Half of all heatpipes in CPU coolers are handicapped

[Zap]

Heatpipes work best if the cool part is higher than the hot part, right?

Most CPU heatsinks that use heatpipes have roughly half of them pointing the wrong way at any given time in a typical tower case.

From SPCR's review of the Scythe Katana:
"Finally, there is a caution regarding the "Proper Direction for the Installation". Three of the four mounting positions are OK; the fourth, which causes the ends of the heatpipes to be pointing downward, is not approved. This has to do with the need for grvitational pull to flow the condensed liquid in the heatpipes back to the base."

If this is true, then any configuration where the hot end is higher than the cool end means lessened efficiency, right? For instance, a typical "tower" heatpipe unit mounted in a tower case facing the rear fan would have half of the heatpipes (the ones closest to the video card) positioned inefficient. Same thing with a "standard" heatsink with heatpipes, such as the Thermalright XP-90. Let's say it is positioned so that the heatpipes are parallel to the video card. Two of the pipes would have reduced efficiency.

I wonder if anyone has done research into this, perhaps benchmarking a heatsink with the motherboard vertical as well as horizontal. Also, more extreme would be to cut the inefficient heatpipes to see how much performance the heatsink loses - effectively how much those pipes assist even when oriented non-optimally.

Something else I was thinking about... looks as if many of the fins are press-mounted onto the heatpipes. Wouldn't this be like using a heatsink without thermal paste? I wonder if performance would get any better if the fins were soldered to the heatpipes, or even if thermal compound were applied to the pipes before the fins get pressed on? Perhaps an almost liquid-thin drying thermal compound to drip into heatsinks at the heatpipe/fin junction. I recall having good, but not outstanding performance with a Thermalright XP-120 I used to have. Touching it with my finger, the heatpipes would be hot but the fins were just warm.

Comments?

 

[JEDIYoda]

Something else I was thinking about... looks as if many of the fins are press-mounted onto the heatpipes. Wouldn't this be like using a heatsink without thermal paste? I wonder if performance would get any better if the fins were soldered to the heatpipes, or even if thermal compound were applied to the pipes before the fins get pressed on? Perhaps an almost liquid-thin drying thermal compound to drip into heatsinks at the heatpipe/fin junction. I recall having good, but not outstanding performance with a Thermalright XP-120 I used to have. Touching it with my finger, the heatpipes would be hot but the fins were just warm.


If I recall something I was reading...

It matters not how the fins are connected just that they make contact!!

 

[JBDan]

Your last point about the fins, to me, is a very good one. I would think if they were hardmounted to the pipes (not necesarily sodder, but maybe welded Al), the pipes would transfer heat to the fins better/more efficiently. SPCR did a test on the XP120 and different mounting positions (vert/horiz), albeit not very extensive, and they claimed no difference. http://silentpcreview.com/article186-page4.html

 

[JEDIYoda]

Ever since SPCR slammed my beloved Zalman over the(supposed) noise issue...

I have been following SPCR very closely!!

 

[Elcs]

The Thermaltake Volcano 7+ played with Skiving.

Cutting 1 large block of copper into a heatsink. No separate fins to attach and no problems with gaps/thermal problems between base and fins. Perhaps this is a way forward?

Seems like a good idea but I dont think they can get the fins wafer-thin like Thermalright have done.

Heatpipes..... never used a heatpipe design but yes, I can see how the process would work better if they were vertical || rather than horizontal =. Heat rising to the top, getting cooled by air, dropping back to the bottom and repeat.

 

[Tiamat]

now if these awesome heatpipe heatsinks could fit into HTPC cases, you'd get the best of everything. No terrible torque action on the ZIFF socket, and full efficiency of the Heatpipes with no "bad" orientation.

 

[Zap]

Originally posted by: JBDan
SPCR did a test on the XP120 and different mounting positions (vert/horiz), albeit not very extensive, and they claimed no difference. http://silentpcreview.com/article186-page4.html

Thanks for the link. Glad to see someone did think about that and test for it.

Originally posted by: JEDIYoda
It matters not how the fins are connected just that they make contact!!

If so, then why do we use thermal compound between heatsink and CPU?

 

[JEDIYoda]

Originally posted by: Zap

Originally posted by: JBDan
SPCR did a test on the XP120 and different mounting positions (vert/horiz), albeit not very extensive, and they claimed no difference. http://silentpcreview.com/article186-page4.html

Thanks for the link. Glad to see someone did think about that and test for it.

Originally posted by: JEDIYoda
It matters not how the fins are connected just that they make contact!!

If so, then why do we use thermal compound between heatsink and CPU?

lets see those 2 are not even closely related........

