A hybrid guide to chill CPUs ?

[faziten]

Hello people, i was doing some research arround ways of cooling CPUs, (i've got myself a darn hot 3570k... silicon lotery D: ) and arrived at this youtube video about PELTIER.

(there's no need to watch the video... continue reading please).

 

https://www.youtube.com/watch?v=DHFGY32hV7w

 

Let me stop right here, I WILL NOT TALK ABOUT PELTIER, CONTINUE READING PLEASE. (if you were moved by "WTF is peltier?", wikipedia is your friend, but still not the main topic of this thread.)

 

After watching the video, i went to comment section. Some random dude threw an amazing idea. He claims to use it as daily driver to have... attention: ONLY 8ºC OVER ROOM TEMP IN A 3570 @ 4.2GHZ in CRYSIS 3, BF4 and 15ºC in Intel Burn Test. That if i recall right is the lowest temp (Liquid nitrogen aside) i've ever heard. And its buyable!.

 

How is it possible? Silverstone NT06 + H60 COMBINED!

 

Let's quote the full comment (just in case it desappears b4 u read it).

 

(...) Best is hybrid system of active water + passive air. Cheap $120, H60+Silverstone NT06 fanless air cooler.

i5-3570 at 4200 Crysis 3 & BF4, hasn't exceeded 8 degC above ambient.

 

 

Colonel Wilhelm Klink

 

 

OC 3570 to 4200 -
Burn test max 46C - ambient 31C - idle 33C

Traducir

 

 

 

Colonel Wilhelm Klink

 

 

+Michael Mikkel Tech Remove heat sink from back of NT06 (2 screws?). Mill fins off heat sink leaving dead-flat surface and reattach to NT06 using arctic silver or similar and screws. Remove NT06 fin shroud.
Place NT06 sideways on cpu. External heat pipes should be visible on right side of cpu. Mount H60 pump-block on the milled surface of the NT06 heat sink. Pump-block is placed so coolant lines exit vertically down from right side of cpu under heat pipes. Rad and fan mounted (normally) internally on rear of case (Antec 900 gaming case). 
May need to move ram to outside slots depending on MB.
I had to on Z77 sabre-tooth. Obviously arctic silver between CPU, NT06 and Pump-block. My H60 (V.1) pump-block surface wasn't flat either. I shaved it too. (H60 V.2 looks to be better quality)

Cheap, easy and looks great.
Should be enough info..

 

Finaly some contextual jadda jadda you might not read:

 

For a i5 3570@4200Mhz (in this case 3570 or 3570K would be the same) in games like crysis3 and BF4 implies almost 100% in CPU usage (maxed out settings and specific zones, the FPU component in the CPU DIE, is stressed out like hell and of course the cheap thermal intel threw inside does not help a tiny bit. Same thing occurs in Far Cry 3, Saint Rows IV, and other modern games). In this particular chip, it's expected to have arround 20 to 30ºC over room temperature at those clocks even with noctua NH-D14 of H100i coolers, AiO. So, if this is 100% true it would be outstanding 60+% thermal performance with 70% pricetag, and very little inconvenience to PC master race rookies. 

I have no money to test this right now (i lost my job few mounths ago) and even given the money i might not have change to get my hands on theese set of coolers (shitty government import policies). Anyway if some of you think this might work of have something similar in mind please share

 

May the tech be with you.

[Butterfish]

Can't really visualize the concept without some pictures  :wacko:

[vern021]

Can't really visualize the concept without some pictures  :wacko:

^this. I can't even think how that air cooler will be modded to the H60.

[faziten]

+

 

Remove the black tiny heatsink at the end of the copper pipes in the Silverstone, attach the H60 to the silverstone's naked copper pipes end.

It ends CPU > Thermal goo > silverstone copperpipes > Thermal goo > H60 cooling pipes. in theory, the heat remaining after the H60 is maxed out is discipated by the silverstone fins. 

