Sign in with
Sign up | Sign in
Your question

Water or Air-cooling solution for HD6950 in 90° Rotation (FT02)

Last response: in Overclocking
Share
January 3, 2012 10:37:07 AM

Hey guys!

Hope you can help me because I am getting a bit desperate.

I have got a Silverstone Fortress 2 (FT02) case where motherboard is in vertical position, so all the hot air is coming out from the top and not the back.
I also have a MSI R6950 Twin Frozr III Power Edition/OC graphics card, which gets quite loud when playing games. I wanted to buy a aftermarket GPU cooler for it,
but after researching information on the internet I found very conflicting opinions about effectiveness of aftermarket air coolers. The problem is - as the motherboard is in a 90° rotation, the heat pipes in the cooler have to work against gravity unless the heat source is located below other parts of the heat pipe.

After I couldn't find really effective solution (because for every person with a working aftermarket cooler in vertical motherboard, there was another person saying that for him the same cooler is not working) I figured out - maybe I should try water block cooling.

Unfortunately EK does not make a full block water cooler for my non-reference GPU, so at the moment I don't know what to do. I know that there are options of making a custom cooling system for GPU, but I have no experience whatsoever with water cooling and I have limited budget of about 80 pounds for my cooling solution. Maybe there are other water cooling manufacturers who make full water blocks for this card, but as I mentioned - I have no experience with liquid cooling so its difficult to know where and what to look for.

Will appreciate any advice.

Thanks.
a c 324 K Overclocking
January 3, 2012 12:42:16 PM

Well your budget eliminates watercooling altogether. I would suggest sticking to an aftermarket air cooler if this is the case.

I don't agree with your statements about heatpipes working against gravity...how do you think all heatpipes work? Also, video cards are designed for placement that differs from normal heatpipes, yet they still make heatpipe coolers for them. Going further, CPU coolers that are heatpipe design are actually sideways when mounted inside a case...so there isn't any real orientation to a specific up or down when using these kinds of cooler designs.
January 3, 2012 12:55:13 PM

There is enough information from users, testers and Silverstone itself to prove that you should be worried about this issue.

This is the official info from Silverstone

The problem is with the heatpipes going "upside-down U", or "groove pipes" as Silverstone calls them: http://www.silverstonetek.com/qa/qa_...=FT02&area=usa

Quick snippet:
There are two main types of heat pipes used in popular aftermarket coolers, they are groove and powder. Groove heat pipes are very susceptible to gravity while powder heat pipes are less so. To achieve best performance in either heat pipe technology, they need to be placed horizontally or have the heat source side located below the other end of the heat pipe. We recommend choosing and installing components with heat pipes carefully by taking into consideration of the following examples:

The illustration here shows a VGA cooler that will not work well in the FT02 because the heat source side (touching the GPU) ends up being located higher than the other end.


Basically, heatpipes in GPU coolers aren't very well made and are designed to perform horizontally (parallel to the floor). If vertical, the heatpipes don't distribute heat properly because they're working against gravity. You can read more on the technical reasons here: http://www.thermacore.com/frequently...efault.aspx#73

There are plenty of users who have confirmed this first hand in FT02 and RV02 cases.
January 3, 2012 1:11:10 PM

rubix_1011 said:
Well your budget eliminates watercooling altogether. I would suggest sticking to an aftermarket air cooler if this is the case.

I don't agree with your statements about heatpipes working against gravity...how do you think all heatpipes work? Also, video cards are designed for placement that differs from normal heatpipes, yet they still make heatpipe coolers for them. Going further, CPU coolers that are heatpipe design are actually sideways when mounted inside a case...so there isn't any real orientation to a specific up or down when using these kinds of cooler designs.



Also this is from another forum user:

----------------------------------------------------------------
I wanted to report my experience of using Thermalright's Shaman VGA cooler in a 90° rotated environment like Silverstone's Fortress FT02 case.

Put simply, it does not work well.

I've mounted it on a HD 6970 reference design rev2 (blue PCB). The card is overclocked to 950 / 1450, stock voltage. Stress tests are run with Afterburner's embedded Kombustor utility.

First, I ran a test in the standard position, by putting the case on its front panel, so the card would be horizontal and Shaman's fan would blow upwards.

