 'heat soak'.

I think that should be summer actually and it depends on the temperature and fluid type.
For the heat pipe design, it requires the hot spot's temperature over the boiling point of the fluid and the cold spot's temperature under the boiling point so the fluid can circulate through continuous evaporation and condensation. When the cold spot (usually at the fin array for coolers) is too hot, the fluid doesn't condense and thus no circulation, making the heat pipe just a hollow copper tube with awful thermal conduction.

-been wondering if Synthetic/Pyrolytic Graphite / Graphene could be used in lieu of a vapour chamber or heatpipes?

I don't think those could help since although they conduct heat fast but that is compared solid materials. Heat pipes rely on the convection of fluid so the heat transfer rate would be a complete different level.

But doesn't Newton's law of cooling state that the greater the temperature differential the better the cooling? Sure the heatpipe may lose some efficiency, but who's to say it isn't offset by the increased cooling capacity of cooler air?

Isn't that Fourier's law? Anyway, that's for thermal conduction, which is heat transfer within substrance without it moving, yet heat pipes majorly rely on the flow of the fluid inside and that's what makes it conduct better than solid materials.

 

[80251]

Not if the fluid inside cannot vaporize and condense properly.

With no facts or research data this is just idle speculation.

No it has nothing to do w/Fourier. Newton was the first man on earth to use a differential equation to model a thermodynamic system.

I haven't noticed my Noctua NH-D15s cooling LESS effectively at 60°F than it does at 80°F.

 

[Count von Schwalbe]

60°F

Is not outside of its temperature range. Try comparing 0 and -30 degrees.

 

[joemama]

With no facts or research data this is just idle speculation.

You want something like this?

https://www.sciencedirect.com/science/article/pii/S0895717711003888

 

[claes]

I think that should be summer actually and it depends on the temperature and fluid type.

It’s both, depending on the design

 

[Ruru]

I'd guess that the performance difference would be so marginal that it's just not worth it. Practically we've somewhat achieved the best that an air cooler can do.

 

[claes]

Don’t think you can make that statement without including water coolers. Unless you have space for a 360mm radiator (or, let’s be real, a triple tower), we’ve reached the thermal limits of atx and chipset design.

 

[londiste]

Not really because it also takes a given amount of mass to absorb and move the heat along.

A copper block can and will move more thermal energy because of it's extra mass, meaning it takes more heat to thermally saturate the mass of material before it stops wanting to absorb more heat to pass along.
Another thing related to how quickly it can release the heat to atmosphere so more heat can move up and be dissapated too - Every link in the chain from actual source (Silicon) to destination (Atmosphere) matters and it's only going to work as well as the weakest link in the chain.

Not to mention the more links you have the less efficiency you'll tend to have too.

That's why for example you don't see pots for Ln2 made with pipes - EVERY bit of cooling efficency matters and best results are obtained from a solid piece instead of things like pipes because there is more mass present for it to do as described.

Heatpipes use phase changes, boiling and condensation and those make the whole thing quite efficient. They are limited in terms of heat capacity, temperature range, directionality etc but all that is not relevant for their performance in what they are intended for. The use case of a heatpipe (or vapor chamber) is to move the heat. Either spread it locally to a wider area in case of vapor chamber or to move it further away from where it was generated in case of heatpipes. That is exactly what they are good at and high heat capacity would actually be detrimental here.

Releasing heat to atmosphere is already the next step and normally done by fins of some kind.

 

[shoskunk]

Most of the replies hit the mark. Space and weight.

I'm pretty convinced we've already close-to-maxed-out the wicking properties of chambers and pipes with new cpus and gpus.

You don't see more of them because they don't return liquid to a chamber or set of pipes that is already heat saturated.

 

[damric]

Prior to 7nm, my TPC coolers were pretty awesome. Well actually my TPC-812 could indeed passively cool my Ryzen 3600 with small overclock (4.3), but just barely without throttling. It seems like heat density is so severe now that water is needed. Except Cezanne which the monolithic die is cooled quite well with my TPC 612.

