Notices

Overclockers Forums > Hardware > Cooling > Water Cooling
Water Cooling Discussion devoted to blocks, pumps, radiators, reservoirs, tubing, and everything else to get you running smooth on a water loop
Forum Jump

Radeon 7990 Custom super cooler

Post Reply New Thread Subscribe Search this Thread
 
 
Thread Tools
Old 02-24-13, 10:42 PM Thread Starter   #1
ZytheEKS
Registered



Join Date: Dec 2012

 
Radeon 7990 Custom super cooler


I'm posting this thread because I want suggestions/recommendations on my design.

So I'm going to be getting a 7990, and so far all the waterblocks I've seen are complete crap. I have found some that work really well, but I HATE how they look and seeing as how I have a mostly clear DIY chassis I don't want it looking like scat so I'm going to be making a custom waterblock and I figure if I'm going to be putting in the effort to make a custom full coverage block why not go overkill with a super cooler.


So here's the idea I'll get something like this with peltiers mounted on both sides
http://www.shop.customthermoelectric...7.qscstrfrnt06

Have it configured so that one peltier sits with the cold side on the GPU and the other peltier sits with the cold side pulling heat from the fluid in the waterblock

Then mount that on both GPU's, then use a vacuum bag and seal it flush with the card. Then cast the mold in a ceramic, then use basic casting stuffs to reverse the mold. Then run copper tubing along all the rest of the card. Then put that tubing in the mold and smelt some copper so I have a full coverage block.

Then a chamber outside the full coverage pipes and outside the GPU blocks that would be encased in a clear acrylic chamber for visual effects.

I would have the entire block with only 1 in barb and 1 out barb. The water would flow in the following sequence

in barb -> full coverage copper tubing -> GPU #1 block -> GPU #2 block -> freeflow acrylic chamber -> out barb

I'm have various ideas for heatsinks for the second peltier set that pulls heat from the liquid, so don't assume I have overlooked that. I have not, I just haven't decided if I want to use a second set of liquid cooling for that, or if I don't to use fans, or most likely I will have a passive heatsink for it




ANY IDEA ;D
ZytheEKS is offline   QUOTE Thanks
Old 02-25-13, 12:14 AM   #2
Knufire
Senior Member

 
Knufire's Avatar 

Join Date: Feb 2010
Location: Terre Haute, IN

 
Wait, so you have a GPU cooled by a pelt cooled by a waterblock cooled by another pelt?

I get the first pelt, gives you awesome GPU temps. But I don't get the purpose of the 2nd pelt, that would give you better water temps, but do those even matter given that you have good GPU temps from the first set? They seem to provide no benefit and just increase the inefficiency/total heat you have to disperse. Why not just have a pelt on each GPU, and make a custom block out copper/acrylic to cool those and be done with it?

Last edited by Knufire; 02-25-13 at 12:21 AM.
Knufire is offline Benching Profile Heatware Profile   QUOTE Thanks
Old 02-25-13, 12:34 AM   #3
azanimefan
Member

 
azanimefan's Avatar 

Join Date: Oct 2012
Location: Phoenix, Az

 
don't forget the 2nd water chamber/waterloop cooling the 2nd peltier... he's got a 7990... which has 2 7970 cores on the card itself.

So he's planning on making a custom water loop which is really just system for cooling the peltiers he'll slap onto the two cpus... (2 blocks & 2 peltiers -> 2 cpus on one loop)... then he's building a custom water-chamber on the back of the gpu... he'll stick a 2nd peltier on the other side of the block to counteract the heat from the 1st peltier and submerge that side in the custom water-chamber, which i assume will then be part of it's own loop, separate from the water in the loop cooling the gpus...

several technical issues come to mind.