Idle : 33 °C
Kombustor 5 min : 57 °C
Kombustor 10 min : 57 °C
Kombustor 20 min : 57 °C

Shaman is a GREAT cooler in its meant positioning. This result also tells us the cooler is properly mounted.

I then put the FT02 back in its normal position, so the video card now hangs vertically, with Shaman's fan blowing to the card horizontally.

Idle : 33 °C
Kombustor 5 min : 67 °C
Kombustor 10 min : 85 °C
Kombustor 15 min 30 sec : 97 °C - test aborted.

There you. Some first hand test results.

I had a hell of a time finding such information at the time of shopping for my build's parts, so I thought I would post them back.

Now I need to either buy a VGA cooler that will work well in the FT02 setting, or to buy a new case.
---------------------------------------------------------------------------------


Also, even there is not too much info on the internet about this problem, quite a few people report having issues with this. Also, one tech website (if you want the link I can try and find it) tested some graphic cards in normal motherboards and in 90 degree ones and reported that most of the air coolers had a difference of 12-20 degrees, depending of rotation (the worst results of course having in 90 degree boards).

So I would say that this thing is more serious than you were thinking. And the biggest problem is inconsistency of reports, as I mentioned in the initial post. If I would see 5 people reporting good results in 90 degree motherboard with the same cooler, I would definitely go for it. Unfortunately I can't find such reports. Pretty much for every report or two of good results there is one report about unacceptable results with the same cooler. Some mentioned it could be quality control issue within the manufacturing plant or something.
a c 324 K Overclocking
January 3, 2012 2:10:47 PM

Some of those seem to be poor cooler design/quality rather than heatpipe orientation. This would be an interesting test...
a b V Motherboard
a b K Overclocking
January 3, 2012 2:12:55 PM

The problem with most aftermarket coolers is the airflow. Airflow is everything in the RV/FT cases.


A majority of aftermarket coolers pull air from the expansion slot and exhaust into the case. The problem with this is that the fans are pulling air opposite the direction of flow in the Raven/Fortress cases (vertically downward while air is moving vertically upward).

This causes a stagnation point in the airflow in the GPU, which prevents air from cooling the heatsink attached to the GPU, eventually causing overheating and crashing.


Most reference coolers work the opposite way, in that they pull air from the case and exhaust out of the expansion slot. In this case, the air is moving with the direction of airflow in the case (vertically upward while air is moving vertically upward).

This is an extremely efficient setup for obvious reasons, allowing you to cool much better with less effort. Plus, heat naturally rises, which doesn't add that much to your cooling ability, but it probably does help somewhat.


My initial suggestion would be to see if you could take apart the frame on the Twin Frozr cooler and flip the fans over to switch the direction of airflow. That way, your cooler is operating similarly to the reference GPU setup I discussed. You just have to remove a few screws to move the fans - there's a few holding them down.

Other than buying a good aftermarket cooler and trying to swap the fan direction or buying a new reference GPU, I don't know what else I can suggest.
a b V Motherboard
a b K Overclocking
January 3, 2012 2:31:58 PM

must be the msi heatsink design

in my raven rv02 my asus gtx570 direct cu overclocked to 880mhz core and 4400mhz memory runs much cooler

and quieter than it did in my haf x

so with this card the top to bottom airflow works great

idles at 27c with fans at 12% and when gaming at 1920 x 1200 the fans only go up to 23% can barely hear it

over the other stuff in the raven case where as in the haf x it was pretty loud

a c 117 V Motherboard
a b K Overclocking
January 3, 2012 2:44:17 PM

boiler1990 said:
The problem with most aftermarket coolers is the airflow. Airflow is everything in the RV/FT cases.


A majority of aftermarket coolers pull air from the expansion slot and exhaust into the case. The problem with this is that the fans are pulling air opposite the direction of flow in the Raven/Fortress cases (vertically downward while air is moving vertically upward).

This causes a stagnation point in the airflow in the GPU, which prevents air from cooling the heatsink attached to the GPU, eventually causing overheating and crashing.


Most reference coolers work the opposite way, in that they pull air from the case and exhaust out of the expansion slot. In this case, the air is moving with the direction of airflow in the case (vertically upward while air is moving vertically upward).

