Noctua NF-S12-1200 or Nexus fans on Thermochill ?

noir · Post by **noir** » Mon Sep 03, 2007 2:34 pm

Hello everyone,

This is my first post, although I've been reading through the forums for the last few years

. I did a search about similar topics, but no result turned up for thermochill & noctua

I came to the conclusion that i want to use Thermochill triple radis for my next computer build, as they seem to be the optimum in both heat dissipation and silence.

Now only the question about Fans is bothering me. Nexus 12cm fans seem to work really well. But the Noctuas are rated for higher CFM than the Nexus at similar dB levels.

Now Noctuas are said to be bad with heatsink cooling, because they don't generate as much air pressure.

But the Thermochill radiator's fins are spaced very largely, so maybe the pressure drop is negigeable?

Could anyone please enlighten me on what Fan is the best for Thermochill PA radiator use, the Noctua or the Nexus?

Many thanx in advance

ntavlas · Post by **ntavlas** » Mon Sep 03, 2007 3:15 pm

I would use the noctuas as exaust fans, maybe on a skythe ninja but not an a radiator.

whiic · Post by **whiic** » Mon Sep 03, 2007 3:31 pm

Welcome to SPCR, noir-chan.

I hope you'll get your answers to build your ultimate cruching machine as quiet as possible. It was a OCed dual core system you had in mind, with two Corsair PSUs for sharing the load to keep the fans from ramping up, if I remember correcly. Was that the system you're building?

I assume you've read most articles under "Fans & Controls", including both first and second 120mm round-up? Those don't focus on how backpressure affects sound character and volume, though.

christopher3393 · Post by **christopher3393** » Mon Sep 03, 2007 5:00 pm

I'm probably the guy that was going to use 2 corsair PSUs in a Mountain Mods case. I scrapped that build and went smaller.

There are a number of variants that will lead to your choice of optimal fans, so I don't think there is a blanket answer to your question. I've seen people use the Noctua 1200s effectively with thermochill rads, which are expensive but really good for high performance. Much will depend on your config., so providing as much info as possible will be really helpful.

noir · Post by **noir** » Tue Sep 04, 2007 3:16 am

Ok, about my config... this might take a while explaining. It's a bit of a crazy plan.

The case i will be using is a LianLi V2000B i still have lying around here.

The core of the cooling system will be two thermaltake p120.3 radiators. I will use a shroud of 4cm (the thermaltake one) to "connect" the 2 radiators together, making an air tunnel between them, and i'll make that airtight. So 3 fans mounted on one radiator will suck the air through the first radiator through the wind tunnel and then through the second radiator.

The first radiator will be for the "cool" waterloop: one Laing with Plextop in a silencing box, then a reservoir, then a CPU cooler, a flowmeter, and back tothe radiator. So only the CPU will be connected to this loop (an overclocked e6750. i expect temperatures of the water to stay cool). Eventually i will replace that cpu with a penryn in January.

The second radiator will be the hot loop: 2 grahic cards (overclocked or not) probably HD2900XTs, then the reservoir, Laing Pump silenced, and flowmeter to radiator again.

I expect wattage of this system to be around 600w under load, so i plan to use 2 Corsair 620w psus: one to power the system and the second to power the graphic cards. the goal is to have both PSUs at 300W load, so that their fans don't ramp up. The Corsairs also seem very stable on 12V lines, so 2 hd2900xt should work out.

The question right now is which 12cm fans will pull enough airflow through the 2 radiators in series while operating in silent mode (7volt-ish).

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 6:05 am

WOW Sounds very cool and well thought out. I'm going to assume you've decided that you have adequate motherboard cooling even w/ the OC.

A high performance gaming rig that also runs cool and very quiet is a great project imo!

I'm wondering if you wouldn't be better off with Scythe S-flex.

noir · Post by **noir** » Tue Sep 04, 2007 6:15 am

Thank you very much , i appreciate it

Glad my ideas aren't total nonsense (i'm still new to watercooling).

About motherboard cooling... i don't know yet. I'm sure i will be getting an x38 Motherboard, because i just managed to get my hands on an e6750 managing 4ghz@500fsb at 1.48v. So i need a motherboard doing hefty FSB.

I planned on just using a Noctua 12cm as Casefan to suck air from the Heatpipes of the Motherboard. If that isn't sufficient, i will just have to add Mosfet/Southbridge/Northbridge water cooling... but i'd prefer not having tp do it.

