Overclock.net › Forums › Cooling › Water Cooling › Heavily modifying an A/C evaporator coil to use as a radiator for my loop.
New Posts  All Forums:Forum Nav:

Heavily modifying an A/C evaporator coil to use as a radiator for my loop.

post #1 of 21
Thread Starter 
New schtuff on page 3.

Currently the radiator is 48-pass and gives me about .5GPM flow on my pump, which is obviously not enough. So I decided to do some serious changes about how the water circulates through it.

Here are 3 (or 4) designs that I'm thinking of. What do you think will be most effective?



Grainy-picture goodness:

Edited by Khaotik55 - 5/14/11 at 7:17pm
post #2 of 21
Without detailed modelling of the flows and heat transfer it would be difficult to say which will perform the best thermally. Personally I would go with option 1 as it will be easier on your pump, allowing you to maintain a higher flow rate.

One thing I can tell you though, have the hot water entering on the air exhaust side, not the inlet side. This makes it into more of a counter flow heat exchanger than a parallel flow (A on diagram) which are more efficient:



From the Wikipedia article
Main
(21 items)
 
HTPC
(10 items)
 
 
CPUMotherboardGraphicsRAM
i5 2550k P8P67 Pro Sapphire HD 7950 G.Skill RipJaws X 1600 Cas 9 
Hard DriveHard DriveHard DriveCooling
Corsair Force 120 WD Blue 500GB WD Caviar Green 1TB XSPC RayStorm 
CoolingCoolingCoolingCooling
RX240 MCR 220 EK 7950 Copper Acetal  DDC-1T 
OSMonitorMonitorKeyboard
Windows 7 64-bit Dell U2311H Oculus Rift DK2 Ducky Shine 3 MX Brown 
PowerCaseMouseAudio
Corsair TX 750W CoolerMaster CM690 II G500 Klipsch ProMedia 2.1 
Audio
Asus Xonar DX 
CPUMotherboardRAMHard Drive
A10-6800K Gigabyte GA-F2A85XN-WIFI G Skill 1600 CAS9 Kingston SSD Now 60GB 
Hard DriveOptical DriveCoolingOS
WD Caviar Blue 1TB LG Slim Blu-Ray player Silverstone NT06-PRO  Widows 7 Home Premium 
PowerCase
Silverstone Sfx Series ST45SF 450W Silverstone SG05 
  hide details  
Reply
Main
(21 items)
 
HTPC
(10 items)
 
 
CPUMotherboardGraphicsRAM
i5 2550k P8P67 Pro Sapphire HD 7950 G.Skill RipJaws X 1600 Cas 9 
Hard DriveHard DriveHard DriveCooling
Corsair Force 120 WD Blue 500GB WD Caviar Green 1TB XSPC RayStorm 
CoolingCoolingCoolingCooling
RX240 MCR 220 EK 7950 Copper Acetal  DDC-1T 
OSMonitorMonitorKeyboard
Windows 7 64-bit Dell U2311H Oculus Rift DK2 Ducky Shine 3 MX Brown 
PowerCaseMouseAudio
Corsair TX 750W CoolerMaster CM690 II G500 Klipsch ProMedia 2.1 
Audio
Asus Xonar DX 
CPUMotherboardRAMHard Drive
A10-6800K Gigabyte GA-F2A85XN-WIFI G Skill 1600 CAS9 Kingston SSD Now 60GB 
Hard DriveOptical DriveCoolingOS
WD Caviar Blue 1TB LG Slim Blu-Ray player Silverstone NT06-PRO  Widows 7 Home Premium 
PowerCase
Silverstone Sfx Series ST45SF 450W Silverstone SG05 
  hide details  
Reply
post #3 of 21
Thread Starter 
Quote:
Originally Posted by GingerJohn View Post
Without detailed modelling of the flows and heat transfer it would be difficult to say which will perform the best thermally. Personally I would go with option 1 as it will be easier on your pump, allowing you to maintain a higher flow rate.

