Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

Dare I say it, but I think you're all using fans wrong.

Chen G
 Share

People talk about airflow management but I think everyone has been doing this wrong all this time. Conventional wisdoms like mounting fans on radiators, or having exhaust fans etc, it's just wrong and I have a build to prove it:

Repeating the video description here, the fan speed on this build is controlled by the front panel speed for the demonstration, and it is on MAX SPEED. This is as noisy as it will ever get.

I'm also not sticking some super lower power stuff in there, what you're looking at is a 7980XE and a 2080TI. Admittedly I am not overclocking the crap out of them but they are running above stock.

So what about the thermals, am I just constantly butting up against 99 degrees and throttling? Absolutely not, a general figure I can give is 70 degrees maximum for GPU at 500w CPU+GPU power draw at 23 degrees room temperature. CPU thermals will depend on the interface, delid, lapping, LM etc, because that's going to be the biggest bottleneck.

I'm sure this level of silence and thermal performance can also be done on a huge elaborate loop maybe with dual 480 rads or something. While a loop is required, I am using a pretty basic EKWB kit.

 

DSC09989.thumb.jpg.a6f1ad592fb5dc1e7e62bc863ce87ab6.jpg

 

To explain what I did here, I'm going to point out the top three problems conventional fan placements have:

 

1. Wasted airflow. If you get air into your case and it just flows directly back out without cooling anything, you're generating noise for no thermal gains.

This is extremely common with the standard front intake top exhaust case layout. Think about that top corner, what's going on there? The front fan sucks in air and it immediately goes back out the top, you've basically generated some noise for nothing. 

 

2. Sub-optimal temperature difference. When air flows past a fin, the amount of heat transfer is determined by the difference in temperature, if you put a radiator at your intake, the air you take in have reduced cooling effect on everything else in the case, this again make you generate more noise with no thermal gains.

34rfe.thumb.png.359436857af2b681d90b7ceea0c7adfb.png

Look at this here, we've got these radiators on the intake, which means your hybrid cooling cards are no longer hybrid they're basically just running on one 120mm rad, because the hot air isn't going to cool the rest of the air cooling heat sinks all that much.

 

And then look at that traditional read exhaust fan, what's the point of that? Aren't you just sucking cold air out? Wouldn't it be better if all your available air either go through the top radiator or the graphics cards?

 

3. Aerodynamic disruption. Placing fans right up against something, whether it's a solid piece of glass or a semi-permeable structure like a radiator makes it far less efficient, even for the best fans.

To see how much of a difference this makes all you have to do is take off one of your fans, turn on the computer, and hold the fan either in open air, or up against its regular position. If you're running any significant fan speeds, the difference will be very noticeable. You're making a lot more noise, but defiantly less air. All those impressive fan specs you see on Noctua's website, well most of those are for open air operation, not up against a radiator.

 

I basically designed a build specifically to eliminate all of these problems, and there is exactly one case not just compatible, but perfect for what I'm trying to do here:

 

The Cooler Master SL600M

 

Now if you read the reviews, you'll notice that it doesn't have the best reviews, a common complaint is high CPU temperatures, which is to be expected, as the graphics card is not only blocking the air path to the CPU, but also pre-heating the air, exacerbating the 2nd problem I mentioned above. However, the solution to that, is just forget about how the air interacts with the parts inside, but focus on the entrance and exit, because those are the spots where we can be certain all the air will have to go through.

DSC09919.thumb.jpg.1b9191f05dbe4b070dc3a7c01c39121e.jpg

There's two giant 200mm intake fans, so we know air must come in here, and then since the front panel is closed, and there's no exhaust fan on the usual spot, we know the air must exist through the top side, where a radiator mount is conveniently located. At first I still had the traditional mindset of sticking fans on radiators so I had the fans over the radiator to suck out the heat, it worked but noise did not reduce compared to my previous build with the same 360rad and fans.

 

But then it dawned on me, that there's just no reason to have those fans, if I just seal the case such that all air must escape through the radiator, then I will get an airflow through the radiator because the bottom fans are constantly taking in air, there's no where else for the air to go!

DSC09910.thumb.jpg.e9d381f1549b5941dbe9f72d7a35cb37.jpg

 

There is actually no attempt from CM to seal the case for this use case, there's a huge gaping hole around the power switch which I had to cover up with cardboard. If I didn't do this none of this would work because the air would just go through those holes rather than through the radiator which has more resistance. There's also the issue of expansion card brackets. Most cards for cooling reasons will put exhaust grids on their brackets, but we can't have that in this build because the air will just escape through those grids rather than through the top radiator.

 

Now let's look back at our three airflow problems and see how they've been solved:

 

1. There is no wasted airflow at all, because all air must escape through the top radiator, which cools the CPU and GPU, so we do not generate any unnecessary noise.

2. The air that gets to the top radiator will barely be carrying any heat because well, the CPU and GPU's heat are in the radiator, so we get maximum heat transfer.

3. While there is an air filter against the bottom fans, there's basically zero obstruction near the fans, so they can operate very close to their open air performance.

 

In fact, by doing this with this specific case we get some more additional benefits:

4. Because the intake fan is 200mm, and max fan speed is only 800RPM, the noise to airflow ratio is extremely high and unmatched by any 120mm or 140mm fan.

6. Because of the huge intake area of 2x200mm fans, intake airspeed is very low. Not only does this mean low noise, it also sucks in less dust/particles. the filter doesn't need to be cleaned for at least half a year or more.

7. There's only two fans in the system (the one 120mm you see behind the panel is for back-up when the side panel is open), and it is extremely easy to set up an almost ideal fan curve to further minimize dust build-up (because noise is already minimal). I just stick a thermal probe in the radiator (or if you want to be fancy, a water temperature probe), and all you need to do is change the fan speed with the water temperature. You get an extremely gradual ramp, and you get exactly what your system needs for keeping cool.