This is an extremely efficient setup for obvious reasons, allowing you to cool much better with less effort. Plus, heat naturally rises, which doesn't add that much to your cooling ability, but it probably does help somewhat.


My initial suggestion would be to see if you could take apart the frame on the Twin Frozr cooler and flip the fans over to switch the direction of airflow. That way, your cooler is operating similarly to the reference GPU setup I discussed. You just have to remove a few screws to move the fans - there's a few holding them down.

Other than buying a good aftermarket cooler and trying to swap the fan direction or buying a new reference GPU, I don't know what else I can suggest.


I am just going to add to this, I have a FT03(The little fortress case) with a stock 5870(blower at the far end). It works just as well in the 90 degree orientation as it did in the normal case(Was in a Fractal Define R2 before).

On thing it may be interested in, is would this issue have some effect on a desktop orientation case? I know many heatsinks had listed ways to install so that the heatpipes are parallel to the floor as well.
January 3, 2012 2:48:40 PM

boiler1990 said:
The problem with most aftermarket coolers is the airflow. Airflow is everything in the RV/FT cases.


A majority of aftermarket coolers pull air from the expansion slot and exhaust into the case. The problem with this is that the fans are pulling air opposite the direction of flow in the Raven/Fortress cases (vertically downward while air is moving vertically upward).

This causes a stagnation point in the airflow in the GPU, which prevents air from cooling the heatsink attached to the GPU, eventually causing overheating and crashing.


Most reference coolers work the opposite way, in that they pull air from the case and exhaust out of the expansion slot. In this case, the air is moving with the direction of airflow in the case (vertically upward while air is moving vertically upward).

This is an extremely efficient setup for obvious reasons, allowing you to cool much better with less effort. Plus, heat naturally rises, which doesn't add that much to your cooling ability, but it probably does help somewhat.


My initial suggestion would be to see if you could take apart the frame on the Twin Frozr cooler and flip the fans over to switch the direction of airflow. That way, your cooler is operating similarly to the reference GPU setup I discussed. You just have to remove a few screws to move the fans - there's a few holding them down.

Other than buying a good aftermarket cooler and trying to swap the fan direction or buying a new reference GPU, I don't know what else I can suggest.



Well, the problem is - Twin Frozr III alrady works that way. Both fans are pulling air in from the inside of the case and exhausting it through the top of the case (expansion slot). As I didn't trust my hand completely I used piece of dental floss - it clearly was getting sucked in through the fans and if I put it on top of the expansion slot if clearly tried to move away along with the hot air I was feeling.

About good aftermarket cooler - I don't mind buying any of the best and most expensive air coolers, but it's pretty much a game of luck. People are reporting such a different results, that I am quite sure that quality of the same product model can vary greatly. Don't want to waste 60 pounds on something that doesn't work.

Of course thanks for your help anyway. :) 
a c 324 K Overclocking
January 3, 2012 3:17:09 PM

Case airflow makes much more difference than about anything else. You can also try removing the stock cooler, replacing thermal paste and re-mounting the stock cooler. This often gives decent performance benefits, 5C or so at load in many instances.
a b V Motherboard
a b K Overclocking
January 3, 2012 5:47:18 PM

Most of the Twin Frozrs dont work that way AFAIK. I wonder if the case airflow is strong enough to essentially reverse the flow in the GPU cooler.

You could always verify by watching which way the fans spin and looking at the blade shape.
a b V Motherboard
a c 100 K Overclocking
January 3, 2012 8:28:09 PM

I didn't think heat pipe orientation had anything to do with anything. After all most benchmark rigs have the mobo horizontal, so the heat pipes are vertical on CPU coolers, yet they perform very well.
a c 324 K Overclocking
January 3, 2012 8:46:10 PM

Quote:
I didn't think heat pipe orientation had anything to do with anything. After all most benchmark rigs have the mobo horizontal, so the heat pipes are vertical on CPU coolers, yet they perform very well.


This was exactly my understanding, as well. You can usually mount most CPU coolers as you want, which would put the heatpipes in either vertical or horizontal orientation which has been a topic often discussed if a cooler can exhaust out the back or out the top...either works just fine. There are also many NB heatpipe solutions that have pipes oriented in various positions as well.

