Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › optimize the cold loop in a TEC-chilled dual loop system
New Posts  All Forums:Forum Nav:

optimize the cold loop in a TEC-chilled dual loop system

post #1 of 9
Thread Starter 
Hi!

I’m changing from a traditional single loop (reservoir/pump > triple-radiator > cpu > gpu >) into a dual loop TEC chill cooling system:
[hot loop] reservoir/pump#1 > triple-radiator > chill-TEC > gpu >
[cold loop] reservoir/pump#2 > chill-TEC > cpu >

To start in the ‘TEC world’, I’m planning to run it with a controller for sub-ambient temps, but above dew point (not to have to insulate, at this moment). CPU is the i7-3930k.

I understand that most principles from a traditional water-cooling single loop would apply also for the cold loop, but at least two things are different: the cold loop is running sub-ambient temps; and, as a very simple loop, is much more flow unrestricted.

So, like the title says, I would like to ask your expert advice on how to best optimize the cold loop in a TEC-chilled dual loop system?… more specifically, the following:

1) Reservoir: big-fat-one, small-reservoir, or no-reservoir-at-all?

My n00b idea: I’m tempted to go with no reservoir at all in my cold loop, to keep it simple; filling the system with a T, but I’m afraid to lose efficiency for the extra difficulty to bleed the air out. Going no reservoir is a good idea or not? What do you think?

2) Pump: flow or head pressure?

My n00b opinion: for my cold loop, I have spare a D5 (24v) and a DDC (12v), so I can choose. By the specs, the D5 has better flow, but the DDC has higher head pressure. I was thinking to go with the D5 ‘full-out’ (at 24v), but isn’t that a bit overkill for such a simple and unrestricted loop (so I can safely run it regulated for less noise)?

3) CPU block: what type of CPU block could be labeled ‘ideal’ for the cold loop of a TEC-chilled system?

My n00b thoughts: currently most CPU blocks are being designed flow-agnostic or high-flow, because are meant for single loop systems with several blocks/radiators, where restriction can impair performance, especially if the pump is not powerful enough. However, I wonder if the same ‘gimme-gimme-more-flow’ theory is ideal also for such a simple (pump>cpu) loop?
Other thought is on the sub-ambient temperatures… I wonder if the size (and weight) of the cooler block would affect its performance for a cold loop, running sub-ambient temps? Because the cooler block itself, in contact with (warmer) air, would also transfer back into the air some of the coldness… and if that is an advantage for a warm loop system, it actually seems a disadvantage for a cold loop.
So, here I’m lost without a clue… and since I have to buy a new CPU block would appreciate very much your advice on this (I looked at some candidates ‘EK-Supremacy nickel’, ‘Koolance 370’, ‘XSPC Raystorm copper’, ‘Swiftech Apogee’… some include options for more/less restriction).


4) Thermal paste: anything goes or some perform better at sub-ambient temps?

My n00b finding: I’m a fan of the old good ‘artic silver’ (conductive, yes... but if one cant handle it without smear paste over the entire motherboard, maybe shouldnt be installing CPUs, any way...). I recognize other products came up and currently perform better. However, since most tests, reviews, benchmarks and comparisons (I know of) were done with air-coolers or hot-loop water-cooling, I’m not sure if we can extrapolate such performance for a TEC-chilled cold loop… can we? Or these thermal pastes can perform different depending on the loop temperature? For my cold loop I have here, spare, ‘artic silver 5’ and ‘MX-2’, as options… but don’t know what is tested, trusted, and proved very good for a TEC chilled system. Any help?

Many thanks!
Edited by TAC TEC TIC - 7/24/12 at 3:38pm
post #2 of 9
Here is my opinion:

1) Reservoir:
Two ways (or balance between them):
1) a small reservoir, for fast temperature pull down and easy management of the loop. If you have space limitation then this is the way to go.
2) a very large reservoir (insulated) for stable low temperature during benching (especially in winter when dew is close to/below zero). You could pull down the water temperature to close zero and then bench. It will take a while before water temp climbs back.