DSC09974.thumb.jpg.75a53a47ead76c8690914e871226b5a8.jpg

 

Again, this is at max fan speeds, and I'm placing the phone right against the exhaust port. You can see on the graph the noise peak is at ~150hz, a very subtle low frequency humm, if you have good enough ears to even hear it.

 

464733228_ScreenShot2020-05-17at12_05_33AM.thumb.png.2ee30656d8acba541d31536b6f59ebde.png

 

The main reason most DIY cases don't do this is probably the requirement on sealing. The approach is actually not uncommon with purpose designed machines like this Mac Pro. While it doesn't have a water loop, there's similarly no fan on the CPU or GPU heat sinks. Instead the fans are at the front, which gets tunnelled to the heatsinks. This way you don't reduce the efficiency of your fans.

Other examples of this are the trash can Mac Pro, and the new Xbox Series X, the ideas are the same. You somehow make a tunnel with all your heatsinks inside, and all you have to do is put one fan at the start or end of the tunnel to cool everything in the tunnel. No wasted airflow, and your heatsinks always get the coldest air possible. Could even take some slight advantage of the chimney effect.

 

DSC00013.thumb.jpg.7cd28880d1b117236376268fc7b2cc05.jpg

 

This is an experimental build I just made, while the thermal and acoustic performance of this is nowhere near as good as the big build. To my defence, there is only room for a 120mm rad and this machine, and despite being ITX, I think is as good as any full sized build with 120mm AIO. The airflow is less clean on this, but there's still no fan on the AIO radiator, the air pressure will find its way out of the system so it doesn't need to be linear.

DSC00026.thumb.jpg.2936625d40f904f9c72b2fdeab8584f0.jpg

So Again I've completely sealed the case with packing tape, except for the 120mm AIO radiator, and the exhaust port of the blower 5700xt. Yes the blower 5700XT, and I actually specifically picked the blower because the regular third party rack cooler design would just allow air to escape through the rear bracket without passing through much of the GPU's heatsink. The 2 140mm RGB fans are both intake, and once the air gets in, they can only escape either through the GPU blower, or the CPU radiator.

 

This is ITX and I'm working with only 2 140mm fans but thermals/noise efficiency is actually better than open air. And the reason is quite simple, I am effectively using 2x 140mm fans to push one blower 5700xt and one 120mm radiator. If it was just open air, I'd only have 1x 120mm and 1x ~50mm fan to do the same thing. So another way to think about this approach, is it allows you to mount more or bigger fans on your radiator than you other wise could, and run those fans at increased efficiency.

 

The interesting thing is I got this case second hand, and the original owner gave me a mesh side panel and told me to use that to improve thermals. The fact that Phanteks sells that just means most builds in this case have worse thermals than open air. Also just from watching Gamer's Nexus, I get the impression that the accepted norm is that almost any case is not as good as open air. However by actually using fans correctly here, I have somehow achieved better than open air performance in an ITX case. Of course it won't beat a proper third-party rack cooler plus bigger AIO on open air, but that's not the point.

 

I'm not trying to pick on anybody here, it's a sick build I love it but I'm just taking a random example to illustrate the problem.

So he's got a radiator on the intake, which seems to be quite common now days. Provided the CPU isn't that hot so the pre-warming effect won't be too significant, the GPU is air cooled so it still has to work extra hard than if the radiator was at the exhaust. Just this one placement difference reduces the thermal/noise efficiency of your GPU.  There's three exhaust fans up top but they're too far away to be sucking the GPU's exhaust out so I'd guess they're sucking out a lot of the air from the radiator, but since your CPU is so efficient, you're moving a lot of air to cool a not very hot CPU, wasting existing airflow capability.

Another stunning build, but this dual rad on both intake/exhaust layout is also pretty common and it doesn't maximize thermal/noise efficiency. The first obvious problem is fans against the radiator, so i already talked about that. But then if we think about this, you are basically taking the exhaust air of the first rad into the case, which is bad for the passively cooled components, and then you run that through the second rad, which will be no where near as efficient as the first one, but it costs just as much and generates just as much noise? 

I'm not sure how much I can do better with a traditional case like this, because this is definitely partially the case's fault, there's really no room to mount unobstructed fans other than the traditional exhaust spot, which is not only unnecessary but counter productive since again, you'd rather have more air go through your radiators than through nothing. I guess what I would do here is just have one radiator as before, remove the fans, and put all other fans on intake, but then there's no dust filter on some of these so that's gonna be much harder to maintain.

 

This looks psychedelic and interesting, but there's just massive inefficiency going on here, a lot of air will escape before they even get a chance to touch anything, especially through that top hole, And even if that was filled with a 360 rad instead it's not much help because the air through that fan will have generated the noise of 2 120mm fans, but without cooling much of anything except 1/3 of your CPU.

 

Looking at all these builds here, I realize mine is seriously lacking in aesthetics, and that's going to be the focus point of mk.2 I am currently planning. No more cheap EK kits and industrial black tubes, this time I'll be using full EK Quantum line of parts. However I am in a big bind here, there's no way I'm going to buy another 2080ti at this point, so most likely the build won't be finished until next-gen GPUs come out, until then I'll probably just run some generic card with stock cooler, it won't look very impressive. I also haven't decided on if I wanna do hard tube, while that looks good it'll be such a pain to change anything, like for example upgrade the CPU.

 

In terms of cooling, it'll mostly be the same as before, single 360mm rad. While there is space on the bottom intake for radiators, like some builds have done, I don't think I'll actually do that because I think that will give me overall worse thermal/noise ratio because I won't be able to use the 200mm fans. Unless I just stick a thin 360 rad over the 200mm fans, but then I might as well just make the top radiator thicker, it's effectively the same thing, which is exactly what I'm going to do. Instead of the old 38mm rad on top, I'm ordering a 45mm rad. Now there's technically no room but I found the radiator rack can be raised and not interfere with the outer cosmetic shell. So I'm going to just use some MB standoffs to raise the radiator further up so it can be thicker. The bottom panel above the fans will now be filled with the flat radiator I really really love, and hopefully it'll cover up the ugly I/O and fan headers at the bottom of the MB.