[Thorium19]

I get the concept, and I can see what the guy means, but unless someone here was to test it and provide evidence of results, then I'm not sure I will be trusting it to be that good in performance. It'll probably give a boost, but that big of a boost is unrealistic IMO on air and water, especially when custom WC loops often don't get that low.

[Marvzl1357]

It seems...well...bullshit.

Just coz you added a watercooler to a metal heatsink does not mean that suddenly the temperature will drop so drastically. Don't you think someone (oh, I don't know, silverstone themselves) would have already come up with such a cooler if the gains were that insane?

 

also:

It ends CPU > Thermal goo > silverstone copperpipes > Thermal goo > H60 cooling pipes. in theory, the heat remaining after the H60 is maxed out is discipated by the silverstone fins. 

 Unfortunately that's not how thermodynamics works, the heat would go in parallel to the silverstone fins and the h60 cooler until it reaches the equilibrium of input heat=heat dissipated.

 

so more correctly: CPU>silverstone heat pipes>silverstone fin array/h60 heat plate. So unfortunately it's most probably bullshit.

Sure you'll get more performance because of the extra fin array from the silverstone cooler, but it won't be so much that it only gives 15c increase in temp even under intel burn test.

[DigitalHermit]

-snip-

-snip-

 

Remove the black tiny heatsink at the end of the copper pipes in the Silverstone, attach the H60 to the silverstone's naked copper pipes end.

It ends CPU > Thermal goo > silverstone copperpipes > Thermal goo > H60 cooling pipes. in theory, the heat remaining after the H60 is maxed out is discipated by the silverstone fins. 

 

Seems feasible... But won't there be some inefficiency with the thicker block?

 

It's that or he added a peltier pad between the CPU heat spreader and the copper pipes...

[faziten]

It seems...well...bullshit.

Just coz you added a watercooler to a metal heatsink does not mean that suddenly the temperature will drop so drastically. Don't you think someone (oh, I don't know, silverstone themselves) would have already come up with such a cooler if the gains were that insane?

 

also:

 

 Unfortunately that's not how thermodynamics works, the heat would go in parallel to the silverstone fins and the h60 cooler until it reaches the equilibrium of input heat=heat dissipated.

 

so more correctly: CPU>silverstone heat pipes>silverstone fin array/h60 heat plate. So unfortunately it's most probably bullshit.

Sure you'll get more performance because of the extra fin array from the silverstone cooler, but it won't be so much that it only gives 15c increase in temp even under intel burn test.

 

sincerely i don't know (and also i didnt mean the heat would prefer to go first into the H60 or the other), in theory would't the sum of discipation coeficients be the final result? Even that i couldn't find the numbers to directly compare.  Silverstone claims up to 65W TDP on this unit, Corsar maybe 95? giving a (feel free to call this bs) a 160W TDP ?

 

In this link, some dude asked noctua about their D14 TDP rating...

http://www.pcper.com/category/tags/noctua

Surprisingly the supposed noctua guy replied 130W was easy work for the D (no pun intended). So giving the previous approximation in theoretical therms... maybe it's not magic but can be close to first tier units... 

 

still, i dont know for sure.

[HexaGuy]

I think this is complete and utter made-up bullshit to make that guy seem like he is smarter than he really is. That, or he's trying to troll you.

 

This makes no sense; both in the sense of thermodyamics, and in simple, utter logic.

 

Think about it.

 

You are talking about 8C above room temperature in modern games, and 15C above room temperature for Intel Burn Test. Not even a H100i or a 360mm radiator setup could produce that, and you're talking about a not very thick 120mm radiator with some cheap heatsink attached to it? It would probably increase your temperatures due to thermal inefficiencies of mounting the H60 to the round copper pipes and mounting the CPU to the round heatpipes. 

 

If your friend could provide some proof, with some nice, reliable data, and maybe some pictures, I would then believe him. I think he's just trying to boast/brag/trying to impress you with his little made up thought.