2) Pump:
If you use a less restrictive cpu block, I would choose DDC for its small size and lower heat dump into the cold loop. But It also depends on how restrictive your chiller is. In dry days you may run water close to freezing point, high flow will reduce the risk of freezing/blocking.

3) CPU block:
Any good performance high flow block. Less metal surface exposure will help retain cold (quite significantly if water temperature is low).

4) Thermal paste:
Yes, some TIM is not good at low temperature. Most manufacturer don't even give this information. You may ask LN2 or phase section about the selection of TIM. I have used MX-4 down to -25C and it seems good.

What kind of chiller/TEC are you going to use?
Obsidian
(22 items)
 
CPUMotherboardGraphicsRAM
i7-3930K ASUS P9X79 Deluxe XFX HD 7970 Black Edition Double Dissipation Patriot Viper Xtreme Series 4x 4GB DDR3 2000 9-... 
Hard DriveHard DriveHard DriveCooling
OCZ RevoDrive3 X2 OCZ Vertex 4 WD 640GB  Swiftech 360 Radiator 
CoolingCoolingCoolingCooling
MCP 35X pump Swiftech Apogee XT waterblock Foxrena's Water Rack KOC TEC and Fan controller 
CoolingOSMonitorKeyboard
KOC 62mm TEC block Windows 7 Ultimate Dell UltraSharp IPS 27'' 2560x1440 Razor Black Widow Ultimate 
PowerCaseMouseAudio
Corsair AX1200i Corsair 800D Logitech G9X Logitech Z4 
  hide details  
Reply
Obsidian
(22 items)
 
CPUMotherboardGraphicsRAM
i7-3930K ASUS P9X79 Deluxe XFX HD 7970 Black Edition Double Dissipation Patriot Viper Xtreme Series 4x 4GB DDR3 2000 9-... 
Hard DriveHard DriveHard DriveCooling
OCZ RevoDrive3 X2 OCZ Vertex 4 WD 640GB  Swiftech 360 Radiator 
CoolingCoolingCoolingCooling
MCP 35X pump Swiftech Apogee XT waterblock Foxrena's Water Rack KOC TEC and Fan controller 
CoolingOSMonitorKeyboard
KOC 62mm TEC block Windows 7 Ultimate Dell UltraSharp IPS 27'' 2560x1440 Razor Black Widow Ultimate 
PowerCaseMouseAudio
Corsair AX1200i Corsair 800D Logitech G9X Logitech Z4 
  hide details  
Reply
post #3 of 9
Quote:
Originally Posted by TAC TEC TIC View Post

Hi!
I’m changing from a traditional single loop (reservoir/pump > triple-radiator > cpu > gpu >) into a dual loop TEC chill cooling system:
[hot loop] reservoir/pump#1 > triple-radiator > chill-TEC > gpu >
[cold loop] reservoir/pump#2 > chill-TEC > cpu >
To start in the ‘TEC world’, I’m planning to run it with a controller for sub-ambient temps, but above dew point (not to have to insulate, at this moment). CPU is the i7-3930k.
I understand that most principles from a traditional water-cooling single loop would apply also for the cold loop, but at least two things are different: the cold loop is running sub-ambient temps; and, as a very simple loop, is much more flow unrestricted.
So, like the title says, I would like to ask your expert advice on how to best optimize the cold loop in a TEC-chilled dual loop system?… more specifically, the following:
1) Reservoir: big-fat-one, small-reservoir, or no-reservoir-at-all?
My n00b idea: I’m tempted to go with no reservoir at all in my cold loop, to keep it simple; filling the system with a T, but I’m afraid to lose efficiency for the extra difficulty to bleed the air out. Going no reservoir is a good idea or not? What do you think?