 

=======================

 

There's an additional complication with the 2nd problem with conventional fan placement I mentioned towards the beginning. That is, paradoxically, stronger cooling is less efficient.

What do I mean by this? Well think of a given CPU cooler and air flowing past it, if the flow is constant, the amount of thermal transfer happening here is NOT constant, it depends on the temperature difference between the air and the cooling fins. This means the hotter the fins, the more cooling you can do, and this usually means higher CPU temperatures. Therefore, in reverse, if you have a big 360mm rad on your 65w CPU, and it's 55 degrees under load, all that means is you are pushing a lot more air to cool 65w, than if the radiator was 120mm and the CPU is say, 75 degrees under load. Basically, you need exponentially more airflow (and therefore noise) to cool the same thermal power, if you want to keep your heatsinks cool.

 

There is a way around this, and that's to reduce the temperature delta between the cooling fins and the thing you're trying to cool, it works because you aren't changing the temperature of your cooling fins, so you maintain the same amount of heat transfer between the fins and air, but you still make what you're trying to cool, cooler.

 

So for the efficiency minded builder, you need to keep your priorities straight:

 

1. Do not try to achieve unnecessarily low temperatures, cooling fetish is a real thing but you gotta remember the reduction in performance is exponential as your heat sink gets cooler. Just what kind of temperatures are good well that's a whole new can of worms we won't get into here.

2. First thing you do to improve cooling is to improve the thermal interface, better paste, liquid metal, delid, lapping etc... Because these methods of reducing temperature are "free" and don't require more air flow / noise. This is one of the reasons why water cooling is so ridiculously good, you can minimize the temperature delta between lots and lots of cooling fins, and your CPU. Where as an air cooler's fins get colder very quick as they get more distant from the CPU.

3. Increasing surface area of cooling fins. Air is just the worst at conducting heat, so a few millimetres between two cooling fins usually means there's like 1 millimetre of air in the middle that's not getting any heat transfer. So even with the same amount of flow, there's usually ways to dump more heat into it before the air leaves. This could mean denser fins, thicker rads, bigger rads, those all have their own pros and cons obviously, but find ways to dump more heat into the same amount of air. Average temperature of all the exhaust gas should ideally be as high as possible, that's a sign of an efficient cooling system.

4. Increasing airflow should always be the last resort. Higher fan speed is less efficient, more flow separation, more vibration, more noise, but not as much more airflow. More speed also means the air has less time to contact the fins, less heat transfer per unit volume of air. The ideal situation you want to have is a big radiator at high temperatures, and a uniform, slow airflow through it. Not a jet engine with a small heat sink inside.

 

Link to comment
Share on other sites

Link to post
Share on other sites

This is really interesting! I find all this talk about airflow and cooling performance recently quite informative. I have a couple thoughts/questions. I'm not and expert and have no experience testing this kind of stuff so bear with me if I say anything stupid 😁

 

 

25 minutes ago, Chen G said:

2. Sub-optimal temperature difference. When air flows past a fin, the amount of heat transfer is determined by the difference in temperature, if you put a radiator at your intake, the air you take in have reduced cooling effect on everything else in the case, this again make you generate more noise with no thermal gains.

This doesn't necessarily seem like a case against fans on radiators to me, rather just a trade off. In a scenario with a front rad with fans on it as an intake, the temp difference for the components further into the case will be smaller because that air is already warmer from the rad. However, the difference between the rad and the air it's intaking from outside the case is greater. So basically the loop would be cooled better, but dump more heat onto the other internal components?

 

 

30 minutes ago, Chen G said:

But then it dawned on me, that there's just no reason to have those fans, if I just seal the case such that all air must escape through the radiator, then I will get an airflow through the radiator because the bottom fans are constantly taking in air, there's no where else for the air to go!

The sealing of the case to force all of the air through the rad without fans is super interesting to me. I've long been an advocate of positive pressure because it seems to me that you're working with air pressure rather than against it (and keeping the dust out!) but this is an extreme I'd never considered. 

 

This makes me think: Maybe fans on that rear radiator running at a low speed might have the same effect as the seals? Even at very low RPM, I would think they would coax almost as much air through the rad as the seal does, and wouldn't add much if any noise. A low or even flat curve on those with a higher one on the intakes might do the trick.

That's a lot less effort than sealing up the whole case. 

 

Lenovo Ideapad 720s 14 inch ------ One day I'll have a desktop again...

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Chen G said:

The 2 140mm RGB fans are both intake, and once the air gets in, they can only escape either through the GPU blower, or the CPU radiator.

I have an Evolv Shift system. Did you compare this configuration vs other config  on this case?

 

In my case, early fan config in this was,

 

1) Twin 120mm rads (1 on the bottom front and 1 on the case floor) both as intakes, and 140mm fan upper front and 120 slim fan underneath the hood as exhaust, the overall system temps was okay, good on the GPU temps but not so much on the CPU, system, and storage drives, they all were heated more than I expected.

 

2) Changed it to the 140mm upper front fan as intake and the rest were unchanged.

It ended up worse than before.

 

3) Changed to 140mm upper front as intake, 120mm bottom front as intake, 120mm bottom exhaust, hood fan as exhaust. The result not much different than the 2nd config.

 

4) Changed to 140mm upper front as intake, and all the rads as exhaust, hood as exhaust. This is the best fan configuration I can get with 2 rads on the bottom of the case, as they doesn't dump alot of heat anymore inside the case which heats up the motherboard, VRM, and storage drives. Overall system temps are considerally better.

