Need a loop for 4 x 290 xfire

Soldato
Joined
19 Feb 2011
Posts
5,849
Ok so I have 1 x 290 but plan on adding another 3 as I think a superflower 1200w plat can handle them. I need to know what rad amounts I will require, how much tubing, fittings etc, basically everything, I guess might aswell cool the 4770k also.

Case is a Phanteks Enthoo primo
 
I don't know about 4, but i have 3 of these and 2x280mm rads can barely keep them at 55-65 degrees while mining. That is with the CPU and mobo in the loop but they are sitting idle most of the time. You can use this as a reference.
 
I think general advice is two 120mm fans-worth per GPU and one for the CPU and mobo combined.

You might find it well worth to look at the external Phobya 1080 (9x120mm) super-rads.
 
240mm of rad per high end components, afaik.
So 2 480 rads or as Daniel mentioned, have an external Phobya 1080/1260. Phobya makes all the accesories to their rads, even a special cages which you can attach to case side panel or even to the wall.
I am having 3 thick 360s for 4770k and 3 780s.
 
290 do run hot (i found this out when changing to it) so 240 is what i would say for each of them unless you want high speed fans all the time.
 
GOT Rads

High Res Pic : http://s801.photobucket.com/user/Karlitosss/media/P1010757_zps130598ba.jpg.html?sort=3&o=42

P1010782_zps51cbe8b2.jpg

P1010757_zps130598ba.jpg
 
Would be aiming for 360mm+ per two 290x's. In the end the cheapest best option is to use an external 1080/1260 rad with quick disconnects for ease of use.

For blocks, EK blocks or heatkiller get my vote for the 290x. (have been eyeing up the heatkiller black edition and i may have to do some comparisons soon, its just sooooo tasty!)

If you go for it, be careful with the thermal pads for the VRMs as they run hot and need every bit of help to disperse the heat

My VRM temps have always been fine, what block are you using?

Did you put a tiny bit of paste under the thermal pad to unsure good contact with components closer to the PCB?
 
I just stuck three 290's on a MO-RA 1080 (4x180mm 700rpm) with a EK blocks and a parallel bridge and they all hit just under 50°C (ambient around 23°C) when mining at stock clocks/volts 1040/1375 fwiw.
 
Last edited:
Yeah, easier to manage cables too :D
I would like to feel a bit more air passing through the rad though tbh, but Silverstone AP181's are a bit pricey compared to these Phobya jobbies.

ED: As for the OP, it would be a shame not to use the available space in that case.
I'd say to grab a couple of reasonably thick, high fpi 480's like GTX extremes (if the width allows for one in the bottom!) and run them in push/pull.
A lot more expensive than a single large external rad though...
 
Last edited:
My VRM temps have always been fine, what block are you using?

Did you put a tiny bit of paste under the thermal pad to unsure good contact with components closer to the PCB?

That's what i was talking about. I didn't put paste and must have slipped the pads on one of the cards during the mounting of the block as the VRMs are at 65 on 2 of them but around 80 on the third. Which is still much better than air where they go to like 100. Couldn't be bothered resetting the block this time, i will do it next time i'm re-working the loop in a month or two.
 
Yeah, no point in breaking it down if it happily does 80 on vrms all day. A small splodge of mx-2/4 on the mosfets do the trick, as it is non conductive and has a certain thickness which helps keep the thermal pad down.
 
I have fujipoly Ultra Extreme on the VRM (without Thermal Paste, why thermal paste with pads anyway?) and I never see the VRM get above 40'C even with 1225/1600. Water temp around 30'C during load with Quad 290x.

Why put paste on the pads?? Only EK "suggests" to add paste in case it falls off. But adding more paste will add more materials between the Mosfets (or VRMS) and the blocks will just decrease the heat transfer increasing the temps of the component to cool. The pads won't fall off if you put the block on the card before flipping it around to screw it.

I use thermal paste only on chips where the gap is under 0.5mm. (aka CPU or GPU)


P1010714_zps7d07afbb.jpg

P1010715_zps194923b4.jpg
 
Last edited:
Why put paste on the pads?? Only EK "suggests" to add paste in case it falls off. But adding more paste will add more materials between the Mosfets (or VRMS) and the blocks will just decrease the heat transfer increasing the temps of the component to cool. The pads won't fall off if you put the block on the card before flipping it around to screw it.

Paste on VRMs don't increase temps when properly used. While you see the increase of temps with chips when you add too much, MOSFETs don't have even surfaces and you get better heat transfer with a tiny bit of TIM, just enough to let the pad stick to it.

While GPUs are pretty straight forward to mount, many motherboard blocks which cool both VRMs and chips are not. TIM goes a long way when holding pads into place as the chip screws are always screwed down first, with the rest to follow evenly. This is different to GPU's in the way that on most motherboard full cover blocks, the chip is not centered, so once you screw the chip screws down, the VRMs do not always end up with great coverage and slipping thermal pads are common (the CHVF-Z to name one of the most annoying vrm+chip blocks).

