One Radiator or Two?

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
TiredOldMan
Posts: 22
Joined: Wed Nov 02, 2005 1:52 am

One Radiator or Two?

Post by TiredOldMan » Sun Nov 13, 2005 7:44 pm

This is my first post, so if I sound stupid, go easy on me. :)

I have an SLI gaming system based around the Asus A8N-SLI Deluxe and two 6600gts. It’s housed in a CM Stacker (one of the original ones that were for some reason designed to take power supplies at both the top and bottom.) I am tired of all the whirring and whining, and so am looking to install a WC system, with an eye to quiet performance.

I would love to install a 12x24 radiator, but there is just not room, unless I cut the top of the box--something I want to do. I was thinking about cutting the back, but that would involve cutting through a structural support. Again, something I don't want to do.

So, what I do have space for is, one 12x12 radiator, and one 8x16 radiator. I was looking at a pair of Sharkoon Eagle 1000s for the 8x16 (though all I’ve read so far is company hype).

Does anyone have any experience with installing more than one radiator? Reading Chylld’s fantastic manual above, it seems that the fewer parts I’ve got, the better. Would I be able to cool an SLI system with just the one 12x12 radiator?

Thanks for anything you have to offer!

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Fri Nov 18, 2005 10:57 pm

I doubt a 12x12cm rad can handle your CPU & 2 GPUs with near-silent fan(s). But WTH do you have in that box? A Stacker has a lot of room to play, and I'd expect you could manage at least a double-120 rad.

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Sat Nov 19, 2005 7:41 am

HammerSandwich wrote:I doubt a 12x12cm rad can handle your CPU & 2 GPUs with near-silent fan(s). But WTH do you have in that box? A Stacker has a lot of room to play, and I'd expect you could manage at least a double-120 rad.
And for what I understand, not every rad needs to have case side space. You could for example to a setup like this with 120mm fans and single-120-rads:

[fan] [rad] [fan] [rad] [fan]

With proper shrouding I'd expect this to be both effective and easier to fit.

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Sat Nov 19, 2005 8:16 am

zds wrote:[fan] [rad] [fan] [rad] [fan]
Not a good idea! Just see what Cathar wrote.

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Sat Nov 19, 2005 8:43 am

HammerSandwich wrote:
zds wrote:[fan] [rad] [fan] [rad] [fan]
Not a good idea! Just see what Cathar wrote.
The crucial difference here is that I proposed *three* fans. Of course it would be a bad solution if you tried to do it with just one fan. In what I proposed there would be three fans blowing air through the tunnel, so while there is double the restriction, there is three times the blowing force.

I need some hard data, not just assumptations, to believe it wouldn't work :-). The warmening of the air is of course a problem, but I do not think it severely affects the end result - after all, many people are actually blowing air through the radiator to the inside of the case, and still it works.

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Sat Nov 19, 2005 8:47 am

zds wrote:I need some hard data, not just assumptations, to believe it wouldn't work :-). The warmening of the air is of course a problem, but I do not think it severely affects the end result - after all, many people are actually blowing air through the radiator to the inside of the case, and still it works.
Hmmm, this started intriguing me... Luckily I have spare heater core in my possession, now I only need to get/loan a second one and set up a test system :-)

Of course if you have enough space, go for double-120 rad and four 120mm fans, but I still think the sandwich solution would be a lot better than a single 120 rad with two fans.

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Sun Nov 20, 2005 12:27 pm

zds wrote:I need some hard data, not just assumptations(sic), to believe it wouldn't work :-).
BillA's Thermochil review has a lot of good info in it, including C/W and airflow charts. Get fan P/Q curves, and you can generate a pretty decent estimate.
zds wrote:Of course if you have enough space, go for double-120 rad and four 120mm fans, but I still think the sandwich solution would be a lot better than a single 120 rad with two fans.
But this is Silent PC Review, and we're allergic to fannage. Because radiators are pretty darned efficient with low airflow,

