Node 304: tower vs low-profile cooler (impact on airflow?)

[piokos]

I want to replace a CPU cooler (currently Intel stock) in my Node 304.
Originally I was thinking about a tower Noctua (either D9L or U12S).
But I had a lazy Friday evening and started thinking about it. And thinking always leads to problems, but just occasionally to solutions... ;-).

The case is basically an air tunnel, with big intake/outtake fans at both ends (2x92mm and 140mm). These fans are independent of what's going on in the case.
If we put a medium-sized tower cooler with PWM fan, e.g. Noctua U12S., it takes roughly half of the case cross-section left for the airflow.

Here's the thing:
1) at low CPU load (cooler airflow < case airflow) the cooler will effectively become an obstacle - forcing air to move around it. Not perfect, but acceptable.
2) at high CPU load (cooler airflow > case airflow) the cooler may try to suck more air than case airflow provides...
So theoretically it'll either be inefficient (limited), force more airflow through intake and gaps (turbulence/noise/dust) or suck from behind the cooler (warm air and turbulence/noise as well).
In reality it should be a stable state combining everything mentioned.

So now I'm curious whether getting a low-profile cooler, pushing air downwards, is not a better idea (e.g. Noctua L12S).
(2) will still happen, but in a more organized way (less turbulence)
(1) will be avoided, since the cooler will suck from a clean, almost laminar airflow above it

What do you think?

 

[tarmc]

curious which cpu youre cooling? overclocked?

 

[mindless1]

You're overthinking it. A tower sink oriented so its fan is blowing the same direciton as the chassis airflow is prefect for that case and you might not even need a fan on it if the heatsink rating is for around double the CPU rating. If the tower heatsink fan were blowing opposite the chassis airflow it would be a bad choice, but practically any tower sink using a *standard* fan would allow simply flipping the fan over to blow the other direction.

1) At low CPU load, if the heatsink fan is powered (spinning at all), it is increasing the velocity of air through the 'sink, not hindering the flow the chassis fans provide. At any given power a brushless computer fan will increase (or decrease) in RPM based on the static pressure environment it's in.

2) Not a problem as long as heatsink capacity is enough to keep the CPU under its max temp spec. Also not a problem for other case components if the chassis fans move a sufficient amount of air for the total case heat production. They are somewhat independent of each other because each component that needs heat removed should have it's own mechanism (heatsink, and fan on it if needed) to do so within a fair temperature range and amount of airflow. With the size fans you have this should be no problem except in a very hot room, ambient temperature.

The Noctua L12S is quite expensive for what it is and I would only use one in a case requiring that design for clearance reasons. It's practically always best to have the heatsink fan blowing the same/similar direction the chassis airflow is already moving, unless it's a setup where you are at the edge of the motherboard design power and need more flow on a chipset 'sink or mosfets in the VRM next to the CPU in which case a downward blowing fan will put more airflow on them.

This was more of a problem back before very low RDS 'fets hit the market or with extreme overclocking, or trying to use a cheap budget board (with only 4 pin PSU connector) on a CPU over ~ 140W which Just Shouldn't Be Done At All. Even in that case, you might need the heatsink performance more or to switch motherboards instead of trying to depend on a CPU heatsink to do double-duty.

 

[tarmc]

would the evo 212 work? way cheaper option? noctua does make a decent low profile cooler as well. doesnt leave room for overclocking though

 

[piokos]

curious which cpu youre cooling? overclocked?

Just a humble i5 at the moment, but I'm going for 8C next time (9700K or Intel 10th). No overclocking.

would the evo 212 work? way cheaper option? noctua does make a decent low profile cooler as well. doesnt leave room for overclocking though

212 is much larger and somehow loud. But loudness aside, I don't like the sound it makes.
Heatsink is actually very good, but since the cooler is cheap, it has to use an unrefined fan.

If I got a 212 now, I'm 90% sure the fan would have to be replaced at some point. That moves the total cost into Noctua territory.
Also, since 212 heatsink is very dense (57 fins), it needs a high pressure fan - these are usually louder.

You're overthinking it. A tower sink oriented so its fan is blowing the same direciton as the chassis airflow is prefect for that case and you might not even need a fan on it if the heatsink rating is for around double the CPU rating.