To be honest there isn't any realistic downside to putting TIM on MOSFETS. If you find the VRM temps have gone up, then you probably have used about half a tube of TIM, since the VRMs dont produce enough heat to make even a bit too much TIM an issue or even measurable using temp monitoring software. While the upside of using it is keeping the pad in place, which doesn't sound like much.... till you feel the need to break down your loop and make sure the thermal pad hasn't slipped during the mount.
 
Paste on VRMs don't increase temps when properly used. While you see the increase of temps with chips when you add too much, MOSFETs don't have even surfaces and you get better heat transfer with a tiny bit of TIM, just enough to let the pad stick to it.

While GPUs are pretty straight forward to mount, many motherboard blocks which cool both VRMs and chips are not. TIM goes a long way when holding pads into place as the chip screws are always screwed down first, with the rest to follow evenly. This is different to GPU's in the way that on most motherboard full cover blocks, the chip is not centered, so once you screw the chip screws down, the VRMs do not always end up with great coverage and slipping thermal pads are common (the CHVF-Z to name one of the most annoying vrm+chip blocks).

To be honest there isn't any realistic downside to putting TIM on MOSFETS. If you find the VRM temps have gone up, then you probably have used about half a tube of TIM, since the VRMs dont produce enough heat to make even a bit too much TIM an issue or even measurable using temp monitoring software. While the upside of using it is keeping the pad in place, which doesn't sound like much.... till you feel the need to break down your loop and make sure the thermal pad hasn't slipped during the mount.

This is misinformation. Why use thermal paste when you already have pads??? I don't understand your logic.

Cost me 130$ of thermal pads for my 4 x 290x but lets put 5$ thermal paste underneath them??? This is plain stupid to me.

Thermal pads and thermal compound are used to fill air gaps caused by imperfectly flat or smooth surfaces which should be in thermal contact;[1] they would not be needed between perfectly flat and smooth surfaces. Thermal pads are relatively firm at room temperature, but become soft and well able to fill gaps at higher temperatures.

Do you need more paste to fill the air gaps?? Just get the correct thickness of pads :confused: .

Adding material will just decrease the heat transfer... I think its logical, no?

----- and when mounting a block, I don't see why the pads would move.
 
Last edited:
The height of the MOSFET transistor from the PCB is greater than the thickness of thermal pads recommended. Heat is not only dissipated from the MOSFETs themselves but the PCB very near them, so you gain a fair bit of surface area by adding a tiny bit of TIM in between the transistors.

You are idealizing the uses of thermal pads too much, thinking the correct thickness will give you near perfect coverage.

I am going to stop after this post since you seem to just be repeating yourself but your logic is based on idealized assumptions:

1. correct thickness of thermal pads gives you as much area contact of the components and surrounding PCB as the pad and paste.

It does not, you can see this when you lay the pad down on the MOSFETs and check to see if the whole pad is in contact with transistors AND pcb or not

2. You assume you use a significant amount of thermal paste when applying it

Literately just touch the end of the syringe without pushing any out. whatever comes off the end is enough, $5 of tim is far far too much even for 4 cards. Also find the money argument quite funny after admitting to $130 of thermal pads, hundreds more on wcing equipment and thousands for actual components. I am sure if you have all this stuff, a dab of TIM could be spared for some thermal pad security.

3.You assume that thermal pads dont move if you mount the block 'your way'

Someone in this thread has admitted that they think this is the reason for their temps. EK guides also recommend it. Blocks like the VRM + nb block in my sig (left hand side in black acetal) is an example of particularly difficult to mount blocks, where the chip is at one end and the MOSFETs are at the other. In cases like these, the chip screws ALWAYS get tightened first, which causes the block to lift at the VRM side where the pad is. This makes it liable to slipping until the rest of the block is tightened.

No need to quote what TIM and pads are for, have been using them before i had ever used them for PC cooling.
 
Last edited:
Whatever dude. Don't need to get so offensive. Do you put thermal pads underneath the thermal grease on your CPU ... In case the grease fall off. (facepalm) Because it seems so hard to mount.

You know, you take Mobo mosfet block as example, I understand some blocks are hard to mount (even though I never put grease under the pads) BUT MY POINT IS: grease WILL NOT increase the thermal transfer. IT IS MUCH BETTER to not use it for lower temperatures.
 
Last edited:
Thanks man. But it's a Watercool Mo-RA3 Pro LT Black. With 9 x Gentle Typhoons Ap15s, a fan controller and a temp sensor at the output of the Rad.

P1010800_zpsa59a4b11.jpg


P1010761_zps0e753725.jpg


P1010666_zps505d0851.jpg


P1010717_zpsee3cd6fd.jpg


P1010719_zpsf0ed4429.jpg


P1010718_zps8291232e.jpg



P1010738_zps2c6bdc26.jpg
 
Last edited:
Back
Top Bottom