1) full copper tubing for the ingress would expose the incomeing water to the heated peltier cooling loop water. which would result in warming that loop prior to hitting the gpu.
2) peltiers put out something silly like 2 to 3 times the heat on their backside then one's frontside can cool. which means IF your backside peltiers CAN cool the water in the block enough to counter the extreme heat from the chipside peltier, the backside peltier will likely be putting out 4 to 6 times the heat energy the CHIPSIDE peltier is "cooling"...

sounds like you'll need a boatload of rad to cool that backside acrylic water loop

__________________
CPU: AMD Phenom II x4 965 OC 3.8GHz
MB: Asus M4A78T-E
RAM: G.Skill 8gb DDR3 1600
GPU: MSI HD 7770 OC/GHz Ed. 1GB GDDR5
HD: 2 Seagate Barracuda 40GB in RAID 0
HD: 1 WD Black 250gb
Case: Antec Nine-Hundred Two
OS: Windows 7 Ultimate (x64)

Folding User Stats
azanimefan is offline   QUOTE Thanks
Old 02-25-13, 12:47 AM   #4
Knufire
Senior Member

 
Knufire's Avatar 

Join Date: Feb 2010
Location: Terre Haute, IN

 
I just don't get the point of the peltier on the back side. Two pelts and two blocks for two GPUs...that just seems simpler, same performance, and less overall heat to cool. He could even just stick a custom giant single block to cool both pelts (one on each GPU). Why more than two pelts are advantageous, I don't understand.
Knufire is offline Benching Profile Heatware Profile   QUOTE Thanks
Old 02-25-13, 01:17 AM Thread Starter   #5
ZytheEKS
Registered



Join Date: Dec 2012

 
Knufire you make a good point, but the point of having that second peltier is so that the waterblock won't put out immense amounts of heat for my rads. The second peltier would not be submerged as azanimefan thought, it would have a passive cooler. Maybe an aircooler, not sure on that yet.

After reviewing your posts I think it would make more sense to put

in barb -> GPU #1 block -> GPU #2 block ->f ull coverage copper tubing -> freeflow acrylic chamber -> out barb


The freeflow acrylic chamber serves absolutely no significant purpose. It is for aesthetic purposes only ( I will probably add a temp probe and a flow rate sensor too, but it still doesn't serve any cooling purposes)

Basically after the cooling block has done it's job the fluid exits into a chamber, then the chamber has the fluid exit the system at the other end of the acrylic chamber.

Again, the acrylic part is mainly just for show.



The second peltier is crucial to this setup otherwise I will need probably another radiator to let this block work properly. Peltiers put out a LOT of heat, that second peltier is so that the heat from the first peltier is mostly removed from the water BEFORE it exists the waterblock and goes to the rest of my system
ZytheEKS is offline   QUOTE Thanks
Old 02-25-13, 01:24 AM   #6
Knufire
Senior Member

 
Knufire's Avatar 

Join Date: Feb 2010
Location: Terre Haute, IN

 
The physics just doesn't make sense here.

Say one GPU on the thing is 175W. Assuming the peltier is 50% efficient, that's 350W the 2nd peltier has to remove from the liquid. That means the hot side on the 2nd pelt is like 700W. You're going to passively cool 700W?

Those are nice numbers too, because a 7970 is like a 230-250W chip and pelts are potentially less efficient than 50%. You have an outright massive amount of heat off the 2nd pelt that needs some serious cooling.

Last edited by Knufire; 02-25-13 at 01:30 AM.
Knufire is offline Benching Profile Heatware Profile   QUOTE Thanks
Old 02-25-13, 01:24 AM Thread Starter   #7
ZytheEKS
Registered



Join Date: Dec 2012

 
Knufire you make a good point, but the point of having that second peltier is so that the waterblock won't put out immense amounts of heat for my rads. The second peltier would not be submerged as azanimefan thought, it would have a passive cooler. Maybe an aircooler, not sure on that yet.

