Hi!

Did some reading about heapipes a while ago and as i remember, the condenser-side has to be situated above the evaporator-side when using standard heapipes as in this article.
Don't remeber the minimum angle but i think there was one.

Was different positioning of the condenser and evaporator tried?

I´m sure i will be corrected right away if i´m wrong about this


/Dm0Nx

well yes think of any of the heatsinks that use Heatpipes they are all at a 90 degree angle to the base. so i would say that they dont need to be higher.

Hmmm...well i guess technolgy moved forward....when i read about heatpipes some time ago the only heatpipe stuff that allowed strange positioning of the condenser/evaporator was the TS Heatronics Heatlane stuff http://www.tsheatronics.co.jp/english/t ... index.html

Apparently thats all changed now...

quote:
"A heat pipe is said to be operating against gravity when the evaporator (heat source) is located above the condenser. In this orientation, the working fluid must be pumped against gravity back to the evaporator. This is in contrast to gravity-aided operation, where the heat source is located below the condenser, and the recondensed water is returned to the evaporator by gravity.

Thermacore uses a sintered wick structure in their heat pipes. The sintered wick is uniquely capable of returning working fluid to the evaporator against gravity by using capillary pumping action. Screen and grooved wick structures are not capable of returning the working fluid against gravity at angles greater than -5°.

The finer the pore radius of a wick structure, the higher against gravity the heat pipe can operate."

Still interested if the positioning does affect the results of the pipes in this article.

/Dm0Nx

Older heatpipe design used to require gravity return of the condensed water; but, that was a couple years back. Most companies today are using sintered powder cores that allow use at any angle. Our tests here at Aerocase have shown less than a 10% drop in heatpipe efficiency when placed in the down (reversed) position.

SG

The Borg heatpipes are designed for use in the HFX case, which suggests that they should be effective in a horizontal orientation. It's hard to believe that mCubed would have released them if they weren't...

What would be interesting to see tested would be minimalistic water cooling system being used to pump the heat with the same testbed. Once I get my bed done, maybe I'll do it, unless someone else does it before me..

In forced cooling it's usually better to use radiator as water to air heat exhanger, but in case of passive cooling there is no real difference here, except that with water it's easier to pass the heat to higher surface area. But it would be interesting to see the thermal loss while using water when all the other components are same/similar.

In fact, I guess you could even use the mCubed heat-plates if you picked narrow enough copper tubing and passed water through it. It'd make the loop a lot more restrictive, but with pump with high head pressure and this small loop it shouldn't be an issue.

Naturally if the heatpipes do the job, in so small case as that Hiper, they most likely make sense. But just out of curiosity and for larger enclosures it would be nice to have some comparisons.

I also see no reason why you couldn't combine those two.. I in fact plan to do so for the components producing less heat, ie. use heatpipes to gather the heat to a single waterblock and then run water through it. This should provide sufficient cooling for regulators, main memory, southbridge etc.

It'd also be very interesting to build water-cooled system to that Hiper case.. That'd certainly be an excersice in miniaturization . But given the flexibility of water cooling, it'd certainly be possible.. even some good silent pumps measure like 80x80x31mm (DDC with acrylic top), which is less than most 1U heatsink+fans, and something like 1/3 of all the waterblock would nicely with within 40mm height, lowest needing mere 11mm of clearance (Koolance GPU-180-H06).

nice test setup:)


in the editor's note, please check the product name , should be thermaltake, not thermalright

In Brendan's test, both the heat source and the Thermalright heatsink/fan are intrinsic to the test. Changing the HSF could have a significant effect on the measured °C/W value of the system and the apparent heat transfer of the heatpipes, just like lowering the heat at the source did. Because a heatsink with greater heatsink area (say, the Scythe Ninja or Thermalright Ultra-120) was not tried, we can't know for sure whether the heatpipes or the HSF was the greater bottleneck.

Thanks jmke!

With regard to the angle of the heatpipe and its effectiveness, I did actually tilt the whole test-bed and there was very little change in the results. The sintering material used must do its job well.

The greatest "suprise" I had using these heatpipes was how easily bent into position they are. For tight bends I did use a pipe bender however, as the pipe sides did tend to "squash out".

argo = Brendan I suppose?

do you think you could create a case similar to Zalman TNN series with a large collection of these heatpipes? Would it be more cost effective?

If you had four heat pipes couldn't you run two pairs to two seperate heatsinks?

If you had a large case with two 120MM fan mounts at the back, you could mod the heatsinks to the exhaust fans. Dumping the heat directly out of the case.

Good "guess"

Yes a similiar case to the TNN could be built, in fact IMO mCubed have done a great job with their heatpipe kits which would greatly assist with this type of project.