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Sun Nov 20, 2005 12:43 pm

zds wrote:I need some hard data, not just assumptations(sic), to believe it wouldn't work :-).
BillA's Thermochil review has a lot of good info in it, including C/W and airflow charts. Get fan P/Q curves, and you can generate a pretty decent estimate.
zds wrote:Of course if you have enough space, go for double-120 rad and four 120mm fans, but I still think the sandwich solution would be a lot better than a single 120 rad with two fans.
But this is Silent PC Review, and we're allergic to fannage. Radiators are pretty darned efficient with low airflow, so the air coming out is nearly as warm as the water. Feeding that air into a 2nd rad won't offer much cooling. The greater airflow from three fans in series might address this, but what's the noise penalty? Not only will 3 fans be ~5dBA louder than 1, but air turbulence through the small frontal area will become a problem as well.

Adding a second rad in airflow series is a lot like doubling a rad's thickness. Look at available radiators. Air flows across the smallest dimension for a good reason.

TiredOldMan
Posts: 22
Joined: Wed Nov 02, 2005 1:52 am

Post by TiredOldMan » Sun Nov 20, 2005 8:40 pm

Thanks so much, guys. I've got the original Stacker, which means that it is impossible to put the 2x12 up at the back of the case. The only other options are to either run 1 12cm rad and one 2x8 rad, or to run a 2x12 out of the front of the case, warming my knees in winter and giving me extra sweat where I don't need it in summer. :shock:

BUT, since posting this, some land we were looking at suddenly came on the market, and so for another year, I'm not going to have the time or money to even consider this project. Damn. :x

Thanks again, (and see you next year?)

TiredOldMan

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Fri Nov 25, 2005 6:15 pm

HammerSandwich wrote:But this is Silent PC Review, and we're allergic to fannage.
Not only will 3 fans be ~5dBA louder than 1, but air turbulence through the small frontal area will become a problem as well.
Yes, but as we know, if there's more than minimal amount of heat to be transfered, less fans is not always better. It depends on how you set up fans, overall case airflow, voltage etc. And it also makes a big difference are the fans inside the case or in the edge of it.

I do not claim adding fans to your system is always better, but it's a sum of too many factors to be declared to be the other way either.
Last edited by zds on Wed Nov 30, 2005 5:22 am, edited 1 time in total.

Marci
Posts: 68
Joined: Thu Oct 27, 2005 5:49 am
Location: West Yorkshire, UK
Contact:

Post by Marci » Wed Nov 30, 2005 4:49 am

Mount PA160 radiator in case with a single Nexus 120mm fan. Will perform ever so slightly (1 or 2 degrees) behind a 2x120 rad with 2x120mm nexus fans installed, and will radically outperform 1x120 and 2x80 combination with much less noise due to simple equation of 3 quiet fans = louder than 1 quiet fan.

TiredOldMan
Posts: 22
Joined: Wed Nov 02, 2005 1:52 am

Post by TiredOldMan » Thu Dec 01, 2005 3:34 am

And there's the answer.

You're the Marci that I've seen (I think) on ProCooling and other expert forums, aren't you. It's good to have your input.

So, once this house-moving thing is all squared away, I'll probably run a duct along the bottom of the case, with my 2 drives at the front and the PA160/Nexus setup you describe in the middle.Then, I'll put a quiet fan in my PSU and let it take care of the rest of the heat dissipation.

Thanks!

Marci
Posts: 68
Joined: Thu Oct 27, 2005 5:49 am
Location: West Yorkshire, UK
Contact:

Post by Marci » Thu Dec 01, 2005 6:05 am

You're the Marci that I've seen (I think) on ProCooling and other expert forums, aren't you. It's good to have your input.
Aye that be me... aka Technical Manager of ThermoChill & Over-Clock UK Ltd.

Also, recommend investigate mounting PA160 flat on floor of case... minor dremel work cutting a rectangle out - use the PA160 wings kit to fix it there... Shroud on inside protruding into case... fan on shroud... then a simple 120mm flexible duct to the air inlet at front of case, so air is sucked in front of case, ducted to the rad fan, and vented out thru the rad and thru the base of the case.