The problem isn't them getting saturated, because the fins barely even feel warm to the touch. It's the hellish heat flux density and getting that through the IHS and into the baseplate to the vapor.

 

[Mussels]

With no facts or research data this is just idle speculation.

No it has nothing to do w/Fourier. Newton was the first man on earth to use a differential equation to model a thermodynamic system.

I haven't noticed my Noctua NH-D15s cooling LESS effectively at 60°F than it does at 80°F.

They do have limits, but they're a curve on a graph not a yes/no situation. 20c less ambient making it 5% less efficient means nothing because its 20c cooler.
If 80% of the length of the heatpipe is at boiling temp for the fluid, it's not travelling the entire length any longer.
Same if its too cold to boil, it sits at the bottom. We can't know those ranges since the fluid and pressure used control that.

Heatpipe fluid can send some of the heat to the far end of the heatsink faster - it's used as a cost saving measure (less copper total) but a way to move some of the heat through the pipe a little faster and get some quicker benefits from the tallest fin stacks, because normally copper has to heat up evenly.
It's really only a way to transfer small bursty bits of heat and not going to help with maximum temperatures. The mass of the copper does that, not the fluid.

You can think of a heatpipe (or vapour chambers) best benefit with something like the 5800x and its heat density issue - it can transfer the heat to a larger cooling area much faster, but only if you have multiple heatpipes perfectly lined up to transfer it evenly. This works on a GPU because each cooler is die specific - and why we have issues with AMD's chiplets and intels E-cores, without a solid plate to spread the heat to the pipes in the first place.


The fluids used are likely either proprietary knowledge or plain old distilled water - they dont share it.

If you want heatpipes in a below freezing situation, you'd use a different fluid. PC coolers wont do that because condensation would kill the PC's in those ranges, so they can optimise them for the lets say 20-100c range

 

[Bones]

Heatpipes use phase changes, boiling and condensation and those make the whole thing quite efficient. They are limited in terms of heat capacity, temperature range, directionality etc but all that is not relevant for their performance in what they are intended for. The use case of a heatpipe (or vapor chamber) is to move the heat. Either spread it locally to a wider area in case of vapor chamber or to move it further away from where it was generated in case of heatpipes. That is exactly what they are good at and high heat capacity would actually be detrimental here.

Releasing heat to atmosphere is already the next step and normally done by fins of some kind.

Yes, I agree and understand they are indeed efficient.

It's when you need something beyond efficient is when the other kicks in.
Heatpiped coolers work well for the purpose they are intended for but at the same time they can be limited in the amount of heat they can move vs one with a base to them, which is important if dealing with a chip of a higher wattage/core count which equals more heat to remove - Esp if the chip will be running under a heavier, sustained load.

That's when a cooler with a base works since it's easier to thermally oversaturate a standard heatpiped cooler under a sustained, heavier heat load.
When the fluid inside has all gone to a gas due to thermal saturation, that's it - It loses alot of it's ability to pass heat along until at least some of the gas can go back to a liquid so the process of liquid to gas can start over again, restoring it's cooling efficency.

One with a base does a little better because the mass of it isn't affected in this exact way.
It still keeps trying to pass it along at it's max thermal dissapation rate (BTU) even if it's thermally saturated so that's one difference between them.
It's also harder for a standard heatpiped cooler to control thermal spikes under load which one with a base can more readily soak up and pass it along, meaning there is a tendency for less of a temp spike/swing when it happens.

You can get one with more pipes and that would help but that's also when a cooler can start getting to a ridiculous size (By it's sheer volume) because of the liquid to gas expansion requirement it has for it to work in the first place.
One with a base can be affected this way too but they tend to be of a smaller size overall, esp if they are designed well and the fins are made properly too of the correct material but, TBH that's really all of them.

For most anything used as a common desktop setup, heatpiped coolers are fine but I still prefer a cooler with a base.