All depends on how efficient the new X38 will be.

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 6:29 am

Nice OC on that 6750! Can I assume you'll be using a Laing D5? Yah, the X38s look like they have adequate heatsinks. But will probably be hot anyway w/ that OC. What memory and how much OC?

noir · Post by **noir** » Tue Sep 04, 2007 6:36 am

nothing special on RAM.
I will use CellShock DDR2-666

The fancy thing about them is that they use D9GMH Micron chips.. the best DDR2 in the industry. So i will run em at 1:1 (thus ddr2-1000) with tight timings

As for the pump, i was thinking about a Laing DDC with special top:

Two of them

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 6:46 am

I was scratching my head about pump choice until I saw the little note under the pic. Have you compared the specs for this top w/ the Petra's tech delrin top for this pump? http://www.petrastechshop.com/peddcudeddcp.html
You also might wanna email Alex at Petra's. I asked for advice on an H2O build and he wrote a whole page in response and invited further questions. He knows his stuff and is a regular in HardForum's Water Cooling section. Would also recommend Xtreme Systems forum for advice on a project of this scale.

noir · Post by **noir** » Tue Sep 04, 2007 7:00 am

thanks for the detailed links

I did think about getting a ddc with petra's top, but there were 2 reasons i decided not to go for them:

1Â° I live in Europe, so its a lot cheaper and easier for me to get local products

2Â° This little box:

It is made especially for the DDC with that plexi top, to silence it.

I want to make that little pump inaudible, the pressure and liters/minute don't interest me that much, because the DDC is a very potent pump, and my system isn't restrictive at all.

So that's my reason why i want this lil pump

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 7:13 am

That box looks great! Who makes that top and box? Man, that should be ultra quiet if it works well. Are you using the 9W or the 18W version of that pump?

noir · Post by **noir** » Tue Sep 04, 2007 7:27 am

Both are made by http://www.watercool.de/

It *seems* like its a 10W version, although I'm not sure as they don't mention it on their page. Would be enough for my small loop anyway

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 8:09 am

Looks like this box is only available in Europe.

An early look at X38: http://www.anandtech.com/cpuchipsets/sh ... spx?i=3077

noir · Post by **noir** » Tue Sep 04, 2007 8:33 am

yeah. i read that report as well (RSS feeds are a wonderful thing)

i fear that i will need to watercool the chipset, mosfets and northbridge as well... for my high FSBs. Unfortunately, because i always want to keep my setups as simple as possible.

But maybe it is illusionary to want 500 FSB without active cooling

Ah yes, the box for the laing is only available in Europe as watercool is the maker of the Box. They also do very nice CPU coolers which score among the best worldwide.

migueld · Post by **migueld** » Tue Sep 04, 2007 10:17 am

noir wrote:Both are made by http://www.watercool.de/

It *seems* like its a 10W version, although I'm not sure as they don't mention it on their page. Would be enough for my small loop anyway

I like that little box

But keep in mind that the DDC 10w is already silent with the plexi top

christopher3393 · Post by **christopher3393** » Tue Sep 04, 2007 10:38 am

ya, looks like the box is intended for the newer noisier revision.

noir · Post by **noir** » Tue Sep 04, 2007 3:23 pm

yeah, probably it is

but i still want to eliminate every potential source of noise. for now the flowmeters are one of my biggest problems as they're supposed to make quite audible clicking noises. i need them though for an emergency shutdown of my computer through a controller.

the goal, after all, is a PC worth of SPRC

I will surely post my building log about the computer on this forum... and on Xtremesys as well. the goal is a computer compatible with the goals of both forums (i love doing the impossible)

it seems like with the answers i got til now, Nexus fans seem more suitable for sucking air through 2 thermochill radiators. am i right with that assessment?

Marci · Post by **Marci** » Thu Sep 06, 2007 2:15 am

1) Search for "Radiator Stacking" as that's essentially what you're doing with your radiators.

Radiator stacking simply reduces the efficiency of every radiator after the first as you're using warmer air...

Because at the sorts of air pressures that most PC cooling fans can provide, coupled with the fin density of the heat exchangers being used, a single radiator will typically achieve between 90-98% thermal transfer efficiency for the amount of air-flow that a single fan is capable of pushing through the singular radiator.