Cost effective, hard to say. Each D6 pipe was ~US$9 and Austrian post to NZ was $15 for four pipes. They also sell the heatsinks which would be expensive to ship (heavy!). If you just bought the heatpipes from mCubed and sourced the heatsinks yourself (or made them yourself) from say a second-hand computer/electronics parts warehouse then sure!

BTW to find mCubed do a search on "hifiatx", which is model series name mCubed use for their passive-cooled cases.

I saw that case at Cebit

here's their PR for that case: http://www.madshrimps.be/forums/showthr ... adid=13152

So you did - JM!

I know styling is in the eye of the beholder, but out of interest what did you think of its looks?

I don't really care about looks, for me it's the thermal performance that counts, if it has killer performance buts looks horrible I'd still buy it and just hide it away

I'm the wrong person to ask...


but there are several companies out there which do what you suggested in the article, they use it for low profile cases for the CPU cooling, I saw several products which used heat pipes to get the heat from the CPU to heatsinks at the side of the case.

I can't exactly remember the stand, Silverstone I think.

I wonder when this system is inserted inside a closed case, whether it would help to insulate the pipes at their midpoint so that the pipes are not dissipating heat back into the case near the cpu.

Ultimately a good deal of the efficiency will be tied to how cool you keep the cool end, so a water cooled option would help, but for a slimline case it runs counter to saving space, same as with a tower HS like the Ninja.

If space allowed, perhaps running a single pipe directly to each of the 4 sides of a case could work, or 2 to the 2 closest sides. If the case is designed to dissipate heat passively that might be the best bet. If all 4 pipes go to the same point, that spot on the case will be hotter and the temperature difference would be lower, making it less effective at moving the heat, wouldn't it?

Thanks for the great review!


This is a good idea, do you have a suggestion of what to use to insulate?

I am building a passive HTPC case with the borg coolers for the processor and the graphics card, you might find inspiration or answers if you are building one: Stylencio: Passively cooled custom designed HTPC

It's not really necessary to insulate it, as most of the heat is given up when the vapor condenses at the end of the heatpipe, but if you wanted to you could use some kind of closed-cell foam covered with reflective aluminium foil.

Would it be possibel to test pure aluminium and copper "pipes" to see the difference?

From the Editor's Note:

[i]When thinking about heat transfer, it's useful to consider thermal pathways, points of constriction, and the speed of heat transfer. The fact is that a heatpipe or any thermally conductive path can transfer even thousands of watts — if the pathway is narrow, it will just take longer[/i]...

Here's a quick analogy:
Temperature -> voltage
Power -> current
degrees/Watt -> resistance

So degrees/Watt is thermal resistance, exactly analogous to electrical resistance. The editor's note is saying that a resistor can transmit arbitarily large current, so long as you apply sufficient voltage. (In the real world, this only works so far, and then the magic smoke escapes.)

In a computer, you typically have a given quantity of heat-power to dissipate, into air at a given temperature via a heatsink/radiator. The electrical equivalent is a constant current source (CPU heat) via a series of resistors (thermal compounds, heatsinks, heat pipes, convection etc.) to electrical earth (the air in your room.)

In a typical computer system, you'll have something like the following resistances: Silicon chip to a heat-spreader (part of the CPU packaging) to thermal compound to heatsink base to heatpipes to radiator fins to air inside the case to air outside the case. The resistances are in series, so while any improvement is useful, it is generally most cost effective to deal first with the large resistances. There's little point in paying big money to upgrade the thermal compound resistance from 0.03 deg/W to 0.02 deg/W if is followed by a cheap 1 deg/W heatsink, or to put in a highly efficient heatsink and then not to ventilate your case - this will merely ensure the air trapped in your case is just as hot as the CPU.

Hi Guys

Couple of practical points I found when using the heatpipes. First is that there is no need to think of heatpipes as having a cool end and a hot end, in fact forget about the pipe having an end at all (other than when the end is to short to reach!). Simply put I noticed energy transferred from the hotter part to the cooler part (no suprises there!), no matter where this was on the total length of pipe. The significance of this is that the cpu and hsf can be clamped anywhere on the pipe, it does not change the performance. I found when using the full length of the pipe, the copper surface felt (using my skin as a thermomenter) equally hot at any given point between the cpu-sim and hsf. A simple way to think of this is that my hand was cooler than the pipe it was touching, so energy (heat) travelled into my hand.

So to answer bruenor's question in theory lagging the exposed pipe would be good. However practically the pipe area exposed is fairly insignificant in the scheme of things. Also remember air is very good insulator (the reason expanded plastics are used as insulation).