I recommend no resevior, more turns on the water the better and more responsive your chilled loop will be. Less water in the loop equals more turns of that water. I've gotten much better results with a small loop vs a large one.
Quote:
2) Pump: flow or head pressure?
My n00b opinion: for my cold loop, I have spare a D5 (24v) and a DDC (12v), so I can choose. By the specs, the D5 has better flow, but the DDC has higher head pressure. I was thinking to go with the D5 ‘full-out’ (at 24v), but isn’t that a bit overkill for such a simple and unrestricted loop (so I can safely run it regulated for less noise)?
3) CPU block: what type of CPU block could be labeled ‘ideal’ for the cold loop of a TEC-chilled system?
My n00b thoughts: currently most CPU blocks are being designed flow-agnostic or high-flow, because are meant for single loop systems with several blocks/radiators, where restriction can impair performance, especially if the pump is not powerful enough. However, I wonder if the same ‘gimme-gimme-more-flow’ theory is ideal also for such a simple (pump>cpu) loop?
Other thought is on the sub-ambient temperatures… I wonder if the size (and weight) of the cooler block would affect its performance for a cold loop, running sub-ambient temps? Because the cooler block itself, in contact with (warmer) air, would also transfer back into the air some of the coldness… and if that is an advantage for a warm loop system, it actually seems a disadvantage for a cold loop.
So, here I’m lost without a clue… and since I have to buy a new CPU block would appreciate very much your advice on this (I looked at some candidates ‘EK-Supremacy nickel’, ‘Koolance 370’, ‘XSPC Raystorm copper’, ‘Swiftech Apogee’… some include options for more/less restriction).

I'd go with a higher flow block design in my opinion, I'd also go with a high flow plate for the chiller as well. The more flow generally = better c/w and the better the loop will respond to changing loads. Again more turns you can put through your chiller the better off you will be.
Quote:
4) Thermal paste: anything goes or some perform better at sub-ambient temps?
My n00b finding: I’m a fan of the old good ‘artic silver’ (conductive, yes... but if one cant handle it without smear paste over the entire motherboard, maybe shouldnt be installing CPUs, any way...). I recognize other products came up and currently perform better. However, since most tests, reviews, benchmarks and comparisons (I know of) were done with air-coolers or hot-loop water-cooling, I’m not sure if we can extrapolate such performance for a TEC-chilled cold loop… can we? Or these thermal pastes can perform different depending on the loop temperature? For my cold loop I have here, spare, ‘artic silver 5’ and ‘MX-2’, as options… but don’t know what is tested, trusted, and proved very good for a TEC chilled system. Any help?
Many thanks!

I've used mx-2 with good results. I've also used artic silver and had no problems going sub ambient as well as subzero with it. I think the mx-2 is the goto paste for subzero temps.
post #4 of 9
Quote:
Originally Posted by foxrena View Post

Here is my opinion:
1) Reservoir:
Two ways (or balance between them):
1) a small reservoir, for fast temperature pull down and easy management of the loop. If you have space limitation then this is the way to go.
2) a very large reservoir (insulated) for stable low temperature during benching (especially in winter when dew is close to/below zero). You could pull down the water temperature to close zero and then bench. It will take a while before water temp climbs back.
2) Pump:
If you use a less restrictive cpu block, I would choose DDC for its small size and lower heat dump into the cold loop. But It also depends on how restrictive your chiller is. In dry days you may run water close to freezing point, high flow will reduce the risk of freezing/blocking.
3) CPU block:
Any good performance high flow block. Less metal surface exposure will help retain cold (quite significantly if water temperature is low).
4) Thermal paste:
Yes, some TIM is not good at low temperature. Most manufacturer don't even give this information. You may ask LN2 or phase section about the selection of TIM. I have used MX-4 down to -25C and it seems good.
What kind of chiller/TEC are you going to use?

lol.. damn, you ninja posted as I was hitting the submit button..

I agree if you use a big loop, insulation is a must and will take longer to respond to loads. It will be slower to get down to target temps, but as a result will hold lower temps longer with a change in load.
Edited by Mindchi|l - 7/24/12 at 5:10pm
post #5 of 9
One thing I've been dying to try is using the new swiftech block/pump combo with a chiller. It would make that second loop a non issue

post #6 of 9
Thread Starter 
WOW!
Fast and great comments! Thank you both!
Chiller is from Arqtik.


- Reservoir...