If you double up the radiators, one after the other, the second radiator is receiving air that is within 10% of the temperature of the coolant already. Worse, the second radiator will in turn double the air-flow resistance, effectively cutting the air-flow through the radiators by around 30%. So the 10% of possible fluid-air heat transfer capacity that the first radiator didn't achieve that you're counting on the second radiator to scavenge, gets totally overwhelmed by the fact that there's now 30% less air going through to pick up the heat.

i.e. you'd see ~20% higher water temps as a result.

Now if your fan is SO powerful that the first radiator's thermal transfer efficiency is low enough that adding the air resistance of the second radiator is minor in comparison to the amount of heat transfer that the second radiator is capable of scavenging, then that's where it'd work.

In my experience, this isn't going to happen for PC sized 12cm radiators until you start using >200CFM singular fans, or alternately double up some 100cfm fans in a push-pull arrangement, and even then, it's going to be borderline better than the single radiator with the single fan.

If you're going to use two radiators, they need to both be pulling their own clean air source for best effect.

Source: http://forums.overclockers.com.au/showp ... stcount=12

Hmmm, I suppose I should've explained it better.

Air has a finite thermal capacity per volume. It happens to be around 1250J/mÂ³.K (round figure).

i.e. it takes around 1250 Joules to raise 1mÂ³ of air by 1Â°C.

So if the fan is moving 1.5mÂ³/min of air through the radiator, then the air-flow has an inherent thermal capacity of around 31.25W/C, or a C/W of 0.032. i.e. the radiator CANNOT perform any better than the inherent thermal capacity of the air. Even if the radiator was 100% efficient, the C/W will never be lower than 0.032C/W.

Now if the radiator is 90% efficient at this air-flow speed, this means that the radiator will have a C/W of 0.032/0.9 =~ 0.036. i.e. the equilibrium point will be such that the water warm up by 0.036Â°C per Watt of heat input (from CPU + GPU + pump + whatever else is being water-cooled).

If we now add on a second radiator, and this increases the air-flow resistance such that the air-flow through the two radiators is now 1.1mÂ³/min, then the inherent thermal capacity of the air is now 22.9W/C, or a C/W of around 0.044.

i.e. even if we achieved a 100% thermal transfer efficiency of heat from the fluid into the air using the two radiators, at best we will achieve a C/W of 0.044, or about 20% worse than the 0.036 of the single radiator.

Now as air-flow goes up, radiator's thermal transfer efficiency goes down. There is eventually a cross-over point where adding a second radiator in air-flow series to the first radiator does help, but for PC water-cooling radiators, that doesn't occur until you start stacking extremely noisy fans, or using obscenely noisy single fans (Delta FFB/TFB, etc).

Source: http://forums.overclockers.com.au/showp ... stcount=16 (later in same thread)

2) For the fans, see http://www.xtremesystems.org/forums/sho ... p?t=150261 - tis a long read... but will leave you with the answer you're looking for.

two thermaltake p120.3 radiators. I will use a shroud of 4cm (the thermaltake one)

3) THERMOCHILL, not Thermaltake. COMPLETELY different companies, and I'd rather not be associated with Thermaltake if it can be avoided.

christopher3393 · Post by **christopher3393** » Thu Sep 06, 2007 4:21 am

Thanks for the links Marci---very helpful.

noir · Post by **noir** » Sun Sep 09, 2007 4:17 am

Thank you very Much Marci... love the informations in those threads

1) i didn't know it was called radiator stacking. But the insight from these threads really say that using stacking isn't a good idea. So i will stay away from it.

2) Concerning the fans, it's unfortunate that Cathar hasn't tested the Nexus and the Noctua yet. Right now i feel like going for the Nexus, as it is kind of implied in the thread that the Noctuas really wouldn't be able to handle radiators.

3) Ouch... i'm stupid... how did i manage to put Thermaltake instead of ThermoChill. I'll be more careful in the future

So, i ended up with a new design, instead of stacking; the idea still is to fit 2 radiators in the lower compartment of my LianLi v2000, but with a different layout this time:

Both radiators should be able to get fresh air from the front intakes, back intakes, and from the floor too (as the V2200 has "holes" in its floor.

Would this design work along nicely?

noir · Post by **noir** » Mon Sep 10, 2007 1:35 am

sorry for doubleposting, but i thought i might be on to something.