 

Sorry mate.

 

 

EDIT: How would you even mount the CPU to the heatpipes and the H60 to the heatpipes? Zip-ties?

[Marvzl1357]

sincerely i don't know (and also i didnt mean the heat would prefer to go first into the H60 or the other), in theory would't the sum of discipation coeficients be the final result? Even that i couldn't find the numbers to directly compare.  Silverstone claims up to 65W TDP on this unit, Corsar maybe 95? giving a (feel free to call this bs) a 160W TDP ?

 

In this link, some dude asked noctua about their D14 TDP rating...

http://www.pcper.com/category/tags/noctua

Surprisingly the supposed noctua guy replied 130W was easy work for the D (no pun intended). So giving the previous approximation in theoretical therms... maybe it's not magic but can be close to first tier units... 

 

still, i dont know for sure.

well, yeah, the total possible power dissipation would be the sum of each cooler's tdp minus inefficiencies, so probably more like 140W combined(maybe).

...definitely not enough to net a 15c delta T on intel burn test. 

[MasterRaceFanBoy]

tbh i like the idea, i hope some1 will do this and test it.

[faziten]

I think this is complete and utter made-up bullshit to make that guy seem like he is smarter than he really is. That, or he's trying to troll you.

 

This makes no sense; both in the sense of thermodyamics, and in simple, utter logic.

 

Think about it.

 

You are talking about 8C above room temperature in modern games, and 15C above room temperature for Intel Burn Test. Not even a H100i or a 360mm radiator setup could produce that, and you're talking about a not very thick 120mm radiator with some cheap heatsink attached to it? It would probably increase your temperatures due to thermal inefficiencies of mounting the H60 to the round copper pipes and mounting the CPU to the round heatpipes. 

 

If your friend could provide some proof, with some nice, reliable data, and maybe some pictures, I would then believe him. I think he's just trying to boast/brag/trying to impress you with his little made up thought.

 

Sorry mate.

 

 

EDIT: How would you even mount the CPU to the heatpipes and the H60 to the heatpipes? Zip-ties?

I don't know this guy, like i said, i went to watch a vid and saw this crazy idea, wandering the amount of truth in that claim. We all trend to exagerate so i'm not expecting realistic numbers, but i would be fun if someone could test it out .

[KemoKa]

Seems feasible... But won't there be some inefficiency with the thicker block?

 

It's that or he added a peltier pad between the CPU heat spreader and the copper pipes...

Putting a peltier pad directly onto a CPU makes almost no sense. You'd have to shove so much power through the pad to cool it down that much and then you'd have to cool the other side...right?

[DigitalHermit]

Putting a peltier pad directly onto a CPU makes almost no sense. You'd have to shove so much power through the pad to cool it down that much and then you'd have to cool the other side...right?

 

I think so... Haven't used a peltier pad myself yet, but from principle it could work... It'll be inefficient though... (IIRC those USB fridges/can coolers use peltiers...)

 

This (http://en.wikipedia.org/wiki/Thermoelectric_cooling#Performance) says that it's dependent on the material of the thermoelectric...

 

I do like the idea of modding the heatsink and putting the waterblock on top of it...

[KemoKa]

I think so... Haven't used a peltier pad myself yet, but from principle it could work... It'll be inefficient though... (IIRC those USB fridges/can coolers use peltiers...)

 

This (http://en.wikipedia.org/wiki/Thermoelectric_cooling#Performance) says that it's dependent on the material of the thermoelectric...

 

I do like the idea of modding the heatsink and putting the waterblock on top of it...

ehh, I'd just prefer waiting till winter and sticking my radiator outside... throwing another 200W at a peltier pad makes no sense to me. sure, for a USB fridge to gradually cool a can of dr pepper is fine, but the amount of power needed to cool a 100+ watt CPU strictly using peltier pads when there are CPU coolers like IcePipes which cost nothing to run is just unjustifiable.