One thing I can tell you though, have the hot water on the air exhaust side, not the inlet side. This makes it into more of a counter flow heat exchanger than a parallel flow (A on diagram) which are more efficient:



From the Wikipedia article
Yeah otherwise the fan will be blowing the hottest air over the cooler lines. Also the tubing is 3/8OD. I'm just worried about using the #1 option because the water might be moving through it too fast. Since it's meant for gas and not liquid it's not very efficient therefore I'd like the water to pass through it more. Am I thinking correctly?
Edited by Khaotik55 - 5/11/11 at 3:16pm
post #4 of 21
Firstly, I'm interested to know, what is your pump and what are you cooling on that loop?


I guess it's the easy way out, but I did wonder; could you just throw money at it and get a bigger pump or double up the pump you are currently using?


Two pumps in parallel for example would almost double the flow. You may get better results from series however, since increasing the flow will increase the pressure drop on the loop... (thus decreasing flow more)



Also, I think option one would be the easiest of those options too. While it may not be the most thermally efficient, I'd have a hard time imagining it not dumping the heat...



P.S.

I really wish more stuff had pressure drop graphs associated with them... Would make designing perfect liquid cooling loops so easy lol.
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
post #5 of 21
Thread Starter 
Quote:
Originally Posted by Crabid View Post
Firstly, I'm interested to know, what is your pump and what are you cooling on that loop? Nothing you'd normally find on a water cooling loop. Little Giant 2E-38N. Quite weak actually. +CPU only loop, lol.


I guess it's the easy way out, but I did wonder; could you just throw money at it and get a bigger pump or double up the pump you are currently using?
I could.. however that would be another 100-150 dollars.

Two pumps in parallel for example would almost double the flow. You may get better results from series however, since increasing the flow will increase the pressure drop on the loop... (thus decreasing flow more)
What do you mean by series, a 2nd pump directly after the first?


Also, I think option one would be the easiest of those options too. While it may not be the most thermally efficient, I'd have a hard time imagining it not dumping the heat...
It does seem like a lot of time to transfer the heat, but this coil is quite inefficient, and hard to push air through I might add.



P.S.

I really wish more stuff had pressure drop graphs associated with them... Would make designing perfect liquid cooling loops so easy lol.
I replied in bold.

I have another coil that is only about an inch and a half thick but is extremely dense and transfers heat well. It also uses 1/4" OD lines which makes it's even more efficient because the water is flowing through thin copper lines. This is already in a 30 pass configuration but within those 30 passes it has 5 circuits, weird, I know. Each circuit makes water flow in(+1 pass), goes through 4 more passes, and then goes back into the output manifold(+1 pass). and I get about 1.4GPM with this setup.
Edited by Khaotik55 - 5/11/11 at 4:45pm
post #6 of 21
No, sorry, I meant pumps in series...


2 pumps in series have the same maximum flow and double the maximum pressure of one pump

2 pumps in parallel have the same pressure but double the maximum flow of one pump.

Really need a graph to explain this lol, but your loop will have a pressure drop for a given flow based on the restriction of the loop.

So if you increase the flow, the pressure drop increases and the flow will be reduced by the loop more, but because you increased it by double, the net result is an increase in flow.

On the flip side, if you double the pressure, your flow will be reduced less, but your maximum flow hasn't changed.


Let me see if I can't dig a graph out if gingerjohn doesn't beat me to it, I know he has some
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
post #7 of 21
Quote:
Originally Posted by Crabid View Post
Let me see if I can't dig a graph out if gingerjohn doesn't beat me to it, I know he has some
You mean like this?



Credit for this one to MartinM210.

See Martin's testing for proof of series being better than parallel.
Edited by GingerJohn - 5/11/11 at 4:35pm
Main
(21 items)
 
HTPC
(10 items)
 
 
CPUMotherboardGraphicsRAM
i5 2550k P8P67 Pro Sapphire HD 7950 G.Skill RipJaws X 1600 Cas 9 
Hard DriveHard DriveHard DriveCooling
Corsair Force 120 WD Blue 500GB WD Caviar Green 1TB XSPC RayStorm 
CoolingCoolingCoolingCooling
RX240 MCR 220 EK 7950 Copper Acetal  DDC-1T 
OSMonitorMonitorKeyboard
Windows 7 64-bit Dell U2311H Oculus Rift DK2 Ducky Shine 3 MX Brown 
PowerCaseMouseAudio
Corsair TX 750W CoolerMaster CM690 II G500 Klipsch ProMedia 2.1 
Audio
Asus Xonar DX 
CPUMotherboardRAMHard Drive
A10-6800K Gigabyte GA-F2A85XN-WIFI G Skill 1600 CAS9 Kingston SSD Now 60GB 
Hard DriveOptical DriveCoolingOS
WD Caviar Blue 1TB LG Slim Blu-Ray player Silverstone NT06-PRO  Widows 7 Home Premium 
PowerCase
Silverstone Sfx Series ST45SF 450W Silverstone SG05 
  hide details  
Reply
Main
(21 items)
 