 

When both rads were intakes when the system under the load they dissipates heat by 180W from the GPU and 58W from the CPU, even the glass panels feels hot and the VRM heatsink is untouchable.

But now with front 140mm as intake and both rads as exhaust, the internal space of the case is much cooler than before. And max temps now are 5°C lower than before under the load.

SILVER GLINT

CPU: AMD Ryzen 7 3700X || Motherboard: Gigabyte X570 I Aorus Pro WiFi || Memory: G.Skill Trident Z Neo 3600 MHz || GPU: Sapphire Radeon RX 5700 XT || Storage: Intel 660P Series || PSU: Corsair SF600 Platinum || Case: Phanteks Evolv Shift TG Modded || Cooling: EKWB ZMT Tubing, Velocity Strike RGB, Vector RX 5700 +XT Special Edition, EK-Quantum Kinetic FLT 120 DDC, and EK Fittings || Fans: Noctua NF-F12 (2x), NF-A14, NF-A12x15

Link to comment
Share on other sites

Link to post
Share on other sites

The more I think about this, maybe a case that's truly optimized for only cooling performance might include several separate thermal zones? Even just two for a liquid-cooled rig might work. There could be a separate chamber just for the rad(s), this way they could intake cold external air, and then it could be passively exhausted without coming into contact with any of the other system components. Then the main chamber can intake cold external air for itself, and exhaust it (with or without fans) without having to worry about that hot air passing through a rad, making it less efficient.

 

Basically the same theory as a bottom-mounted PSU being its own isolated thermal zone, except for your entire system's cooling setup. Does this make any sense or am I just rambling? Haha

Lenovo Ideapad 720s 14 inch ------ One day I'll have a desktop again...

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, Spork829 said:

This doesn't necessarily seem like a case against fans on radiators to me, rather just a trade off. In a scenario with a front rad with fans on it as an intake, the temp difference for the components further into the case will be smaller because that air is already warmer from the rad. However, the difference between the rad and the air it's intaking from outside the case is greater. So basically the loop would be cooled better, but dump more heat onto the other internal components?

But it is a case against putting radiators on intake fans, which seems to be quite common now days.

If you have the CPU rad and an air cooled GPU, it's hard to say which one is better, radiator intake or radiator exhaust, probably depends on a lot of things. However if you have everything in one radiator, it's better to have the radiator on exhaust, because some passive components require non-trivial amount of air cooling, like the SSD and memory. And if you just need more radiators, it's better to have one thicker radiator on exhaust than one thin radiator on each. It's equally effective but cheaper and less noisy.

8 hours ago, Spork829 said:

The sealing of the case to force all of the air through the rad without fans is super interesting to me. I've long been an advocate of positive pressure because it seems to me that you're working with air pressure rather than against it (and keeping the dust out!) but this is an extreme I'd never considered. 

 

This makes me think: Maybe fans on that rear radiator running at a low speed might have the same effect as the seals? Even at very low RPM, I would think they would coax almost as much air through the rad as the seal does, and wouldn't add much if any noise. A low or even flat curve on those with a higher one on the intakes might do the trick.

That's a lot less effort than sealing up the whole case. 

I don't think that would work as well, you have to effectively use the fan to overcome the resistance of the radiator such that it is significantly lower than other holes on the case, but a hole is a hole so it has almost no resistance.

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, Spork829 said:

The more I think about this, maybe a case that's truly optimized for only cooling performance might include several separate thermal zones? Even just two for a liquid-cooled rig might work. There could be a separate chamber just for the rad(s), this way they could intake cold external air, and then it could be passively exhausted without coming into contact with any of the other system components. Then the main chamber can intake cold external air for itself, and exhaust it (with or without fans) without having to worry about that hot air passing through a rad, making it less efficient.

 

Basically the same theory as a bottom-mounted PSU being its own isolated thermal zone, except for your entire system's cooling setup. Does this make any sense or am I just rambling? Haha

 Bottom mounted PSU is basically for aesthetics, I don't think it helps with cooling at all. The PSU only generates about 10% of your total heat but it's as large as a big CPU heat sink.

 

The reason some products look like they have compartments like the Mac Pro, is because cooling fins have resistance so if you didn't have anything, the air would just rather go around the fins instead of between them. So what you do is make compartments where air have to go through the cooling fin. In my case, since all the air must go through the radiator, it doesn't matter what path they take inside of the case.

 

As for minor components pre-heating the air, obviously given the same air flow, doesn't matter how you arrange the inside you end up with the same overall cooling, equal to the total thermal energy gained by the exhaust air. different arrangements like this first or that first will only affect the distribution of temperature. Like for example hotter VRM versus hotter CPU core. Not sure which is better though, have to think more about that one. My gut feeling is radiator last is better because things like capacitors are probably not as long lasting as the chips themselves.

Link to comment
Share on other sites

Link to post
Share on other sites

So then would the Dune Pro case be suitable for a gaming PC? You have up to three 120mm fans in the front and if you really want to add two 80mm fans for exhaust. 

CPU Cooler Tier List  || Motherboard VRMs Tier List || Motherboard Beep & POST Codes || Graphics Card Tier List || PSU Tier List 

 

Main System Specifications: 

 

CPU: AMD Ryzen 9 5950X ||  CPU Cooler: Noctua NH-D15 Air Cooler ||  RAM: Corsair Vengeance LPX 32GB(4x8GB) DDR4-3600 CL18  ||  Mobo: ASUS ROG Crosshair VIII Dark Hero X570  ||  SSD: Samsung 970 EVO 1TB M.2-2280 Boot Drive/Some Games)  ||  HDD: 2X Western Digital Caviar Blue 1TB(Game Drive)  ||  GPU: ASUS TUF Gaming RX 6900XT  ||  PSU: EVGA P2 1600W  ||  Case: Corsair 900D  ||  Mouse: Logitech G502 Hero SE RGB  ||  Keyboard: Logitech G513 Carbon RGB with GX Blue Clicky Switches  ||  Mouse Pad: MAINGEAR ASSIST XL ||  Monitor: ASUS TUF Gaming VG34VQL1B 34" 

 

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, CommanderAlex said:

So then would the Dune Pro case be suitable for a gaming PC? You have up to three 120mm fans in the front and if you really want to add two 80mm fans for exhaust. 