Just the mere hesat coming off the CPU rerquires thr thermal compound....

Most of the time by the time the heat has risen to the fins its been spread out and dissipated....

anymore questions?

 

[The Boston Dangler]

Originally posted by: Zap
Heatpipes work best if the cool part is higher than the hot part, right?

Most CPU heatsinks that use heatpipes have roughly half of them pointing the wrong way at any given time in a typical tower case.

From SPCR's review of the Scythe Katana:
"Finally, there is a caution regarding the "Proper Direction for the Installation". Three of the four mounting positions are OK; the fourth, which causes the ends of the heatpipes to be pointing downward, is not approved. This has to do with the need for grvitational pull to flow the condensed liquid in the heatpipes back to the base."

If this is true, then any configuration where the hot end is higher than the cool end means lessened efficiency, right? For instance, a typical "tower" heatpipe unit mounted in a tower case facing the rear fan would have half of the heatpipes (the ones closest to the video card) positioned inefficient. Same thing with a "standard" heatsink with heatpipes, such as the Thermalright XP-90. Let's say it is positioned so that the heatpipes are parallel to the video card. Two of the pipes would have reduced efficiency.

I wonder if anyone has done research into this, perhaps benchmarking a heatsink with the motherboard vertical as well as horizontal. Also, more extreme would be to cut the inefficient heatpipes to see how much performance the heatsink loses - effectively how much those pipes assist even when oriented non-optimally.

Something else I was thinking about... looks as if many of the fins are press-mounted onto the heatpipes. Wouldn't this be like using a heatsink without thermal paste? I wonder if performance would get any better if the fins were soldered to the heatpipes, or even if thermal compound were applied to the pipes before the fins get pressed on? Perhaps an almost liquid-thin drying thermal compound to drip into heatsinks at the heatpipe/fin junction. I recall having good, but not outstanding performance with a Thermalright XP-120 I used to have. Touching it with my finger, the heatpipes would be hot but the fins were just warm.

Comments?

If the end result is effective cooling, then the fins are mounted in an acceptable manor.

Not all heatpipes are the same. It's all about the design of the heatpipe, things like: wall thickness, length, active medium, wick, and bends. Any good heasink with pipes (like the current favorites) will work in the PC enviroment. Installation instructions are available before purchase, BTW.

They are a comprimise, but necessary. As a manufacturer, you need to keep costs under control, while making your products usable to as many as possible. Look at the confusion surrounding the XP-90 and the copper version. Can you imagine if one model was meant only for towers, and the other was for flat boxes, but they look identical?

It says in the Katana's instructions, and was documented by SPCR, the Katana is most effective when mounted in the tower style. This puts the pipes horizontal.

You do realize, mounting humongous heatsinks on an upside-down board is not the greatest idea ever, right?

 

[Jimmah]

Originally posted by: JEDIYoda

It matters not how the fins are connected just that they make contact!!

Heatpipe fins are press fitted yes? If so then there is almost 100% contact around the circumference of the pipe and inside of the fin. Might have some minor gaps but it'd be better than solder, unless of course it is silver solder, but don't expect to see that ever.

 

[Zap]

Originally posted by: The Boston Dangler
Not all heatpipes are the same. It's all about the design of the heatpipe, things like: wall thickness, length, active medium, wick, and bends.

They are a comprimise, but necessary. As a manufacturer, you need to keep costs under control, while making your products usable to as many as possible.

Those are reasonable explanations.

Originally posted by: JEDIYoda

Originally posted by: Zap

Originally posted by: JEDIYoda
It matters not how the fins are connected just that they make contact!!

If so, then why do we use thermal compound between heatsink and CPU?

lets see those 2 are not even closely related........
Just the mere hesat coming off the CPU rerquires thr thermal compound....
Most of the time by the time the heat has risen to the fins its been spread out and dissipated....

Transferring heat from CPU to heatsink base - thermal compound is a good idea. Transferring heat from heatpipe to fins - thermal compound not required? That doesn't make sense.

Originally posted by: Jimmah
Heatpipe fins are press fitted yes? If so then there is almost 100% contact around the circumference of the pipe and inside of the fin.

If someone laps a heatsink and CPU to get 100% contact and the heatsink retension mechanism basically clamps the two together, does that mean heatsink compound is no longer needed?