Talked a bit with a mate yesterday about how much a radiator heats up the air that passes through. He said that it a function of CFM and Watt load.

So he gave me this formula:

Fixed air temp raise for given wattage is wattage * 60 seconds per minute / ( 1.293 kg per cubic meter * 1010 J per kg * 0.0283 (cubic meters per minute) per cfm * fan cfm) I think it is that now correctly..In other words 1,6 celsius cfm per watt. (That is 1,6 * wattage / cfm = raise in celsius.)

According to that formula, if i were to put a 200W load on the radiator, with 81 CFM (3x27, Nexus 12cm at 7V) the air delta would roughly be 4Â°C. Meaning that 25Â°C air ambient would become 29Â°C after the first rad. The Thermochill site states that it can handle a 200W load on those exact Nexus 7V fans with a 10Â°C air/coolant delta. (http://www.thermochill.com/PATesting/index.php#PA1203). So at full load, the temperature of the coolant would be around 35Â°c.

Now if we consider the second rad, in series, thus stacked, in the following layout:
Nexus fans (push) --> ThermoChill (1) --> Shroud --> ThermoChill (2)--> Nexus fans (pull)

ThermoChill (2) will have to eat 29Â°c hot air, and might be loaded with 300W. According to the same ThermoChill Site linked as above, the fans can manage that load at 12V at a 10Â°C air/coolant delta. So the coolant would be around 39Â°C. Now even if those fans were to run at less than 12V, they'd still could manage the load, albeit at a higher coolant/air delta.

Naturally all of this makes no sense if we were talking just about one watercool loop. But i think it could make sense in two loops.. the first being the "cold" loop for the processor, the second "hot" one for the graphic cards.

Also i know that the fans will never be able to push/pull through as much CFM as stated in the fan databases, nor that the radiator has perfect thermal transfer. That's why i chose to use very high loads in my calculation, namely the power that the CPU and GPUs will draw from the PSU:

CPU : e6750 [email protected] will actually eat 135W of power (according to http://www.extreme.outervision.com/psuc ... orlite.jsp. But i also want to keep enough reserves for overclocking a quadcore heavily. If i could find a jewel doing 4ghz on watercooling it'd ask for 245W at the same Vcore as the DualCore chip.
GPUs: I decided to rate a HD2900XT at around 150W in full load. Somehow difficult to evaluate, as there seems to be contradicting information on that. So that'd mean 300W for a Crossfire system.

I doubt that all power they drain will be transformed into heat. (Unfortunately i don't have stats that tell how much power is transformed into heat).

One problem with this kind of setup is that it could heat up a lot if the ambient runs high, let's say 35Â°c ambient in summer. Then the water temperature in the first rad would be 45Â°, and 49Â°c in the second. I don't know if that kind of water temperature could still manage to actually cool computer parts. I could ramp up all 6 Nexus fans to 12V, but then the setup wouldn't be really quiet anymore.

I understand that the optimal solution would be to allow each radiator to have its own access to fresh air. But unfortunately I'm a bit "limited" with my case. Concerning the layout i presented in the above post, i was told that such a setup would easily make a thermal circle, meaning that the hot air would be sucked directly into the the rads again... making proper cooling impossible.

So for now, i don't see any other possibility for me than stacking the rads; using only one rad wouldn't manage the heat load. This leaves me with a few questions:

- Would this setup work, considering it uses two watercooling loops, while fully knowing the second one will heat up more?
- Are my calculations concerning airflow heat-up actually correct?

- Is there any information about the actual amount of heat that chips dump related to the wattage they eat?

Thank you very much

Marci · Post by **Marci** » Mon Sep 10, 2007 2:57 am

All looks fairly accurate... it will work as you envision...

The power calculator already takes into account the thermal characteristics as best possibly (hence the TDP dropdown). I always base figures as you have done. Assume max heatoutput = max power consumption, that way you're always aiming slightly over... set to 100% TDP in the power calculator if you intend to use TAT to analyse performance. If you intend to use any other software (and NOT use TAT) then leave it set to 85% TDP. White papers over at Intel's site will give more detailed info...

noir · Post by **noir** » Mon Sep 10, 2007 4:31 am

Thank you very much!

I will proceed to build it this way then.

Once i have all the parts i will make a little building log in here and on Xtremesys. Hopefully itll work all out ass planned