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Thu Dec 01, 2005 6:42 pm

Just don't go 1 x 120 rad to cool all that quietly.

TiredOldMan
Posts: 22
Joined: Wed Nov 02, 2005 1:52 am

Post by TiredOldMan » Thu Dec 01, 2005 8:33 pm

Marci: That's not a bad idea at all. The Stacker already has an oblong vent in the bottom that I currently have blocked off to control airflow direction. (I was amazed when blocking it lowered the sum of my system's parts by about two degrees. Yes, I AM new at this...) It would make for a dead-easy modification.

But I am wondering the same as Snutten: with 2 gpus, an amd cpu and a chipset in the loop, would one PA160 be enough?

Marci
Posts: 68
Joined: Thu Oct 27, 2005 5:49 am
Location: West Yorkshire, UK
Contact:

Post by Marci » Fri Dec 02, 2005 3:26 am

It would do the job with best performance at lowest possible noise levels.
Bigger rads will do it better, but use more fans, therefore more noise.

If PA160 can cool my TEC-setup (half a kilowatt of heat almost - which it can) with a single 7v'd Panaflo M1A, then a non-TEC'd setup would be a breeze...

http://www.xtremesystems.org/forums/sho ... hp?t=77260

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Fri Dec 02, 2005 10:30 am

As usual, we end up in a debate what can be considered quiet :-)
I'm not pretending to set the standards here and I certainly wouldn't argue that Marci's got it wrong or anything.

But, in my book, a 7V panaflo M1A is not even close to quiet, especially not when mounted on a rad.
And remember: everything alse equal, two undervolted fans, pushing the same amount of air as one (less undervolted), makes less noise. It is almost always better to have more fans at lower rpm than fewer with higher rpm. Not to mention that in this case there's also the benefit of doubling the radiator area.

Why are you going to go w/c? For cooling performance alone or for silent operation? Nowadays you can build a really really quiet computer with air cooling and lose the hazzle that comes with w/c. At least waaay quieter than any 7V fan on a rad.
Again, no offense to Marci that probably also has other goals than quiet operation, or just a higher noise tolerance.

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Fri Dec 02, 2005 10:39 am

snutten wrote:And remember: everything alse equal, two undervolted fans, pushing the same amount of air as one (less undervolted), makes less noise. It is almost always better to have more fans at lower rpm than fewer with higher rpm. Not to mention that in this case there's also the benefit of doubling the radiator area.
My thoughts exactly. I also find it hard to believe that given equal single fan doubling the radiator area would _worsen_ the cooling. If nothing else, the actual airflow would increase, since backpressure would decrease almost by half.

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Fri Dec 02, 2005 4:41 pm

zds wrote:My thoughts exactly. I also find it hard to believe that given equal single fan doubling the radiator area would _worsen_ the cooling. If nothing else, the actual airflow would increase, since backpressure would decrease almost by half.
Ummm... depends what radiator you're using. No two radiator designs will have the same fin design, and so doubling the radiator area may still leave you with worse airflow for the same pressure drop if you don't keep everything else identical. The PA.160 was designed with exactly this in mind.
Oh, and noise .vs. aircooling depends a lot on how you do it. I'm using AC fans in mine and what little data I've got indicates they're probably quieter than Nexus fans for more airflow. Unfortunately I don't have any of the reference fans to compare it against.

zds
Posts: 261
Joined: Tue Aug 03, 2004 1:26 pm
Location: Helsinki, Finland
Contact:

Post by zds » Fri Dec 02, 2005 4:53 pm

pdf27 wrote: Ummm... depends what radiator you're using. No two radiator designs will have the same fin design, and so doubling the radiator area may still leave you with worse airflow for the same pressure drop if you don't keep everything else identical. The PA.160 was designed with exactly this in mind.
Oh, sorry. I didn't realize you referred just some exact products. I meant just a case where nothing else changes but surface area of the rad.

Certainly different airflow impedance, surface area between water and metal and surface area between metal and air all count in the equation.

