Watercooling a heatpipe?
 

http://www.frostytech.com/permalink.cfm?NewsID=42017

Im guessing the internals of that thing is just a heatpipe. npowertek doesnt really do anything but make heatpipes for oem solutions. They have another version on their front page that has fins on the outside, dunno what thats all about.

What would the performance of a design like that be? Just a heatpipe with a copper pipe around it.

Separate each part of the heat transfer, Die->TIM->Cu->Heat transfer medium->Cu->water.
Copper is copper.

The idea here is to use the heatpipe to spread the heat over a greater surface area. The build of the block should be fairly simple and inexpensive. The fact that there is an additonal thermal layer present (the heatpipe, they can only be so efficient) makes me wonder as to how well it will perform. I would guess that this is designed for low flow systems.

Yeah, really small fittings on it too. Im not really sure if its a heatpipe inside. But Id bet a good chunk of change that it is because of the outer design and that all the products they have made before all use their heatpipes.

Wonder how it would perform if there was like a disc type design of fins around the heatpipe inside, like what innovatek did with that small rev3 block.

Well, I don't think its a heatpipe inside. Probably another design with a big chunk of copper and a maze of tunnels like the Cuplex Evo that acquacomputer makes. I never cared for that design.

ok?


Looking at their front page I realized they titled the waterblock "NPH-WB-1 Heat column water block." Sounds like its just a heatpipe in a waterblock. From the availability of their heatsinks Im guessing this will be just as hard to get a hold of. Let alone find a review on :shrug:

One thing I've considered before is using heatpipes to move heat from hard-to-wc components such as mosfets and VRMs on motherboards, Power Supplies, northbridges, and the like. I would think you could use a really simple low-resistance block (or even copper pipe) to cool these heatpipes and avoid adding a lot of blocks and resistance into your loop.

The problem is that you'd have to have an estimate of heat load for each part to be able to spec out the heatpipes. It's a good idea otherwise I think.

Well that just gave me an awesome idea for cooling my mosfets, thanks pHaestus. Trying to make a box fanless and couldnt figure out how to watercool my vrms which are stuck between some tall caps and inductors.

picture

1/2" pipe connected to a copper plate.

Make a U shape and it should go around it... or, make a U shape to connect to pipes with the copper plates on the bottom

from above

That was a response to starbucks post mostly. I fail to understand how this is better than just using a standard water block (for CPU's anyway).

I don't see the advantage for CPU's. One could argue that it was cheaper for them to manufacture as all they had to do was modify one of their existing air cooled heatpipe HSFs by cutting out some of the fins and adding a top to seal in the water.

That just gave me an idea....

I have always thought that heatpipes routed to a low restriction block would be a great way to cool RAM. I've just never had the time to sit down and figure out the mounting mechanisms. It could be done though. If someone could design a simple, universal block for cooling RAM, mofsets, northbridge, PSU, etc. they would really have a hell of a product. The more components in a system that are water cooled, the more cost effective the system becomes as the cost of the pump and rad are spread over multiple components. The other advantage to having water cooling over a large variety of components is that the computer can potentially be more quite, more stable, and much smaller (ever thought of building an ATX based computer into an aluminum briefcase? It can be done, though the thermal management aspect leads one directly to water cooling.)

That's my opinion, though it is obviously up for debate.

Can't be done actually. To many different layouts.

The point of using heatpipes would be routing heat from the source to the "universal" block. By using various heatpipes I'm sure there is a way to make this work. Do you know if heatpipes can be bent without an appreciable loss in performance? If so, an manufacturer could sell individual heatpipes in various lengths and thermal capacities. These could then be bent by the end user so that they could be routed to the "univeral" block.

Don't know but that it not a universal block. My Abit KD7 for example would need 8 different blocks at about 3/8" x 3/8" just for the mosfets as many of them are in between caps. Many motherboards have mosfets on back of the mobo. Power supplies? They would have to have several hundered diffferent blocks in stock for every combination there is. Just isn't possible on a business level. It would have to be system spacific for sure.

I don't see why you'd water cool it just stick some fins on teh end of the heatpipes and viola.

I'm also wondering why larger manufacturors would use wc'ing over heatpipes into a small rad in the back.

Your missing my idea here, JD. The flexibility of the heatpipes allows for the use of only one block. You would attatch the heatpipe directly to the heat source. The heatpipes would then attatch into slots on a "universal" block. One might use different size copper blanks to attatch to mofsets, nb, gpu, etc. but this wouldn't be a big deal to manufacture. The end user would then use thermal epoxy to hold the heatpipe in place. Mounting of the "universal" block I haven't thought of. Such a system, if properly designed, would have large applications outside of computers, btw.

edit: I forgot to mention that the heatpipes would be bent to route from the heatsource to the "universal" block.