After reviewing your posts I think it would make more sense to put

in barb -> GPU #1 block -> GPU #2 block ->f ull coverage copper tubing -> freeflow acrylic chamber -> out barb


The freeflow acrylic chamber serves absolutely no significant purpose. It is for aesthetic purposes only ( I will probably add a temp probe and a flow rate sensor too, but it still doesn't serve any cooling purposes)

Basically after the cooling block has done it's job the fluid exits into a chamber, then the chamber has the fluid exit the system at the other end of the acrylic chamber.

Again, the acrylic part is mainly just for show.



The second peltier is crucial to this setup otherwise I will need probably another radiator to let this block work properly. Peltiers put out a LOT of heat, that second peltier is so that the heat from the first peltier is mostly removed from the water BEFORE it exists the waterblock and goes to the rest of my system
ZytheEKS is offline   QUOTE Thanks
Old 02-25-13, 10:05 PM   #8
azanimefan
Member

 
azanimefan's Avatar 

Join Date: Oct 2012
Location: Phoenix, Az

 
mmm... no... knufire's point is valid. its what i was saying. there is gonna be way too much heat energy to deal with passively or otherwise.

__________________
CPU: AMD Phenom II x4 965 OC 3.8GHz
MB: Asus M4A78T-E
RAM: G.Skill 8gb DDR3 1600
GPU: MSI HD 7770 OC/GHz Ed. 1GB GDDR5
HD: 2 Seagate Barracuda 40GB in RAID 0
HD: 1 WD Black 250gb
Case: Antec Nine-Hundred Two
OS: Windows 7 Ultimate (x64)

Folding User Stats
azanimefan is offline   QUOTE Thanks
Old 02-25-13, 11:18 PM   #9
Conumdrum
Wants Old Title Back
Didn't Get It

 
Conumdrum's Avatar 

Join Date: Sep 2007
Location: Las Vegas

 
We seen these before. Once the OP learns more, he'll see why TECs don't work anymore.

It's always the hot TEC side that ruins the easy stuff.

__________________
I want my old title back.
Old beast is torn down.
Will update later.

900 watt (1500VA) UPS
27" ASUS VG 3D/Nvidia V2 3D glasses and 24" Acer H243H
Conumdrum is offline Author Profile   QUOTE Thanks
Old 02-26-13, 09:42 AM Thread Starter   #10
ZytheEKS
Registered



Join Date: Dec 2012

 
Water will be flowing through the block in-between the two TEC blocks.

My fear here is that if I remove the second TEC the GPUs and TEC will put out to much heat for a single loop, and I'll end up having to run in an additional loop. And the second additional peltier won't have the full load of the GPU and peltier, there will be liquid flowing through it. So the passive heatsink wouldn't be cooling anywhere near the roughly 500watts of a peltier+GPU

It cools the liquid flowing through the block so I can keep this all in a single contained device, instead of running in multiple loops.
As you said, 400-500watts is a lot of heat and having a single loop cooling 2 of these, not gunna happen.

As whether a passive heatsink can hand a single peltier cooling the water while it's flowing I guess that will have to be a trail and error process, if not I'll make a fan based heatsink for it/them.
ZytheEKS is offline   QUOTE Thanks
Old 02-26-13, 09:52 AM   #11
Bobnova
Innocent Senior Member



 
Bobnova's Avatar 

Join Date: May 2009
Location: Humboldt

 
Each TEC needs be about 2x the wattage of the heat you want to transfer.
So at stock with your 250w GPU core you're going to need a 400-500w TEC.
That means that the hot side of the TEC needs 650-750w of cooling, the wattage spec of a TEC is how much power it draws.
TEC's don't destroy heat, they only move it, and add their own to it.

Between the two cores and the TEC for each core you're looking at 1400w-1500w of heat your poor loop is going to need to get rid of.

That, of course, assumes that you can find a 400+ watt TEC. You'll need a special 24 V PSU for them, too.

TECs cooling water typically slowly freeze the water and plug the waterblock up.


Bottom line: TECs really don't work for high wattage chips.

__________________
"Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe." -- Einstein (maybe)


How to check your PSU with a multimeter.