If heatpipes were that dependent on their orientation in reference to gravitational pull, why aren't CPU heatsinks built with a 90 degree angle off the base so the CPU cooler 'tower' of heatpipes is always oriented vertically?
January 3, 2012 9:04:17 PM

rubix_1011 said:
Quote:
I didn't think heat pipe orientation had anything to do with anything. After all most benchmark rigs have the mobo horizontal, so the heat pipes are vertical on CPU coolers, yet they perform very well.


This was exactly my understanding, as well. You can usually mount most CPU coolers as you want, which would put the heatpipes in either vertical or horizontal orientation which has been a topic often discussed if a cooler can exhaust out the back or out the top...either works just fine. There are also many NB heatpipe solutions that have pipes oriented in various positions as well.

If heatpipes were that dependent on their orientation in reference to gravitational pull, why aren't CPU heatsinks built with a 90 degree angle off the base so the CPU cooler 'tower' of heatpipes is always oriented vertically?



Could it be because of the differences in architecture in CPU and GPU coolers. Form factor, length of heat pipes and other stuff. Maybe I am just talking bull**** now but there might be some other factors in play which we don't know about, where GPU heat pipe orientation gets more important than in CPU's
a b V Motherboard
a c 100 K Overclocking
January 3, 2012 9:09:14 PM

I thought heat pipes just had a liquid inside them to quickly conduct heat away from the heat source... this liquid would therefore be full inside the pipe, so it won't "accumulate" in any spot. Heat conduction couldn't care less about gravity, so ultimately it shouldn't matter.
a c 324 K Overclocking
January 3, 2012 9:10:29 PM

I wouldn't think this would be the case, but it's a variable that we don't seem to have an answer for. I just find it odd that a CPU cooler can be effective, regardless of it's mounting and orientation, but GPU coolers utilizing the same concepts fail to work if they are not mounted typical to an ATX layout.

If these theories were true, than all of those huge heatpipe tower CPU coolers should work much better when mounted vertically and the case laying down (like an old real 'desktop' case used to be) and the motherboard laying down flat.

I'm not trying to discredit anything you are saying- there has to be something to account for those performance differences. I'd be curious if those guys tried tipping their cases to resemble a normal case layout and see if this corrected the heatpipe issue. It would at least give you an answer as to their orientation and each result.

Quote:
I thought heat pipes just had a liquid inside them to quickly conduct heat away from the heat source... this liquid would therefore be full inside the pipe, so it won't "accumulate" in any spot. Heat conduction couldn't care less about gravity, so ultimately it shouldn't matter.


This was my thinking as well, but I thought it was more of a 2-part solution...kind of like a lava lamp.
a b V Motherboard
a b K Overclocking
January 3, 2012 9:10:56 PM

Quote:
Most of the Twin Frozrs dont work that way AFAIK. I wonder if the case airflow is strong enough to essentially reverse the flow in the GPU cooler.

You could always verify by watching which way the fans spin and looking at the blade shape.


Did some more research and turns out I was wrong.

I agree with the others that the heatpipes shouldn't affect temperatures significantly/at all. The only reason I could see the orientation affecting heat dissipation is if the pipes entirely blocked the air flow (which isn't possible unless they're flat planes).

As for the whole "gravity effect" thing, what you're discussing could also be due to using poor heat conductors as heatsink materials. Gravity is physically unrelated to masses/gravitational forces of objects, so it doesn't make much sense to blame it.


What is your Twin Frozr running at? (Fan speed/% and temps)
a b V Motherboard
a b K Overclocking
January 3, 2012 9:13:24 PM

rubix_1011 said:
I wouldn't think this would be the case, but it's a variable that we don't seem to have an answer for. I just find it odd that a CPU cooler can be effective, regardless of it's mounting and orientation, but GPU coolers utilizing the same concepts fail to work if they are not mounted typical to an ATX layout.

If these theories were true, than all of those huge heatpipe tower CPU coolers should work much better when mounted vertically and the case laying down (like an old real 'desktop' case used to be) and the motherboard laying down flat.

I'm not trying to discredit anything you are saying- there has to be something to account for those performance differences. I'd be curious if those guys tried tipping their cases to resemble a normal case layout and see if this corrected the heatpipe issue. It would at least give you an answer as to their orientation and each result.

