CPU tower coolers, which fan config is best?

[turgsh01]

I'll be buying my 2nd Noctua NH-U12S (or bigger) in prep for the next Ryzen release. With that said, I thought of something interesting that I thought might make for an interesting video, unless the answer is obvious and someone can answer it for me.

 

Which config is best for CPU tower coolers like the one mentioned above; Push-Pull? Pull-Pull? Push-Push?

 

Since the air could come from the sides of the cooler instead of just a straight through setup, but the sides aren't overly thick for mass airflow, I imagine the answer might not be so easy. Then again, maybe it is since the middle of the cooler might not get cooled as well due to air flow never reaching the middle and only cooling the edges, but does that matter that much? etc...

 

Any thoughts on this?

[Cyracus]

any setup where the fans aren't blowing the same direction performance will be hindered.

[freeagent]

I prefer push/pull. I’ve used it with heatsinks and a clc with good results. And it looks badass too. Especially with thick fans.

[turgsh01]

5 hours ago, Cyracus said:

any setup where the fans aren't blowing the same direction performance will be hindered.

Is that your opinion or a fact? If it's a fact, I'd like some evidence. Otherwise, I can't just take your word for it.

[Jeppes]

29 minutes ago, turgsh01 said:

Is that your opinion or a fact? If it's a fact, I'd like some evidence. Otherwise, I can't just take your word for it.

If you are claiming things are dropping upwards, its better that you provide the evidence. You could maybe see a weird result with a double tower where with very low fan speeds you get a tiny bit better temps by pushing air towards empty mid fan slot from sides. But even then you better have an exhaust fan above it if you test inside a case.

 

[57chevy]

7 hours ago, turgsh01 said:

I'll be buying my 2nd Noctua NH-U12S (or bigger) in prep for the next Ryzen release. With that said, I thought of something interesting that I thought might make for an interesting video, unless the answer is obvious and someone can answer it for me.

 

Which config is best for CPU tower coolers like the one mentioned above; Push-Pull? Pull-Pull? Push-Push?

 

Since the air could come from the sides of the cooler instead of just a straight through setup, but the sides aren't overly thick for mass airflow, I imagine the answer might not be so easy. Then again, maybe it is since the middle of the cooler might not get cooled as well due to air flow never reaching the middle and only cooling the edges, but does that matter that much? etc...

 

Any thoughts on this?

 

 

[turgsh01]

3 hours ago, 57chevy said:

 

 

That has nothing at all to do with this topic.

 

4 hours ago, Jeppes said:

If you are claiming things are dropping upwards, its better that you provide the evidence.

What claims? Evidence of what? I'm just asking questions looking for answers.

[57chevy]

1 hour ago, turgsh01 said:

That has nothing at all to do with this topic.

Dude you were curious about fan configs on a tower cooler.  Put a tower cooler in there instead of a radiator and you get nearly the same result.  The main reason it's always a push, pull, or push/pull and never push/push or pull/pull is so the air flows in a certain direction. Usually front to back because most cases have a spot for an exhaust fan.  Cool air comes in the front, blows through the cooler and gets sucked out the back. Putting two fans in a push/push config on your tower just blows the heated air into itself, where it will then bleed out the sides. Now that heated air has to flow around the case hoping to get pushed to the rear and then sucked out by an exhaust fan. In the meantime it could get sucked back in by the fan to heat up even more and then bleed out and repeat the process. If you have a GPU, that warm air bleeding out one of the sides is aimed most likely straight at your GPU. 

 

With pull/pull you're sucking cool air through the middle of the cooler and then blowing it out the ends.  Unless you have exhaust fans on the front of your case, you're blowing hot air straight into a front intake fan which will just blow that heated air back into the case, to be sucked through the cooler again. If you have a GPU, it will have no access to cooler air cause the intake is just blowing warm air around. 

 

Air is like water.  It flows best where there's the least amount of resistance. On a pull only config, you'll get more air being sucked in through the sides of the tower due to there being resistance from the tower itself.  But the sides will have less resistance due to there being less tower in the way.

 

Push config will push more air out through the sides because it has to push air through the tower which is creating a light obstruction.  So more air will push out the sides due to there being less of an obstruction.