I know what you mean. One big block with a bunch of smaller blocks piped to it. Pretty much impossible. Pipes can't bend nor can they be length adjustable if used like the products in this thread (have to have the solid material inside them to transfer the heat, the pipe itself is not transfering the heat). Other alternative is phase change which the tubes probably could be bent but they are not adjustable in length as they have gas in side them.

Even if it was possbile no manufacture on earth would consider such a thing. Your talking 10's of millions of dollars. Being 99% of the computing world uses air cooling there is no way they would make profits off that. They can't even make much profit off current cooling items.

Work on it for a while and prove me wrong.

you mean the heat columns like the ones the npowertek use cant be bent? I can see that. But normal heatpipes are easily bent. Just a metal semi-soft wick inside. I bent the heck out of the heatpipes in a TT silenttower to fit a custom heatsink. They went from straight to a gradual 90degree bend. Just by hand.

I belive I said that. Still dosn't cover the length issue. Each system will have to be completely different along with different length.

Does it not compute in your guys heads that there are hundred of thousands of different configured systems? All having different parts and all laid out completely different. It is hard enough trying to make a universal water block for every type of vid card there is let alone every part on a computer of every kind. Even popular stuff.
Hell look at CPU's. There are how many different sockets? I got 4 boards here and all would require custom fitting of such a heat pipe system. In order for this to work a company would have to build the computer and make the parts to fit it and sell that computer. No way they can make a universal heat pipe system to cool every part on all computers. I would challenge them to make a universal system to just cool the CPU and vid card for all systems.

Yeah fully agree with the non-standard mobo layout preventing any sort of retail heatpipe vrm cooling kit.

Was just commenting on the bending heatpipes thing.

I was thinking of using a heatpipe to cool my vrms but, just as you mentioned, the length issue is preventing me from doing that. Id only need like a 4" heatpipe.

Mentioning the zalman fanless box is note worthy. It supports a lot of boards that have the CPU socket in different locations while not requiring any bending of heatpipes, just sliding the pipes around. Not really a cheap solution for everyone but that sort of design can be applied elsewhere.


But I dont think getting the heatpipe a certain length and position would be the limiting factor. I think mounting them to VRMs or memory would. Not really any option other than epoxy there. bye bye warranty. course replacing the CPU heatsink voids a lot of warranties also. I think there are a lot more people willing to unclip a heatsink then there are those who are willing to permanently glue something to their mobo though.

The Zalman box is such a system as I described above. A prebuilt computer with a cooling system built around it. Sure that is possible and doable but certainly not universal.

Length is a limiting factor. How many people can fill a pipe with phase change material after they cut it to lenght and sealed it. An end user capable of that would not need to buy such a product to begine with.

And yes mounting is a problem. There is no company in existance that would recommend epoxing stuff to your parts. That would also make your system non upgradeable. Also they will have no part in suppling stuff that requires you opening your power supply.

...looks like another idea shot down...

It really is a shame that motherboard and PSU makers can't agree on some set of standards. What about using an amorphous cooler? Something similar to what...gulp... koolance is using for their hard drives?

Well I wouldnt want mobo manufacturers to be forced to use some standard layout. We'd see a lot less stuff like additional SATA controllers and nics and 7.1 integrated sound cards etc...

Minimizing trace length and keeping component costs down is a fine art. Shouldnt give it even more restrictions.

Ok, so I haven't been around fo a long time, but I did do some calculations on heat pipes a while back.

Anyway, if you break a heat pipe up into a series of thermal restrictions, you get the following

1. Conduction from outside of heat pipe across copper wall
2. Vaporisation of heat transfer media
3. Transport of vapour throughout heat pipe
4. Condensation of heat transfer media
5. Conduction from inside heat pipe across copper wall.

The beauty of a heatpipe is that it can move heat from a small heat source and transfer it to a large heat dissipating area.

However, the first 2 stages listed above are not improved if the size of the heatpipe and the attached cooling area is increased. They are determined by the size of the heat source. The 1st stage is relatively small due to the copper walls being thin. The problem is the 2nd stage. While doing calculations on a waterblock incorporating a thermabase, I discovered that this step was limiting and for most processors, was already MORE of a restriction than some high performance waterblocks.

In other words, heat pipes will not rival watercooling for performance applications, but may offer some benefits for silent applications as with zalman.

The value of watercooling a heatpipe is negligable.

I hope this makes sense.

8-ball