Quote:
I thought heat pipes just had a liquid inside them to quickly conduct heat away from the heat source... this liquid would therefore be full inside the pipe, so it won't "accumulate" in any spot. Heat conduction couldn't care less about gravity, so ultimately it shouldn't matter.


This was my thinking as well, but I thought it was more of a 2-part solution...kind of like a lava lamp.


It could also just be poor cooler design - GPUs produce significantly more heat than CPUs do, and OC'ed CPUs with large voltages run about as hot as GPUs (with adequate cooling of both)
a c 324 K Overclocking
January 3, 2012 9:14:34 PM

Quote:
It could also just be poor cooler design - GPUs produce significantly more heat than CPUs do, and OC'ed CPUs with large voltages run about as hot as GPUs (with adequate cooling of both)


Agreed. In many instances, a GPU is around a 2:1 heat wattage producer over a CPU. This is why a laugh at those LCS coolers for a GPU that utilize a 120mm rad...like their CPU counterparts.
January 3, 2012 11:06:49 PM

boiler1990 said:
Quote:
Most of the Twin Frozrs dont work that way AFAIK. I wonder if the case airflow is strong enough to essentially reverse the flow in the GPU cooler.

You could always verify by watching which way the fans spin and looking at the blade shape.


Did some more research and turns out I was wrong.

I agree with the others that the heatpipes shouldn't affect temperatures significantly/at all. The only reason I could see the orientation affecting heat dissipation is if the pipes entirely blocked the air flow (which isn't possible unless they're flat planes).

As for the whole "gravity effect" thing, what you're discussing could also be due to using poor heat conductors as heatsink materials. Gravity is physically unrelated to masses/gravitational forces of objects, so it doesn't make much sense to blame it.


What is your Twin Frozr running at? (Fan speed/% and temps)


When I am gaming fan speed gets to about 60% (which is quite loud, while 40% is something you can barely hear). Not sure about temps - think they don't get anywhere major. I don't use MSI Afterburner, tried in the beginning after buying the card, but then realised it would switch to higher fan speeds sooner or later anyways. I don't know - maybe if I would play around a bit in Afterburner I could setup something with optimal noise and temps, but not really sure. Also as the heater in the room is not too far from PC (about 1m) sometimes it plays some role in how much the GPU fans are spinning.

I will definitely try replacing thermal paste (as I have one tube from Arctic Cooling 3 I think) and see how it goes from there.

If the problem with noise is still big enough I might risk and order one of the aftermarket coolers.
January 3, 2012 11:29:09 PM

rubix_1011 said:
I wouldn't think this would be the case, but it's a variable that we don't seem to have an answer for. I just find it odd that a CPU cooler can be effective, regardless of it's mounting and orientation, but GPU coolers utilizing the same concepts fail to work if they are not mounted typical to an ATX layout.

If these theories were true, than all of those huge heatpipe tower CPU coolers should work much better when mounted vertically and the case laying down (like an old real 'desktop' case used to be) and the motherboard laying down flat.

I'm not trying to discredit anything you are saying- there has to be something to account for those performance differences. I'd be curious if those guys tried tipping their cases to resemble a normal case layout and see if this corrected the heatpipe issue. It would at least give you an answer as to their orientation and each result.

Quote:
I thought heat pipes just had a liquid inside them to quickly conduct heat away from the heat source... this liquid would therefore be full inside the pipe, so it won't "accumulate" in any spot. Heat conduction couldn't care less about gravity, so ultimately it shouldn't matter.


This was my thinking as well, but I thought it was more of a 2-part solution...kind of like a lava lamp.


Well, in my second post you can see at least one case where the guy tried to do this.
---------------------------------------------------------------------------------------------------
First, I ran a test in the standard position, by putting the case on its front panel, so the card would be horizontal and Shaman's fan would blow upwards.

Idle : 33 °C
Kombustor 5 min : 57 °C
Kombustor 10 min : 57 °C
Kombustor 20 min : 57 °C

Shaman is a GREAT cooler in its meant positioning. This result also tells us the cooler is properly mounted.

I then put the FT02 back in its normal position, so the video card now hangs vertically, with Shaman's fan blowing to the card horizontally.