HTPC
(10 items)
 
 
CPUMotherboardGraphicsRAM
i5 2550k P8P67 Pro Sapphire HD 7950 G.Skill RipJaws X 1600 Cas 9 
Hard DriveHard DriveHard DriveCooling
Corsair Force 120 WD Blue 500GB WD Caviar Green 1TB XSPC RayStorm 
CoolingCoolingCoolingCooling
RX240 MCR 220 EK 7950 Copper Acetal  DDC-1T 
OSMonitorMonitorKeyboard
Windows 7 64-bit Dell U2311H Oculus Rift DK2 Ducky Shine 3 MX Brown 
PowerCaseMouseAudio
Corsair TX 750W CoolerMaster CM690 II G500 Klipsch ProMedia 2.1 
Audio
Asus Xonar DX 
CPUMotherboardRAMHard Drive
A10-6800K Gigabyte GA-F2A85XN-WIFI G Skill 1600 CAS9 Kingston SSD Now 60GB 
Hard DriveOptical DriveCoolingOS
WD Caviar Blue 1TB LG Slim Blu-Ray player Silverstone NT06-PRO  Widows 7 Home Premium 
PowerCase
Silverstone Sfx Series ST45SF 450W Silverstone SG05 
  hide details  
Reply
post #8 of 21
Thread Starter 
The coil I'm referring to in my previous post is not square as the coil in my OP was, but extremely rectangle, it has 3 sections of coil that is about 4.5 inches wide, ~1 inch thick, and about 3 feet long. Bah I'll just take a picture. I will post it below when uploaded.





I was actually going to take this coil and un-solder all of the joints, take 2 of the 3 sections and stack them, then cut it in half so I have a normal-sized radiator you'd typically find in a computer.

=============
The final outcome:
It would be 19 pass but 5 inputs(Single input welded into a manifold which splits into 5 fingers) and 5 outputs(same dealio, manifold which joins 5 fingers together into one output), each finger having it's own 4-pass circuit except for one which would be 2 pass(Just the input finger and output finger itself due to the way I'm cutting it). It's half length, and missing 1 section, so I'd probably get more than a gallon and a half per minute due to about 66% reduction in overall radiator size. It's also just about the perfect width for a typical computer fan like a Gentle Typhoon or something. Actually after looking at that stat I just calculated, 66 is a lot gone, I might just keep the full 3 foot length and cut off the third section and just stack the two. so it's only a 33% reduction.
=============
Well that was confusing.

Edit: Here I made a picture to explain it better, maybe...


This is actually my main project but if this fails to take out the heat I will have to modify the large square one.
Edited by Khaotik55 - 5/12/11 at 7:32pm
post #9 of 21
Almost, was thinking more like this:



Source

but with the restriction curves on so you can actually see the increase of actual flow.


I know Martin did an awesome one, but that was a week or so ago and I've been through a lot of post since then
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
post #10 of 21
If you can get the flow high enough to get turbulent flow in your CPU, this will dump the heat. CPU only isn't that much and that radiator is frankly huge.

Edit:

That other coil looks perfect tbh. 1.4GPM...



Also, just thinking about the original radiator...

If you switched from 48 to say 24 pass, you would halve the amount of times it's running through, but also double the channel cross sectional area.

I think you'd end up with about 1/4 of the restriction you had before, so 19 pass would in fact work well in all likelyhood.

That's some pretty rough estimates though.
Edited by Crabid - 5/11/11 at 5:17pm
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
Overclock.net › Forums › Cooling › Water Cooling › Heavily modifying an A/C evaporator coil to use as a radiator for my loop.