17bXw5t51rEBXGavJFMJsC8g7HQgThUGc7

Heatware.
Bobnova is online now Author Profile Benching Profile Folding Profile Heatware Profile Rosetta Profile   QUOTE Thanks
Old 02-26-13, 11:09 AM Thread Starter   #12
ZytheEKS
Registered



Join Date: Dec 2012

 
=.= Damn, didn't think about the possibility of water freezing over the second TEC block. Damn, that would definitely cause immediate issues.

Will have to rethink the rig I guess
ZytheEKS is offline   QUOTE Thanks
Old 02-26-13, 11:30 AM   #13
Knufire
Senior Member

 
Knufire's Avatar 

Join Date: Feb 2010
Location: Terre Haute, IN

 
I don't think you understand how much heat would be coming off the 2nd TEC in this scenario. A trial and error process is almost certainly going to cause you to fry the chip. You would need a heatsink the size of your entire case, and that was with my nice numbers. With the numbers Bobnova posted (which I would trust to be more accurate), that's the amount of heat a small portable room heater would put out.
Knufire is offline Benching Profile Heatware Profile   QUOTE Thanks
Old 02-26-13, 01:49 PM Thread Starter   #14
ZytheEKS
Registered



Join Date: Dec 2012

 
You're assuming the second peltier would be pulling anywhere near the amount of even the chip. I have a pretty wicked rad setup on my cooling loop.
Again, that second peltier + passive heatsink is meant to cool the liquid... just the liquid, not the chip.


In reality if I just rigged my loop in with 3 separate inputs/outputs on that rig with only 2 peltiers (one per GPU) I would be able to easily run it... But again, having 3 cables going in and 3 coming out would look pretty gnarr

Going to have to rethink this all, hopefully I'll be able to get some idea together for it =/
ZytheEKS is offline   QUOTE Thanks
Old 02-26-13, 02:02 PM   #15
Bobnova
Innocent Senior Member



 
Bobnova's Avatar 

Join Date: May 2009
Location: Humboldt

 
The second TEC will be dissipating its heat plus as much as it can of the water's heat. Remember that "heat" is all relative to absolute zero, not to room temperature. There is plenty of heat for a TEC to transfer even if the cold side is already well below freezing.

__________________
"Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe." -- Einstein (maybe)


How to check your PSU with a multimeter.

17bXw5t51rEBXGavJFMJsC8g7HQgThUGc7

Heatware.
Bobnova is online now Author Profile Benching Profile Folding Profile Heatware Profile Rosetta Profile   QUOTE Thanks
Old 02-26-13, 02:06 PM   #16
Nebulous
Seņor Senior Member

 
Nebulous's Avatar 

Join Date: Oct 2002
Location: The Empire State

10 Year Badge
 
Quote:
Originally Posted by ZytheEKS View Post
You're assuming the second peltier would be pulling anywhere near the amount of even the chip. I have a pretty wicked rad setup on my cooling loop.
Again, that second peltier + passive heatsink is meant to cool the liquid... just the liquid, not the chip.


In reality if I just rigged my loop in with 3 separate inputs/outputs on that rig with only 2 peltiers (one per GPU) I would be able to easily run it... But again, having 3 cables going in and 3 coming out would look pretty gnarr

Going to have to rethink this all, hopefully I'll be able to get some idea together for it =/

Hmm, so if I understand this right you're going to use the 2nd Pelt to just cool the liquid ONLY?

What about the heat left from the 2nd pelt cooling that water? Where is it going to go? You're going to use a passive heatsink to cool that? Nope, not possible. I've done TEC/Pelt cooling and there's noway in hell a passive sink is going to remove that much heat. If there was a way, trust me I would know about it.

If you do find a method, let me in on it. Good luck with that. Just be sure to have a fire extinguisher handy.