For pure bench, the idea of insulating the reservoir is brilliant... pre-cooling a set of water to delay the warm up.
Maybe I can use one of these... I have at hand, for camping:

This bottle-parka can handle bottle/reservoir with 1 litter of coolant... and can fit easily inside the case... a 30 gallons barrel won't fit tongue.gif

Based on the same principle, the CPU block would certainly benefit also for some insulation, at least on the exposed metal surface.

For 24/7, I'm still tempted to go with no reservoir, just need to work out the bleeding.


- Pump...

Indeed, all pumps dump some bit of extra-heat into the loop, and I missed considering that when comparing these 2 pumps.
If the DDC pump dumps signifcantly less heat into the system, that is the way to go... however, the flow capability specs of the D5 @24v is more than double the DDC @12v: 18.9L/min (5 gal/min) vs 8.7 L/min (2.3 gal/min)... so much extra flow might pay off...


- TIM...

I will follow your advice and use MX-4... since is from the same manufacture of MX-2 must be quality stuff also.
post #7 of 9
Quote:
Originally Posted by Mindchi|l View Post

lol.. damn, you ninja posted as I was hitting the submit button..
biggrin.gif
Quote:
Originally Posted by Mindchi|l View Post

One thing I've been dying to try is using the new swiftech block/pump combo with a chiller. It would make that second loop a non issue
Quote:
Originally Posted by TAC TEC TIC View Post

Indeed, all pumps dump some bit of extra-heat into the loop, and I missed considering that when comparing these 2 pumps.
If the DDC pump dumps signifcantly less heat into the system, that is the way to go... however, the flow capability specs of the D5 @24v is more than double the DDC @12v: 18.9L/min (5 gal/min) vs 8.7 L/min (2.3 gal/min)... so much extra flow might pay off...

I recommend this swiftech block/pump, too (Unless you don't like its look.)
According to Martins' test, once you reach above 1gal/min, extra flow will not have a noticeable impact. So a DDC may suffice.
Obsidian
(22 items)
 
CPUMotherboardGraphicsRAM
i7-3930K ASUS P9X79 Deluxe XFX HD 7970 Black Edition Double Dissipation Patriot Viper Xtreme Series 4x 4GB DDR3 2000 9-... 
Hard DriveHard DriveHard DriveCooling
OCZ RevoDrive3 X2 OCZ Vertex 4 WD 640GB  Swiftech 360 Radiator 
CoolingCoolingCoolingCooling
MCP 35X pump Swiftech Apogee XT waterblock Foxrena's Water Rack KOC TEC and Fan controller 
CoolingOSMonitorKeyboard
KOC 62mm TEC block Windows 7 Ultimate Dell UltraSharp IPS 27'' 2560x1440 Razor Black Widow Ultimate 
PowerCaseMouseAudio
Corsair AX1200i Corsair 800D Logitech G9X Logitech Z4 
  hide details  
Reply
Obsidian
(22 items)
 
CPUMotherboardGraphicsRAM
i7-3930K ASUS P9X79 Deluxe XFX HD 7970 Black Edition Double Dissipation Patriot Viper Xtreme Series 4x 4GB DDR3 2000 9-... 
Hard DriveHard DriveHard DriveCooling
OCZ RevoDrive3 X2 OCZ Vertex 4 WD 640GB  Swiftech 360 Radiator 
CoolingCoolingCoolingCooling
MCP 35X pump Swiftech Apogee XT waterblock Foxrena's Water Rack KOC TEC and Fan controller 
CoolingOSMonitorKeyboard
KOC 62mm TEC block Windows 7 Ultimate Dell UltraSharp IPS 27'' 2560x1440 Razor Black Widow Ultimate 
PowerCaseMouseAudio
Corsair AX1200i Corsair 800D Logitech G9X Logitech Z4 
  hide details  
Reply
post #8 of 9
Thread Starter 
Quote:
I recommend this swiftech block/pump, too

Thanks... lets add it to the loop, then!

A schematic:

post #9 of 9
Quote:
Originally Posted by TAC TEC TIC View Post

Thanks... lets add it to the loop, then!
A schematic:

I love your schematic.. looks awsome!!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › optimize the cold loop in a TEC-chilled dual loop system