 

And push/pull will have a bit of both but most of the air will go straight through due to the push/pull creating a wind tunnel effect.  All versions work with moving air and dissipating heat.  And one might not be better than the other because eventually you hit a ceiling. Your internal temps are limited by the ambient air the case is taking in. If you've got powerful intake and exhaust fans, then you only need a push or a pull since the intake/exhaust create the airflow direction for you. If you only have a single 120mm intake and 120mm exhaust, then do a push/pull so the air if forced in and out of the tower quicker. 

[LogicalDrm]

The keyword is "airflow". Think about that for a moment. What that word means?

 

Now think, is push-push or pull-pull something that would mean same as "airflow"?

 

Airflow means that air is moving to constant direction. Where is air moving with two fans both pushing towards heatsink? Where the air is coming from if you have two fans pulling it from heatsink. If you would have just heatsink on testbench, it probably wouldn't matter what config you use on it. But inside the case, where to air is coming from and where it goes is very important. Which is why the question, where the air is coming from if you have both fans pulling has no answer. It doesn't come anywhere. You have two case fan slots near. You can really utilize one of them. Same with push config, though there having rear fan as intake and top as exhaust could work. You'd still have to deal with air coming out GPU side.

 

This is somewhat issue with GPU cooling already where fans push air towards PCB, but there's no direct route for air to vent out. Some goes out of back, some goes towards bottom, some towards side panel, some towards front fans. Considering GPU fans point down, you can't effectively use from as exhaust. Except if you have good bottom intake, which is becoming fantasy with current case designs.

 

So this actual question might not have been answered with heatsinks (which are different from radiators as the find on sides are open). In theory we could apply same results as from case airflow tests. What happens with all fans intaking, all fans exhausting vs control. Talking about airflow, thats closest test I know.

 

If you would ask, between push, pull or push-pull, the two first make no difference. Thats something I've tested myself. Same goes with coolers direction between horizontal and vertical. No difference. Overall case airflow makes bigger difference than just how single fan cooler is set. Push-pull does make a difference, but you have two fans making noise, so its trade-off between performance and noise.

[Cyracus]

10 hours ago, turgsh01 said:

Is that your opinion or a fact? If it's a fact, I'd like some evidence. Otherwise, I can't just take your word for it.

Sorry, realized I unintentionally misspoke, any configuration where the fans are opposing each other, performance will be hindered (because a front push top pull or similar atypical fan setup should work fine, possibly better in some cases)
As for evidence, I offer none. Feel free to test for yourself.

[turgsh01]

On 9/6/2020 at 8:42 AM, 57chevy said:

snip

You make way too many assumptions about case setup and assume that a radiator is very similar to a tower cooler despite the obvious differences if you just look at them. I will not be reading most of your post as I can do without the headache.

 

On 9/6/2020 at 11:14 AM, LogicalDrm said:

snip

You too have made assumptions instead of providing evidence, but at least some of what you said sounded logical. Also, I have a CM Storm Sniper BE case. Case airflow, including bottom intake, is not a problem at all. Inside the case stays at ambient even under heavy load (with of course the exception of the parts that typically get hot when being used). I've never seen my heavily overclocked GTX 1060 or my mildly overclocked Ryzen 1600x reach beyond 65C during benchmarks while on air coolers. Though I imagine the RTX 3080 and Ryzen 4600 or 4800 (haven't decided yet) might change all that for me. Also, I could theoretically have all my fans exhaust or intake as it would likely not change much for me since most of the case is filtered mesh, but that's another test for another time.

 

On 9/6/2020 at 11:51 AM, Cyracus said:

Sorry, realized I unintentionally misspoke, any configuration where the fans are opposing each other, performance will be hindered (because a front push top pull or similar atypical fan setup should work fine, possibly better in some cases)
As for evidence, I offer none. Feel free to test for yourself.

I understood your point of view / opinions, I just can't take a complete stranger's word for it based on no evidence at all provided. If you were Linus or Jay or Steve, then I would take your claims more seriously as they have the reputation and experience to back them up. But I don't know a thing about you and you've given me no reason at all to take your word for it.

 

I thought posting in a dedicated tech forum would maybe provide me with answers to tech related things. As a certified PC hardware technician of around 20 years, even I still don't know all there is to know about tech cuz it keeps changing daily, hence why I'm willing to ask questions. But clearly I was wrong to come here for information. It's just filled with a fan base of common knowledge that prefers to make assumptions instead of looking deeper into the unknowns. Looks like I'll be doing my own tests when I get the new parts in the mail and find the time to do it.