__________________

<MEGA-O>
Gigabyte Z77X-UD4H /De-Lid i5 3570K@ 4.5Ghz
16GB GSKILL Sniper F3-17000CL11D-8GBSR
XSPC Rasa Black / RX360 / MCP-655 / MCRes
Corsair Force 3 SSD 120GB x2 Raid~0 / WD VelociRaptor 300GB
Diamond 7870PE52GDD Ghz Edition HD 7870+H80
Sony AD-7220S / Lite On LH-16A1S
SilverStone Zeus ZM1350 1350w
CM Cosmos 1000 / Win7 x64
** Under TITLE II, SECTION 210.340 it is legal to overclock **
Nebulous is offline Heatware Profile   QUOTE Thanks
Old 02-26-13, 02:13 PM Thread Starter   #17
ZytheEKS
Registered



Join Date: Dec 2012

 
Either a passive heatsink or fan based heatsink. I don't plan to use a super powerful peltier for the second one, it's just to pull enough of the heat out so that I can keep it all in a single loop.
http://www.shop.customthermoelectric...7.qscstrfrnt06
Again, 1 peltier has it's cold side lay on the GPU and the hot side on that block, the other peltier has the cold side on the liquid so that the heat from the other peltier is dissipated enough so I can keep the liquid loop in a single loop, instead of having it run in multiple loops for that single device.
Again, after reading all this I know my design is a dud and I'll have to rethink the design.

Going to try and rethink my design so I can keep the peltiers on the GPUs, will get back with a new design when I do.
ZytheEKS is offline   QUOTE Thanks
Old 02-26-13, 02:25 PM   #18
Bobnova
Innocent Senior Member



 
Bobnova's Avatar 

Join Date: May 2009
Location: Humboldt

 
You'll still have 1500w to get rid of if you're sizing the GPU TECs correctly.
A 150w TEC isn't going to make much of a dent in that.

What you need for TEC cooling GPUs is GPUs that draw less than 100w.

__________________
"Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe." -- Einstein (maybe)


How to check your PSU with a multimeter.

17bXw5t51rEBXGavJFMJsC8g7HQgThUGc7

Heatware.
Bobnova is online now Author Profile Benching Profile Folding Profile Heatware Profile Rosetta Profile   QUOTE Thanks
Old 02-26-13, 03:29 PM   #19
ivanlabrie


 
ivanlabrie's Avatar 

Join Date: Jul 2011
Location: Buenos Aires, Argentina

 
Bobnova is spot on...
TEC's don't work for more modern gpu's cause of the massive heatloads.
They are nowhere near efficient and introduce a lot of heat to dissipate added to the gpu cores own.
If you want subzero, get a phase change unit or build one. Or get massive peltiers place them over the gpu cores and cool the hot sides with your uber loop. Only way it would work, and it would require a second psu just for the tecs.

__________________
Folding User Stats
"What it can take scales inversely with how much you want it to not die." Bobnova
ivanlabrie is online now Benching Profile Heatware Profile   QUOTE Thanks
Old 02-26-13, 03:56 PM Thread Starter   #20
ZytheEKS
Registered



Join Date: Dec 2012

 
You know what? **** it, I'm just gunna get an extension plug for the PCI-E and submerge the ****in' card in liquid nitrogen, problem solved (Is joking obviously)
ZytheEKS is offline   QUOTE Thanks

Post Reply New Thread Subscribe


Overclockers Forums > Hardware > Cooling > Water Cooling
Water Cooling Discussion devoted to blocks, pumps, radiators, reservoirs, tubing, and everything else to get you running smooth on a water loop
Forum Jump

Thread Tools Search this Thread
Search this Thread:

Advanced Search


Mobile Skin
All times are GMT -5. The time now is 11:15 PM.
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
You can add these icons by updating your profile information to include your Heatware ID, Benching Profile ID or your Folding/SETI profile ID. Edit your profile!
X

Welcome to Overclockers.com

Create your username to jump into the discussion!

New members like you have made this the best community on the Internet since 1998!


(4 digit year)

Why Join Us?

  • Share experience
  • Max out your hardware
  • Best forum members anywhere
  • Customized forum experience

Already a member?