Even just looking at the pictures, it doesn't look nearly as refined as the actual Mac Pro, the side panel isn't even that flush with the front.

 

That aside, it's just not the same as the Mac Pro, you still need fans on your CPU and GPU, it's just a shallow surface level imitation.

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, Chen G said:

Even just looking at the pictures, it doesn't look nearly as refined as the actual Mac Pro, the side panel isn't even that flush with the front.

 

That aside, it's just not the same as the Mac Pro, you still need fans on your CPU and GPU, it's just a shallow surface level imitation.

Sure of course, you would still need fans on your CPU/GPU but overall, the airflow flows from the front towards the back of the case. 

CPU Cooler Tier List  || Motherboard VRMs Tier List || Motherboard Beep & POST Codes || Graphics Card Tier List || PSU Tier List 

 

Main System Specifications: 

 

CPU: AMD Ryzen 9 5950X ||  CPU Cooler: Noctua NH-D15 Air Cooler ||  RAM: Corsair Vengeance LPX 32GB(4x8GB) DDR4-3600 CL18  ||  Mobo: ASUS ROG Crosshair VIII Dark Hero X570  ||  SSD: Samsung 970 EVO 1TB M.2-2280 Boot Drive/Some Games)  ||  HDD: 2X Western Digital Caviar Blue 1TB(Game Drive)  ||  GPU: ASUS TUF Gaming RX 6900XT  ||  PSU: EVGA P2 1600W  ||  Case: Corsair 900D  ||  Mouse: Logitech G502 Hero SE RGB  ||  Keyboard: Logitech G513 Carbon RGB with GX Blue Clicky Switches  ||  Mouse Pad: MAINGEAR ASSIST XL ||  Monitor: ASUS TUF Gaming VG34VQL1B 34" 

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, CommanderAlex said:

Sure of course, you would still need fans on your CPU/GPU but overall, the airflow flows from the front towards the back of the case. 

so? the direction of airflow really doesn't mean anything.

Link to comment
Share on other sites

Link to post
Share on other sites

6 hours ago, Chen G said:

 Bottom mounted PSU is basically for aesthetics, I don't think it helps with cooling at all. The PSU only generates about 10% of your total heat but it's as large as a big CPU heat sink.

 

The reason some products look like they have compartments like the Mac Pro, is because cooling fins have resistance so if you didn't have anything, the air would just rather go around the fins instead of between them. So what you do is make compartments where air have to go through the cooling fin. In my case, since all the air must go through the radiator, it doesn't matter what path they take inside of the case.

 

As for minor components pre-heating the air, obviously given the same air flow, doesn't matter how you arrange the inside you end up with the same overall cooling, equal to the total thermal energy gained by the exhaust air. different arrangements like this first or that first will only affect the distribution of temperature. Like for example hotter VRM versus hotter CPU core. Not sure which is better though, have to think more about that one. My gut feeling is radiator last is better because things like capacitors are probably not as long lasting as the chips themselves.

Yeah, I suppose I'm sort of ignoring overall thermal output here. I do wonder if bottom-mounted PSUs have better cooling performance though, since they're taking in outside air and therefore would have a greater cooling potential than a top-mounted one taking in pre-heated air.

Lenovo Ideapad 720s 14 inch ------ One day I'll have a desktop again...

Link to comment
Share on other sites

Link to post
Share on other sites

Thanks for sharing my build and calling me in, it is an interesting topic.

 

I have always been fascinated from this topic, it's something that deserves more reseach.

 

In addition to the examples you made (which are air-cooled), the Corsair One is also using this kind of "tunneling" and "passive cooling" approach but with liquid cooling (it has a single 140mm fan on top as exhaust, the air gets sucked in passively from the sides and then goes through the rads). You might want to check it out, Linus and several other youtubers made videos about it. The results of "Convection-Assisted Cooling" (that's how they call it) are impressive.

 

 

 

 

 

 

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

I started to read, but lots to ingest - so either I missed some variables you touched on or you aren't touching on other variables.  

 

I guess "right" and "wrong" convention to airflow is...superseded by the goal of the build, imho.  

 

I have builds with front, top, and bottom intake with only rear exhaust.  Front only intake everything else exhaust.  1 intake everything else exhaust (dust free in mind with this build).  

 

Now I delved deep, a while back, into OPTIMAL vs EFFECTIVE cooling.  If something is effective, is it wrong?  Only if they are looking for the most Optimal (and what is that?  Temps, noise, least/most rad space?) in a setup can someone be told its wrong (imho)

 

I plan on reading and ingesting this whole thing though, wanted to comment so I don't forget.

Current in process build: Project Odyssey - Build Logs - Linus Tech Tips

Workstation Laptop: Dell Precision 7540, Xeon E-2276M, 32gb DDR4, Quadro T2000 GPU, 4k display

Ryzen Rig 3: ASRock B450m Pro4, Ryzen 5 3600, ARESGAME River 5 CPU cooler in push/pull, EVGA RTX 2060 KO, 16gb (2x8) 3600mhz TeamGroup T-Force RAM, ARESGAME AGV750w PSU, 1tb WD Black SN750 NVMe Win 10 boot drive, 3tb Hitachi 7200 RPM HDD, Fractal Design Focus G Mini custom painted.  Lenovo 27" 1080p 60hz placeholder monitor, generic mouse, generic keyboard.  Still a work in progress

NVIDIA GeForce RTX 2060 video card benchmark result - AMD Ryzen 5 3600,ASRock B450M Pro4 (3dmark.com)