[LogicalDrm]

20 minutes ago, turgsh01 said:

You too have made assumptions instead of providing evidence, but at least some of what you said sounded logical.

 

Any assumption I make has facts behind. Like basic fluid dynamics for example. I haven't done testing like having two fans on heatsink blowing any other way apart from following my cases overall airflow. I have done tests between push vs pull and orientation of the cooler.

 

So the first two are theorizing based on what I know about air cooling and physics. The two last paragraphs are noting facts and what I've done and read/watched.

 

20 minutes ago, turgsh01 said:

 Looks like I'll be doing my own tests when I get the new parts in the mail and find the time to do it.

 

Please do. Always better to test something yourself if there's no tests made, or what is made has conflicting info. I actually could do this also. I have time, heatsink and fans to do it. Its also quite quick to test out. 30min CPU stress test with all setup, removing my CPU cooler fan settings (aka setting them to auto while testing). Only thing I need is to find my Y-splitter.

[Quadriplegic]

1 hour ago, LogicalDrm said:

Please do. Always better to test something yourself if there's no tests made, or what is made has conflicting info. 

Or when you dismiss what everyone says and act stubborn. Totally not aimed at OP

[turgsh01]

1 hour ago, Quadriplegic said:

Or when you dismiss what everyone and act stubborn. Totally not aimed at OP

Would you believe me if I told you there will be an RTX 3090ti released next year without providing any evidence what-so-ever, or would you be skeptical and assume I might be creating fake information until proven otherwise? I'm not being stubborn, I'm being intelligent in not believing every random thing I read on the internet without first seeing some evidence.

 

 

Here's what I see in the fan tower dilemma: With a normal push pull config, you only get the airflow quantity of a single fan with a bit less airflow restriction due to the added fan helping it along. You would have a small dead zone of no airflow in the absolute center of the cooler due to fan design. With a push push config, the airflow would of course collide with each other and be redirected out the sides. With modern fan designs (especially Noctua's fan designs), static pressure is far better than it used to be, so the air should in theory be less impacted by flow redirects than it used to be. Of course the dead zone of no airflow in the center of the cooler would be much greater, though by how much depends on the cooler fin design (and maybe cutting or drilling some of the center out completely could help to minimize that dead zone). The heat pipes on a tower cooler always seems to be on the outer edges where the airflow is most active for both configs as well. Whether or not the total airflow of a push push config would be more than the equivalent of 1 fan or more is up for debate and testing. It's also up for testing on whether or not the dead zone of no airflow in the center of a non-modified cooler would be large enough to make cooling capacity lower than a straight push pull config. And yes, case fans might need to be adjusted to help direct the flow better, but that's usually no big deal, besides, I'm more interested in best case scenario with open test bench which is what I have for my media / game hosting PC.

[LogicalDrm]

On 9/9/2020 at 5:28 PM, turgsh01 said:

-snip-

You are making assumptions with that too. Like that air would somehow not move between fan centers. When blade design is there to eliminate any possible dead zones right next to fan. Another point is that how far from fan air is effected. This you can actually see from smoke tests (while I don't like them for any other testing purposes). Considering how slim heatsinks are, the dead zones in the middle aren't that important that you would notice it, unless you purposely select fan which has short range on output side.

 

Another thing to think is what heatsink/radiator is actually built to do. Fins are there to move heat from heatpipes to larger surface area which can then be used as dissipation surface. As temperature is trying to reach equilibrium between two surfaces, air inside the case heating and fins cooling trying to match. Purpose of fan is to provide cooler air than what it would be without fans. For this reason even passive cooling works, when dissipation area is large enough.

 

**** THEORETICAL ***

If we assume that there's dead zone as cylinder/cone where fans motor/center is, having push-pull, push-push, pull-pull wouldn't make any difference. The dead zone would still be there, unless fans would have offset to counter it. My logic on why fan config wouldn't matter is that air only moved by fans inside the case. If air is pushed by fans that don't direct it towards center of the fan by default (= causing dead zone in the middle), then why would two colliding airflows that move straight forwards or away from center suddenly change to move towards center? Wouldn't the path of least resistance be where rest of the airflow is already going? Same logic applies with pull-pull. Kinda as the dead zone might be larger as air is only moved from outside of the heatsink. With radiator the problem might not exist since the vacuum factor might be a thing.

***

 

I don't have anything planned for weekend, so I could test this out.