Idle : 33 °C
Kombustor 5 min : 67 °C
Kombustor 10 min : 85 °C
Kombustor 15 min 30 sec : 97 °C - test aborted.
---------------------------------------------------------------------------

But yeah, I agree - if everyone would do this to test their coolers we could see if it's really the problem with orientation, or maybe in some cases it is just bad installation of aftermarket cooler or just a bad quality cooler altogether.
a c 324 K Overclocking
January 4, 2012 12:03:03 AM

Still seems iffy to me...mostly cooler quality and/or too many variables while testing. You could also rule out airflow by blowing a house or desk fan on high into the case in both orientations...the high CFM of air should offset any airflow issues caused by the natural airflow in the case in either direction during testing.
a c 207 V Motherboard
a c 145 K Overclocking
January 4, 2012 12:21:38 AM

http://benchmarkreviews.com/index.php?option=com_conten...

Quote:
In every case where the CPU cooler used a "U" shaped heat pipe rod in their design, having the unit positioned with the rods running horizontally proved to offer a cooling benefit. ....This goes to show you that heat-pipes are in fact prone to suffer the effects of gravity in their design. Once the vapor cools and becomes liquid, it seems to have an easier time completing the thermal circuit from side to side than it does from top to bottom.


January 4, 2012 9:48:10 PM

First - wanted to tell you guys - you have been great. I appreciate you trying to help me and offering solutions. As I understood - it is still pretty much a game of luck and there are way too many variables.

Somebody recommended me the Accelero Twin TurboII, which I will buy and try my luck with it. Will report my experience and test results here.
a c 324 K Overclocking
January 4, 2012 10:59:23 PM

Yeah, seems like there is some hit or miss solutions when it comes to heatpipe coolers...which isn't what I expected. Well, to be honest, you always expect some solutions to be far better than others, but not with such difference with basic design principles and even the end-use of a cooler reacting differently depending on how it is mounted. I'll admit- there are a few misconceptions that I assumed were just sure-shot concepts that I didn't think existed with such controversy.
February 13, 2012 6:00:17 PM

Ok, it's been a long time, but my feedback about Arctic Cooling's Accelero TWIN TURBO II finally is here. I have to thank forum member who goes by the name "may1", for recommendation.

This is sligthly edited copy of the post I made for him.

---------------------------------------------------------------------------------------------------
Following the recommendation I ordered Arctic Cooling Accelero TWIN TURBO II. There was a freakishly long delay, reason being lack of stock, but after almost 6 weeks of waiting I got it delivered.

When I removed the stock cooler (which was MSI Twin Frozr III) I discovered that there are no heat-sinks whatsoever for memory or any other components. The whole cooling process was happening through the main cooling plate (not sure how it is called, but you know what I mean). So following recommendation once again, I did not bother with installing supplied heat-sinks, but simply replaced the cooler. After that I realised I made a mistake, because I mixed up and used the wrong spacers which were too tall and resulted in lack of contact between cards plate and coolers plate. Luckily I noticed that quickly and changed the spacers. I used the pre-applied thermal paste, because that was easier and I always could tinker around and change it for my Arctic Silver 5 if needed.


So here are the results. I used the 12V power connector for the fans.


I run a MSI Combustor stress test (98% GPU load). After about 4 minutes the temperature peaked at 48c degrees. Run it for full 10 minutes, temp stayed the same.


Today I decided to test it a bit more and selected 'extreme burn in' option in MSI Combustor. In 5-6 minutes temperature reached 63c degrees. Run the test for full 25 minutes. Temperature staid the same (63c).


In both cases it was impossible for me to tell by the noise if and how much the fans are spinning, because the Twin Turbo II is amazingly quiet. For all I know, at idle and under extreme burn in test the noise was exactly the same - i.e. I couldn't hear the Twin Turbo II over the Silverstone Fortress 2 case fans on low setting (and I have replaced the top case fan with GentleTyphoon D1225C).


So for me this is clearly a success and even better result than I expected.

P.S. I also assume that some of the bad results people have experienced could be because of errors in installation process, as I almost made such a mistake with the spacers. Anyhow, I am happy to report that for me this cooler works with excellent results.
May 15, 2012 12:48:15 AM

thanks alot im sitting here with my 680 trying to figure out if the twin turbo II would work well in my ft03b, just got my answer, thanks!!!!!
!