Ryzen Rig 2: ASrock B450 Pro4, Ryzen 7 1700 @ 4.2ghz all core 1.4vCore, AMD R9 Fury X w/ Swiftech KOMODO waterblock, Custom Loop 2x240mm + 1x120mm radiators in push/pull 16gb (2x8) 3600mhz V-Color Skywalker (or 4x8gb DDR4 2666mhz for large tasks), Corsair HX850 PSU, 250gb Samsun 960 EVO NVMe Win 10 boot drive, 500gb Samsung 840 EVO SSD, 512GB TeamGroup MP30 M.2 SATA III SSD, SuperTalent 512gb SATA III SSD, CoolerMaster HAF XM Case.  Zalman K600S keyboard, Zalman ZM-GM1 mouse, Viotek GN24C 24" 1080p 144hz curved and Hannspree HF207 as 2nd monitor

https://www.3dmark.com/3dm/37004594?

Ryzen Rig 1: ASUS B350-PRIME ATX, Ryzen 7 1700, Sapphire R9 Fury Tri-X Nitro 4gb HBM, 16gb (2x8) 3200mhz V-Color Skywalker, ANTEC Earthwatts 750w PSU, MasterLiquid Lite 120 AIO cooler in Push/Pull config as rear exhaust, 250gb Samsung 850 Evo SSD, Patriot Burst 240gb SSD, Cougar MX330-X Case.  Zalman K600S keyboard, Zalman ZM-GM1 mouse, Acer XF270HU 2560x1440 144hz IPS monitor

https://www.3dmark.com/3dm/37628874?

Dell OptiPlex 3040 MFF: Dell 0MGK50 A02, i3-6100T, 2x4gb DDR3 1600, Team Group 120gb SSD, 500gb Seagate 7mm HDD attached storage, Windows 10 Pro, Logitech K400+, USB Wifi adapter all vesa mounted to the back of a 37" 1080p TV 

Linux Box: Toshiba Laptop, i7 620M, NVS graphics, 4gb ram tinker toy at the moment.  Running Manjaro XFCE at the moment.

Home Security: ZOSI 8 channel CCTV (4 used at this time, 1080p) DVR H.265+, 3tb HGST Enterprise HDD, ASUS monitor for display

Link to comment
Share on other sites

Link to post
Share on other sites

so to point out the mac pro is a server build with server air flow like other's have pointed out. just like btx were are flow was optimized but never did catch on for some reasone maybe they wanted btx to be for the "Hi end" servers maybe?

 

in some cases making the air flow OPTIMAL in a small case can give better tems then in big cases with many fans. yes bigger the fan the better. there one case brad that is the best that that the siverstone rave case with the 90 degree mb layout and big fans on the bottom. but this is before we got the vertical gpu mount and well can mout you gpu anywhere even out side that case.... this will make it so non 90 degree mb can also be OPTIMAL.

 

in taking cool air thru a rad and exsosting out side is the best way of doing it....but like most case are not built for that.

 

all fans in the front were the air take a 90 or even an 180 and having less then 1 inch forn the front fan and the front panel is ideal but most case don't do this.  better yet have no front panel at all is better.

 

case trend have mostly stopped putting like 3 fans on top mostly 2 or 1 fans for the aio. you have to spend like $150+ to get 3 on top and 3 to 4 in front.... let alone some on the bottom

 

back in the old days pus on the top that fan did spin and it would take the air out. now a days the pus fans don't even spin..

 

look up an dellt3500 and look that it only have 2 fans in front and no cpu fan and there a gap for at lest an inch between the fan and heat sink. note that the nb , ram anf gpu all point in the same deretion.

gpub.jpg

 

 

oh side note is there people just dont care they build the build for looks.

Edited by thrasher_565
Link to comment
Share on other sites

Link to post
Share on other sites

On 5/18/2020 at 12:45 PM, Tristerin said:

I started to read, but lots to ingest - so either I missed some variables you touched on or you aren't touching on other variables.  

 

I guess "right" and "wrong" convention to airflow is...superseded by the goal of the build, imho.  

 

I have builds with front, top, and bottom intake with only rear exhaust.  Front only intake everything else exhaust.  1 intake everything else exhaust (dust free in mind with this build).  

 

Now I delved deep, a while back, into OPTIMAL vs EFFECTIVE cooling.  If something is effective, is it wrong?  Only if they are looking for the most Optimal (and what is that?  Temps, noise, least/most rad space?) in a setup can someone be told its wrong (imho)

 

I plan on reading and ingesting this whole thing though, wanted to comment so I don't forget.

Well if we want to be precise, anything non-optimal is "wrong".

 

If you could've arranged what you have differently, or if you could have bought something else with the same amount of money that are more optimal, then it's "wrong.

 

But of course one could decide that their current arrangement is optimal if taking aesthetics into consideration. Aesthetics are performance certainly do not necessarily go hand in hand.

Link to comment
Share on other sites

Link to post
Share on other sites

6 hours ago, Chen G said:

Well if we want to be precise, anything non-optimal is "wrong".

 

 

Your title to this thread is a profound statement.  So we have to be quite precise after saying (typing) that.  