That is correct. People have successfully cooled 65W CPUs with just a big cooler and this case's airflow. The rear fan is so close to the heatsink it almost acts as a "pulling" fan on the radiator.
But I'm planning a bigger CPU in the future.
The PC runs 24/7 and I'm looking for a cooler that's unaudible at low load, but can handle hours of 100% when I need it - also after CPU upgrade.

I'm totally aware of Noctua price premium, but to be honest - I'm going to replace the 3 case fans with Noctua's as well (2x NF-A9 and NF-A14). Low noise is really important for me (PC in bedroom). Going for the best fans is not a big price for the enormous comfort. I can save money somewhere else - e.g. eat beans for a week (actually, I like beans).

 

[ElFenix]

i run a quad core haswell xeon with a fanless hyper 212 in a node 304. it's constantly loaded. it's fine. the rear fan draws enough air. air doesn't go around the cooler. it isn't mercury. if you're really concerned you can make a duct out of some cardboard or corrugated plastic to draw air through the tower.

if you get a more power hungry cpu later you can always add a fan then.

 

[mindless1]

9700K is still only a 95W CPU? It should not be hard to keep cool enough with any median grade heatpipe 'sink and a fan at low RPM. Does your motherboard not allow setting fan ramp speeds based on CPU temperature?

Regardless, a traditional vertical heatpipe sink will work best with that case airflow. If you want a very expensive fan for it, that's an option as long as the 'sink chosen has a bracket that works with standard fans or you make one or just strap it on with nylon wire ties or whatever, though if the case fan is that close, I would think about making a shroud/duct over the heatsink to let the case fan pull more air through it, then do some full load testing before putting a fan on the heatsink.

 

[piokos]

i run a quad core haswell xeon with a fanless hyper 212 in a node 304. it's constantly loaded. it's fine. the rear fan draws enough air. air doesn't go around the cooler. it isn't mercury.

I've read a bit since yesterday and checked some builds on pcpartpicker.
U12S is out of the race. The fan is not great (not for the price Noctua asks, anyway).
So it leaves me with either D9L or U14S. I actually though the latter doesn't fit. It does. Just...
U14S is above my initial budget, but I'll make it up somewhere else. It's just money.

There's just something special about D9L's compactness that makes me want to buy it.

Testing with and without a fan is going to happen, obviously. And I have a nasty plan for airflow / turbulence experiment as well - I just need a tiny camera. ;-)

if you're really concerned you can make a duct out of some cardboard or corrugated plastic to draw air through the tower.

I actually though about it few days ago. I have a friend who can make me a 3D-printed element.
It's clearly an option worth trying out -maybe not to force air through the cooler, but just to push it downwards once it exists the disk area. It's not like there are many things worth cooling in the top half of the case...

9700K is still only a 95W CPU? It should not be hard to keep cool enough with any median grade heatpipe 'sink and a fan at low RPM. Does your motherboard not allow setting fan ramp speeds based on CPU temperature?

95W in base clocks. It boosts to around 130W.
Noctua provides a CPU guidance instead of TDP (I love the idea).
https://noctua.at/en/cpu/Intel_Core_i7-9700K
D9L is described as not enough to cover full 9700K potential (U14S is), which usually means fans can get to max RPM and get loud.
That said, I'm pretty sure I'll provide more case airflow that the "average case" Noctua uses for testing.

 

[aigomorla]

Personally unless your going to overclock til no tomorrow, i honestly think you'll be fine on the D9

The U12S will be better because noctua's 120mm fans have a better noise to performance ratio.
120mm fans are generally where the best noise to performance ratio ends up on, even greater then 140mm when you go to the higher performance segment.

The fan is not great (not for the price Noctua asks, anyway).

the stock fan is a NF-F12 where did you hear the fan is not great?
Its quiet and pushes a lot of air, again for most likely whatever you intend to overclock at, unless your going after high overclocks, which then your using the wrong case to begin with.

 

[piokos]

Personally unless your going to overclock til no tomorrow, i honestly think you'll be fine on the D9

I'm 100% covered by D9L when it comes to heat transfer. It's only a matter of noise.
Fan in D9L is 400-2000 RPM.
Fan in U14S is 300-1200 RPM (still with higher max airflow).
So whereas a CPU could push D9L to 1600+ RPM (where it gets relatively loud), U14S will stay around 800-100.
Price difference is big on a cooler alone (+40%)), but not that painful in the big picture.
I mean: the big picture is painful anyway.
3 case fans + D9L ~= 110EUR
3 case fans + U14S ~= 130 EUR
Spending all that money to learn that under heavy load D9L ruins the experience - priceless. ;-)

the stock fan is a NF-F12 where did you hear the fan is not great?