Another thing to remember from school physics is that heating and cooling is generally measured in Watts (sometimes BTUs). However Watts is Joules per second. To paint a picture: think Joules as the amount of water in a tank (ie the "heat" or energy), temperature as the pressure of a water pump and Watts as the diameter of a hose moving water out of the tank. To empty the tank faster we can: increase the pump pressure (increase temp), get a bigger hose (a more effective heatpipe), or start with less water (Joules). To complete the analogy our CPU is sitting over the tank dumping "water" in - while we are trying to empty this tank! If we can't get water out of the tank equal or faster than the CPU is dumping it in, then the tank overflows (CPU overheats). The heatpipe is our hose!

Actually this analogy fits the observation of feeling heat on the heatpipe. If our touching the heatpipe is thought of as causing a leak in our hose, we will see water (feel heat) escaping the hose.

Hope this makes sense and helps

McBanjo to get an idea we can look at the thermal conductivity (W/mK) of these materials:

Copper = 400
Aluminium = 237
Heatpipes ~ 7,000

Heh ok, easily solved then
Thanks

Since the pipe felt equally warm along the entire length, and as you said it will dissipate heat at any point if that's where your cooler is located, I wonder then if a pipe were wrapped around the entire case, with the ends coming together at the cpu would work well. You'd have the pipe clamped/fitted along all 4 edges of the case drawing heat off into the outer body. Would that be like having a series of (albiet weak) heatsinks along the entire length of the pipe?

Another way of asking that is to say, would performance improve as you increase the amount of pipe in contact with a cooling surface, whether it be the outer case or an active cooling system?

Even though Brendan is correct is saying the the pipe feels equally hot everywhere, I think it is important to remember that that the heat starts at the CPU end. The coolant recycles wherever it gets cooler -- which is wherever the pipe is "sinked" to cooling fans, airflow, etc. The fact that it feels equally hot everywhere is not surprising, given the extremely fast heat transfer of a heatpipe.

breunor -- the answer is yes, at least in theory. The significance or degree of improvement would be subject to many other factors, however. It's another one of those "silence (or cooling) is in the details" thing. But Brendan may have done other testing not reported on in his artilce that could shed more light on your question.

Hi argo, good article, thanks

People might be interested in an alternative to the M-Cubed range by Aavid Thermalloy, and which are sold HERE... except that the on-line store is down at the moment.

Evaporating water can carry a truly gigantic amount of heat (540 cal/gm). However, it only works if there is water there to evaporate.

My guess as to why the heat pipes were overwhelmed is that under the more extreme test conditions, the evaporating area of the heat pipe "boiled dry" due to limitations on the amount of water the wick can convey from the condensor to the evaporator. Think of this as a conveyor belt: if the consuming end (evaporator) needs more "product" (liquid water) than the conveyor (wick) can carry, it runs out, and Bad Things Happen.

Hi leem - thanks for the heads-up Is this supply relatively new? I did not come across them in searching the Net? The "connector" blocks look a little simple, as only half the circumference is being used to couple heat - still for EPIA wattages I guess it does not matter.

cmthomson - nice conveyor belt analogy . One helpful practical point I experienced was that in extreme conditions the heatpipe did not "shut down" it still worked at full conductivity, but could not carry the "extra" Watts. So to use my water tank analogy, the water was still pumping out at full flow, however there was more water coming in, than going out!

breunor - cool idea (pun intended ) . A miniature version of what you describe is the Zalman VF900
( http://www.sysopt.com/img/2006/04/vf900cu-3.jpg )
Here all parts of the heatpipe length are being cooled with the fins. One important point that I note, is that Zalman do not bring the ends back into the "hub", but rather solder them to the other heatpipe for rigidity. For your concept you would need a long heatpipe, or of course, just couple into another heatpipe at every junction where the pipe connects to the case.

Hi argo, the epiacenter link is 2 years old, and I talked to the Aavid Thermalloy people then, but only found it on bebensee a few months ago... don't know how long they've been stocking it.

Aavid told me that the clips work fine; they use this system on many of their heatpipe applications. They also told me that the quality of the heatpipes was the most important factor in cooling -- they found significant differences between different manufacturers.

The MCubed system is better because it is adaptible to different heatsink mountings, but it will still be a problem on the newer Via boards, which have non-standard mountings for their big heatsinks that cover CPU and chipsets.

Look forward to seeing how you might apply your stuff to a case mod

please be sure to post your Hiper HMC-2x53x sysyem in the general gallery when (if?) u get it.. its such a great looking htpc case (even the current versions) and i'd love to see it silenced.. that is unless of course u are going to be doing an article on it..