 

 

Dare.PNG

Current in process build: Project Odyssey - Build Logs - Linus Tech Tips

Workstation Laptop: Dell Precision 7540, Xeon E-2276M, 32gb DDR4, Quadro T2000 GPU, 4k display

Ryzen Rig 3: ASRock B450m Pro4, Ryzen 5 3600, ARESGAME River 5 CPU cooler in push/pull, EVGA RTX 2060 KO, 16gb (2x8) 3600mhz TeamGroup T-Force RAM, ARESGAME AGV750w PSU, 1tb WD Black SN750 NVMe Win 10 boot drive, 3tb Hitachi 7200 RPM HDD, Fractal Design Focus G Mini custom painted.  Lenovo 27" 1080p 60hz placeholder monitor, generic mouse, generic keyboard.  Still a work in progress

NVIDIA GeForce RTX 2060 video card benchmark result - AMD Ryzen 5 3600,ASRock B450M Pro4 (3dmark.com)

Ryzen Rig 2: ASrock B450 Pro4, Ryzen 7 1700 @ 4.2ghz all core 1.4vCore, AMD R9 Fury X w/ Swiftech KOMODO waterblock, Custom Loop 2x240mm + 1x120mm radiators in push/pull 16gb (2x8) 3600mhz V-Color Skywalker (or 4x8gb DDR4 2666mhz for large tasks), Corsair HX850 PSU, 250gb Samsun 960 EVO NVMe Win 10 boot drive, 500gb Samsung 840 EVO SSD, 512GB TeamGroup MP30 M.2 SATA III SSD, SuperTalent 512gb SATA III SSD, CoolerMaster HAF XM Case.  Zalman K600S keyboard, Zalman ZM-GM1 mouse, Viotek GN24C 24" 1080p 144hz curved and Hannspree HF207 as 2nd monitor

https://www.3dmark.com/3dm/37004594?

Ryzen Rig 1: ASUS B350-PRIME ATX, Ryzen 7 1700, Sapphire R9 Fury Tri-X Nitro 4gb HBM, 16gb (2x8) 3200mhz V-Color Skywalker, ANTEC Earthwatts 750w PSU, MasterLiquid Lite 120 AIO cooler in Push/Pull config as rear exhaust, 250gb Samsung 850 Evo SSD, Patriot Burst 240gb SSD, Cougar MX330-X Case.  Zalman K600S keyboard, Zalman ZM-GM1 mouse, Acer XF270HU 2560x1440 144hz IPS monitor

https://www.3dmark.com/3dm/37628874?

Dell OptiPlex 3040 MFF: Dell 0MGK50 A02, i3-6100T, 2x4gb DDR3 1600, Team Group 120gb SSD, 500gb Seagate 7mm HDD attached storage, Windows 10 Pro, Logitech K400+, USB Wifi adapter all vesa mounted to the back of a 37" 1080p TV 

Linux Box: Toshiba Laptop, i7 620M, NVS graphics, 4gb ram tinker toy at the moment.  Running Manjaro XFCE at the moment.

Home Security: ZOSI 8 channel CCTV (4 used at this time, 1080p) DVR H.265+, 3tb HGST Enterprise HDD, ASUS monitor for display

Link to comment
Share on other sites

Link to post
Share on other sites

oh one other thing is that we use to use fan shrouds (aka cut the fan out and use the shroud) so the fan was not up  against the rad... but it only gave like a 1 degree difference. but still in thirty would be quieter and close to open air.

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Quote

 the reduction in performance is exponential as your heat sink gets cooler.

I am super curious to this statement. 

 

Is this in reference to temperature gradient and cooler performance?

- If it ain't broken, don't fix it! - - Your post codes and beep codes in the drop down below -

Competitive Benching Team - Warp9-Systems 

Save the old OC forums just by reading old school tec

https://forums.extremeoverclocking.com/index.php

They have the best F@H stats too. 

Spoiler

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

On 5/22/2020 at 12:42 AM, ShrimpBrime said:

I am super curious to this statement. 

 

Is this in reference to temperature gradient and cooler performance?

Well to a number of things.

 

Temperature gradient over big air coolers yes, since the heat pipe cannot maintain a constant temperature over its entire length, fins attached to the far sides will contribute much less to cooling, so a water loop with a radiator of the same size will perform better.

 

Improving thermal interface is better than improving the radiator stage of the cooling system. reducing the temperature delta between the radiator and the silicon gives you free temperatures, or in other words, more cooling performance at lower silicon temperature. At no expense of requiring more space or airflow.

 

But also, if you just tolerate say a 10 degree increase in silicon temperature, you can potentially significantly reduce cooling requirement, which may save money, save space, and reduce noise.

 

Link to comment
Share on other sites

Link to post
Share on other sites

On 5/22/2020 at 12:13 AM, thrasher_565 said:

oh one other thing is that we use to use fan shrouds (aka cut the fan out and use the shroud) so the fan was not up  against the rad... but it only gave like a 1 degree difference. but still in thirty would be quieter and close to open air.

 

 

but is that a 1 degree reduction for same fan speed or same noise level?

Link to comment
Share on other sites

Link to post
Share on other sites

11 minutes ago, Chen G said:

Well to a number of things.

 

Temperature gradient over big air coolers yes, since the heat pipe cannot maintain a constant temperature over its entire length, fins attached to the far sides will contribute much less to cooling, so a water loop with a radiator of the same size will perform better.

 

Improving thermal interface is better than improving the radiator stage of the cooling system. reducing the temperature delta between the radiator and the silicon gives you free temperatures, or in other words, more cooling performance at lower silicon temperature. At no expense of requiring more space or airflow.

 

But also, if you just tolerate say a 10 degree increase in silicon temperature, you can potentially significantly reduce cooling requirement, which may save money, save space, and reduce noise.

 

I remember being part of a conversation about running a high temperature gradient on air coolers.

The issue was always that the chips always idle so cool and leap 20-30c in temps from idle to load from sheer transistor density.

Another issue also is running higher VRM temps as a result because some of that Cpu heat is dissipated to the VRMs be it the hot air flowing on it, or heat transfer through the tracings in the mainboard pcb.

 

I've always liked the idea of having more copper mass opposed to less and many chips I've run direct die cooling eliminating passing through 2 interface materials.

 

 

 

- If it ain't broken, don't fix it! - - Your post codes and beep codes in the drop down below -

Competitive Benching Team - Warp9-Systems 

Save the old OC forums just by reading old school tec

https://forums.extremeoverclocking.com/index.php

They have the best F@H stats too. 