Noctua, among other "premium" brands, usually works not just on noise volume, but also on noise quality. The unreachable aim is constant noise spectral density - typical for white noise (which we humans find neutral or even relaxing).
In other words: some 30dBa noises are more irritating than other.
IMO this is the whole point of paying the price Noctua asks (or bequiet!, Blacknoise etc.).

NF-F12 is a very odd fan in Noctua's lineup. It's optimized for static pressure - they made it to attract watercooling users, hence the 120mm size (it's the only NF-F to date).
Blades are flat and there are huge spikes in frequency spectrum. They use a similar design in their Industrial lineup.
In other words: as long as you can't hear it, it's OK. But you do, it's worse than NF-A12 would be.

Sadly, they decided to use NF-F12 in 2 CPU coolers (U12S and L12). I don't know the reason - maybe they simply wanted some media coverage (fans bundled with coolers are reviewed more).

As for my source.
Once upon a time there was a fantastic site that covered PC noise in an advanced, quantitative way: Silent PC Review.
http://www.silentpcreview.com/article1343-page5.html
"The F series may simply have been a short-lived experiment that has reached its conclusion."

But I've already contacted a person who owns this fan. We got a similarly bad result (despite very rudimentary audio recording equipment - his smartphone ;-) ).

 

[ClockHound]

Might consider the new NH-U12A - comes with a pair of of the new NF-A12x25 fans, successor to the P12. And airy tribute to the GTs... While testing shows it keeps up with the U14S, so does the price. A little well. ;-)

If you can one find one, the Scythe Fuma is an excellent small cooler for low cost. As is the Thermalright Macho Direct. Very quiet too.

I've used the D9L with 2 fans in an itx case on a mildly clocked Broadwell. Noise was not obtrusive. I did cheat tho. Made an intake shrowd for a Darkside PWM GT that fed the D9L.

 

[piokos]

Might consider the new NH-U12A - comes with a pair of of the new NF-A12x25 fans, successor to the P12. And airy tribute to the GTs... While testing shows it keeps up with the U14S, so does the price. A little well. ;-)

Sure, great cooler based on the reviews we've seen to date.
But it's also hugely expensive, especially in my case.
I could get it and use one fan for case exhaust, but that makes U12A a 120mm fan cooler for €70. U14S can be had for that...

In fact it's more like €80, because while Noctua asks €30 for the 120mm PWM fan, the 140mm FLX can be had for €20...

If you can one find one, the Scythe Fuma is an excellent small cooler for low cost. As is the Thermalright Macho Direct. Very quiet too.

Very big too. ;-)
I'd prefer a cooler that clears RAM totally - something D9L and U12A provide (U14S sadly doesn't).
Since I'm staying on the mITX, I might want to go for double height RAM modules one day (or Optane).

I've used the D9L with 2 fans in an itx case on a mildly clocked Broadwell. Noise was not obtrusive. I did cheat tho. Made an intake shrowd for a Darkside PWM GT that fed the D9L.

Honestly, I got really interested by this issue over the last few weeks.
I'm going to do some experiments next month. ;-)
I'm almost 100% sure I'll add some air tunnels / baffles. Node 304 is perfect for that. I'm also a huge fan of highly controlled airflow (like in some OEM cases).

 

[ClockHound]

Might consider a smaller Macho (if your ego can take it ;-)

It cools almost as well as some 140mm towers while keeping quiet and saving space for high altitude RAM:

See ehume's review here: https://www.overclockers.com/thermalright-macho-120-sbm-review/

 

[Confirmation]

Might consider a smaller Macho (if your ego can take it ;-)

It cools almost as well as some 140mm towers while keeping quiet and saving space for high altitude RAM:

See ehume's review here: https://www.overclockers.com/thermalright-macho-120-sbm-review/

is not that heatsink sending hot airflow to the mems?
wrong position?

 

[ClockHound]

is not that heatsink sending hot airflow to the mems?
wrong position?

Nope. Just the MB sinks look like stylish heatsunk RAM modules...

The actual stylish heatsunk RAM are on the left in that image. And fan is mounted normally to push air through the fins.