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Sat Dec 03, 2005 7:46 am

zds wrote:Oh, sorry. I didn't realize you referred just some exact products. I meant just a case where nothing else changes but surface area of the rad.
Fair enough. I'm just a test/development engineer at work so stuff like that kind of leaps out a bit at me ;)

Marci
Posts: 68
Joined: Thu Oct 27, 2005 5:49 am
Location: West Yorkshire, UK
Contact:

Post by Marci » Mon Dec 05, 2005 3:29 am

Again, no offense to Marci that probably also has other goals than quiet operation, or just a higher noise tolerance.
Never said the 7v'd M1A was quiet. Merely illustrating performance between two rads.

If PA160 with 1x 7v'd M1A performs just behind a PA120.2 with 2x 7v'd M1As, then the same will apply no matter what the fan... as long as for the comparison, both rads use the SAME fan.

ie: PA160 with 1x Nexus at 7v will perform just behind PA120.2 with 2x Nexus at 7v.
PA160 with 1x SilenX @ 12v will perform just behind PA120.2 with 2x SilenX @ 12v

So, in the scenario above, where you have 1x fan at 7v vs 2x of the same fan at 7v... the singular is obviously quieter.

TiredOldMan
Posts: 22
Joined: Wed Nov 02, 2005 1:52 am

Post by TiredOldMan » Mon Dec 05, 2005 8:46 pm

snutten wrote:As usual, we end up in a debate what can be considered quiet :-) Why are you going to go w/c? For cooling performance alone or for silent operation? Nowadays you can build a really really quiet computer with air cooling and lose the hazzle that comes with w/c. At least waaay quieter than any 7V fan on a rad.
Again, no offense to Marci that probably also has other goals than quiet operation, or just a higher noise tolerance.
Why? Well, I do play a lot of games, and want to get a quiet and safe overclock when I'm playing them, and something approaching pure silence when I'm not.

But really, that's just the justification. The real reason is this: it just sounds like fun. :lol:

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Fri Dec 09, 2005 7:43 am

Marci wrote:If PA160 with 1x 7v'd M1A performs just behind a PA120.2 with 2x 7v'd M1As, then the same will apply no matter what the fan... as long as for the comparison, both rads use the SAME fan.

ie: PA160 with 1x Nexus at 7v will perform just behind PA120.2 with 2x Nexus at 7v.
PA160 with 1x SilenX @ 12v will perform just behind PA120.2 with 2x SilenX @ 12v
Different radiators benifit variably from higher air and or water flow. For example, I believe a rad with tightly set fins often benefit more from a powerful fan than one with wide spaces between its fins. Backpressure varies and narrower fins heats up the air between them faster.

This also explain why higher water flow is likely to result in a very small radiator performance gain if the corresponding fan is spinning very slow. That's why we don't need high water flow in our home radiators, even though the water passes many of them before it's re-heated. Because they are massive with wide spaces and only needs to cope with a very slow air flow passing through them.
Marci wrote:So, in the scenario above, where you have 1x fan at 7v vs 2x of the same fan at 7v... the singular is obviously quieter.
But hopefully he can run 2 x Nexus @ 5V, even down towards 3V when idling, if he adds an active fan controller. I have two rads with 4412s @ 3.3V on a rather power-hungry AMD system. Quite impossible with just one 120 rad I think, but maybe I'm wrong. Nota bene: I know nothing about these PA120/160 rad. But I'd imagine two rads always cool better than one, of the same type.

The fans and the resulting air whoosh make the noise here. E.g: A fan pushing 40 cfm emitting 30dB. Then two of those pushes 80cfm (less if back the case is restricted and back pressure builds up, but that can be avoided) and noise is up to 33dB. The single fan wouldn't push even close to 80, probably not even 60 cfm, at 33dB. The same goes for quiet fans on rads.

Double airflow, double the cooling area, small increase in noise. Am I missing something here?