 

[Banzai042]

On the original topic my persional testing says that the orientation of my XP90 makes no difference in temps. I just ran prime 95, and after a few min i turned the computer on it's side, putting the heatpipes in the supposed "ideal" orientation. After perhaps 10 min i flipped the comp pack up into normal position. When i did this the temps didn't change at all, and i am using an MSI K8N Neo Platinum mobo, which has the socket rotated 90 degrees from normal, and the only way i can install my xp90 is with the heatpipes hanging towards the bottom of the case. In other words i had no temperture change when i moved my xp90 from the theoretically best possible position to the theoretically worst possible position.

 

[BonzaiDuck]

Here's my take on the question at hand. This became a point of contention between a friend and competitor two years ago, when I started using heatpipe coolers, and he continued on the path of massive copper block and fins.

The "gadget" -- heatpipe -- was developed by some company for NASA under contract several decades ago. They were designed to function reliably under complete weightlessness. From this fact, one can deduce:

One set of pipes is augmented, the other may be slightly handicapped by gravity, since the design counts on the "wicking" action of the heatpipe wick, and these sorts of phenomena defy gravity anyway. That is the only way the device would work in complete weightlessness -- through some wicking action.

The company -- I forgot their name but they're still "doin' bidnis" -- posted a white paper on their web-site about the history of heatpipes. The MTBF was estimated at 2 million years.

That MTBF assumes that the pipes are adequately protected from flying debris, the interference of ignorant people, and the curiosity of your house-cat. You're probably covered for the first and last -- the second one depends on you.

Those things are made out of extremely soft and thin metal. You could practically blow on one (or bench-test in Katrina-force winds) and they can bend. Of course, in order to work, they have to be able to withstand SOME air movement.

I would think the interference of gravity on certain heatpipes and their orientation may make a slight difference in the cycling of the refrigerant -- and a slight difference on the "effectiveness" of the disadvantaged heatpipes.

I just don't think it is as significant as the hype and rumor say it is.

 

[GalvanizedYankee]

In responce to Zap's last sentance,re.HS paste.

What if they were both dead flat? Earlyer today i was reading an articel at OCers.com.

The bottom line of a very technical piece, much of it over my head, was thermal paste
will retard heat transfer if both are flat to an angstrom of light. A very light coating of
thermal oil will be best.
In the real world no CPU/HSF will be dead flat and there is the real consideration of the
expansion properties of both. Even a HSF that is all one material type will vary in expansion
because of temperature differential and the alloys use for tubing are prolly different than
the base.


Galvanized

EDIT: forgot an E.

 

[BonzaiDuck]

Per the alloys for the pipes versus the HS base: You've got that right . . . . completely different . . . .

 

[Zap]

Originally posted by: Banzai042
On the original topic my persional testing says that the orientation of my XP90 makes no difference in temps.

Thank you. That mostly answers my first question (along with info from The Boston Dangler).

Originally posted by: BonzaiDuck
massive copper block and fins.

A Thermalright junkie? I think I still have an SLK900 sitting around, NIB even!

Originally posted by: BonzaiDuck
heatpipes and their orientation...
I just don't think it is as significant as the hype and rumor say it is.

That's what it's beginning to sound like. My curiosity is satisfied.

Originally posted by: GalvanizedYankee
thermal paste will retard heat transfer if both are flat to an angstrom of light... In the real world no CPU/HSF will be dead flat and there is the real consideration of the expansion properties of both.

... which is why we use heat transfer compound between CPU and heatsink. Why does this not matter for the fins press-fitted onto heatpipes?

 

[GalvanizedYankee]

re. press fit of fins onto tubes.
*IF* the fit is true,ie.if the swagging process is vg the fit will be incredably close and the fin edge where it contacts the tube will have some degree of spring loading. The internals of a PC are not subject to the hostile enviroment a car's condenser is, so there is no moisture to accelerate corrosion.
I did go to loctite.com and looked for thermal glues, on the order of a watery super-glue.
None were to be found. Loctite supplies the world with product and if there was a market
it would allmost surely be there.
In the thread on the nickel plated XP-90, i posted that in my reading, i was informed that they are plated to permit soldering of dissimilare metals, then most likly replated for appearance.
On non-soldered fins the metals having very similar properties will have a natural attraction for eachother. There will be little or no corrosive action as there would be with dissimilare metals.(that's my take on it).

re. heat pipe orientation.
www.heatsink-guide.com/...Lots of good info there!
A crud heatpipe would have to be vertical and could use distilled water under vacuum for
cooling@ 50C.
The more sophisticated use liquids with high surface tension and internal wicks that draw the condensed fluid back to the heat source. Variations in design for application or marketing is what it's all about. A good heatpipe HS dose not have to be case filling.
Look at the pic at the bottom of the heatpipe page at the link. I like that! The fan under the finned area, blowing away from the mainboard. Benifit there, betca.

Galvanized