Marci
Posts: 68
Joined: Thu Oct 27, 2005 5:49 am
Location: West Yorkshire, UK
Contact:

Post by Marci » Fri Dec 09, 2005 8:01 am

Not quite, you're still missing the point - and need to read up on the PA160. The PA160 performs to within 2 or 3 degrees on average of the PA120.2 with same fans at the same voltages.

PA160 with 1x 120mm nexus at 5v or 3v will still only be 2 or 3 degrees behind PA120.2 with 2x 120mm Nexus at 5v or 3v. Therefore for what equates to 2 or 3 deg, if noise is the important factor, 1x fan at 3v on a pa160 is still quieter than 2x fans at 3v on a double rad.

See www.thermochill.com/pa160.php, specifically links to development discussion and design phases...
Different radiators benifit variably from higher air and or water flow. For example, I believe a rad with tightly set fins often benefit more from a powerful fan than one with wide spaces between its fins. Backpressure varies and narrower fins heats up the air between them faster.

This also explain why higher water flow is likely to result in a very small radiator performance gain if the corresponding fan is spinning very slow. That's why we don't need high water flow in our home radiators, even though the water passes many of them before it's re-heated. Because they are massive with wide spaces and only needs to cope with a very slow air flow passing through them.
I choose the models I refer to carefully... yes, denser fins = more cfm required. Both PA120 and PA160 amongst the lowest FPI rads available today. The PA Series ThermoChill rads were designed by Cathar and produced by yours truly specifically with silent fans in mind, so are designed from the outset for performance in silence. FPI and tube sizes have all been taken into consideration...


The fans and the resulting air whoosh make the noise here. E.g: A fan pushing 40 cfm emitting 30dB. Then two of those pushes 80cfm (less if back the case is restricted and back pressure builds up, but that can be avoided) and noise is up to 33dB. The single fan wouldn't push even close to 80, probably not even 60 cfm, at 33dB. The same goes for quiet fans on rads.
If the radiator is efficient enough, then depending on heatload, 60cfm may not be required... all ultimately depends what kind of noise level is desirable vs what device output temps are desirable.
Double airflow, double the cooling area, small increase in noise. Am I missing something here?
Yes... a PA120.2 doesn't equal double a PA160. If you double a PA160, you end up with a 3x120mm rad (PA120.3 / HE120.3)

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Fri Dec 09, 2005 5:20 pm

Thanks Marci for your patience in explaining. This is what had me going (you wrote about the 1 x PA160 setup):
Marci wrote:It would do the job with best performance at lowest possible noise levels.
Bigger rads will do it better, but use more fans, therefore more noise.
The way I see this we are comparing apples and oranges. If the PA160 is just simply bigger then of course it performs better. So let me repeat and clarify my central point:
It is ALWAYS a good idea to maximize cooling area.
And unless the fans in use can't be heard it is almost ALWAYS better to use more fans spinning slower instead of fewer spinning faster. Not so if the fans motor noise, instead of the air whoosh, is what you hear. Also, of course, not in a serial setup in a non-restricted case, because air-flow would hardly increase.

This means that bigger rads (of same design, like two of the same compared to one) NEVER produce more noise at the same cooling effect. If the rads are somewhat different in size, then there's a weird negating example I can think of: when in between passive and active - E.g. one big radiator needing only a quiet-as-can-be fan is better than two smaller sets where you can't do with only one as active. If the big rad, with one fan @ 90 rpm can do the job, mostly having some slight motor noise left, then adding a second to be able to go to say 60 rpm makes no sense. It would be better to opt for partly or fully passive, or try to make do with the single fan. Now, the Nexus fans are almost void of motor noise at those speeds anyway so with the right equipment this situation shouldn't ever be an issue.

Everything else equal, a bigger cooling area requires less airflow = less noise, not more. If not, you've done something wrong, or the laws of physics must once again be re-written.


So, TiredOldMan, here's my take on the subject. Go 2 x PA160, if you can fit them. In fact, go as big as you can, go huge and get Nexus fans spinning under 200 rpm. That can actually be done with a T-balancer + Attenuators. Chose rads with wide spaces between the fins. The PA series seems to be excellent. Good luck!

Post Reply