Spoiler

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, Chen G said:

but is that a 1 degree reduction for same fan speed or same noise level?

i don't no. there uses to be a website called skinny labs and he tested all kinds of wc stuff. but i think he ran out of money. there were arguments from the cheap yate loons fans fans vs the gentle typhoons witch was presher optomised.  (my dad picked up like 12 gelid 120mm fans pwm for like $4 each) i think maybe there were going to be discontinued but i think there still around but at like $25 a fan. al some fans. i still have 3 of em.

 

anyway the point is there are so many variables its hard to pick a best for each scenario. probly why most people just gave up trying to find the best of the best as there something new every few moths or the testing was not done right or w/e

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, thrasher_565 said:

i don't no. there uses to be a website called skinny labs and he tested all kinds of wc stuff. but i think he ran out of money. there were arguments from the cheap yate loons fans fans vs the gentle typhoons witch was presher optomised.  (my dad picked up like 12 gelid 120mm fans pwm for like $4 each) i think maybe there were going to be discontinued but i think there still around but at like $25 a fan. al some fans. i still have 3 of em.

 

anyway the point is there are so many variables its hard to pick a best for each scenario. probly why most people just gave up trying to find the best of the best as there something new every few moths or the testing was not done right or w/e

That's why I was talking about general principles and not specific parts.

Big fan/ more fans at low speed is always more efficient than small fan at high speed. Fan right up against radiator always less efficient than fan somehow placed away from radiator yet can still push air through radiator.

 

It doesn't matter how the fan is designed, if you move it away from a radiator it will be more efficient.

Link to comment
Share on other sites

Link to post
Share on other sites

30 minutes ago, Chen G said:

That's why I was talking about general principles and not specific parts.

Big fan/ more fans at low speed is always more efficient than small fan at high speed. Fan right up against radiator always less efficient than fan somehow placed away from radiator yet can still push air through radiator.

 

It doesn't matter how the fan is designed, if you move it away from a radiator it will be more efficient.

ya make sense. the rad is just like a fan grill. the the holes take up space were air gets slowed down.

 

i cant seem to find it but  silverstone was testing grill vs no grill and well the no grill was better then the wire grill then the honeycomb we have now.

 

another good YouTubeer i found that test odd ideas is major hardware. like 40mm fans vs 120mm fan on a rad.

 

ya we no that big 200mm fans can turn slow and push a lot of air but not alot of case alow for a 200mm. and the 200mm dose no have stranded holes. and alot of cheap 200mm fans the blade were braking off. i only used cosair hd fans for looks for the fron top bottom and 1 on the back. i do have performance fans on my rads too thow i have the gentle typhoons on 2 of my 4 rads. sadly there in pull and not push like there meant too. but i wanted the hd fans for looks on the outside.

 

the silverstone rave cases are some of the best air flow case around but at a cost....

 

gamer nexus dose case reviews and had an updated chart on what has the best airl flow vs nosie levels. but he dose test only with the stock fans the 90% replace anyway.

 

i wanted a 90 degree case but never did end up getting one. my build right now will have the rads on top and pull air in and out the top want even be any ware ner the cpu or gpu. althow my pc will have like 15 fans on the one side not counting my rads were there be push pull. i could do 200 i think might fit but only argb 200mm are thermaltake. althow thermal take might work on corsair have not tested that.

 

i think the misconception is that people think that when you put air in that it also need another fan to push it out. that air will evaporate but its not the case as you found out. 

 

Edited by thrasher_565
Link to comment
Share on other sites

Link to post
Share on other sites

44 minutes ago, thrasher_565 said:

ya make sense. the rad is just like a fan grill. the the holes take up space were air gets slowed down.

 

i cant seem to find it but  silverstone was testing grill vs no grill and well the no grill was better then the wire grill then the honeycomb we have now.

 

another good YouTubeer i found that test odd ideas is major hardware. like 40mm fans vs 120mm fan on a rad.

 

ya we no that big 200mm fans can turn slow and push a lot of air but not alot of case alow for a 200mm. and the 200mm dose no have stranded holes. and alot of cheap 200mm fans the blade were braking off. i only used cosair hd fans for looks for the fron top bottom and 1 on the back. i do have performance fans on my rads too thow i have the gentle typhoons on 2 of my 4 rads. sadly there in pull and not push like there meant too. but i wanted the hd fans for looks on the outside.

 

the silverstone rave cases are some of the best air flow case around but at a cost....

 

gamer nexus dose case reviews and had an updated chart on what has the best airl flow vs nosie levels. but he dose test only with the stock fans the 90% replace anyway.

 

i wanted a 90 degree case but never did end up getting one. my build right now will have the rads on top and pull air in and out the top want even be any ware ner the cpu or gpu. althow my pc will have like 15 fans on the one side not counting my rads were there be push pull. i could do 200 i think might fit but only argb 200mm are thermaltake. althow thermal take might work on corsair have not tested that.

 

i think the misconception is that people think that when you put air in that it also need another fan to push it out. that air will evaporate but its not the case as you found out. 

 

I know about the raven but but seriously silverstone makes the ugliest cases, they're not like cheap plastic low quality ugly but they're like, tasteless no designer with proper artistic background ugly.

 

Gamer's Nexus also showed how you don't need the 90 degree rotation to have Raven airflow, and that's basically what the Mac Pro does as I mentioned in OP. It's just that most people think more fans + more holes = better cooling and case manufacturers must pander to that perception  rather than what is actually cooler.

 

All you really need is a case with a triple 140 on the front with good intake, no other holes except the traditional 120 exhaust, and some special bracket that turns the unused expansion slots into a 120 fan mount. There you go that's your Raven without the 90 degree rotation.

 

Instead we get these crazy cases with holes on top, holes on bottom, holes going side ways, just tones of wasted airflow.

 

I scrolled through your build log, yea there's nothing innately more efficient about 200mm fans, in fact I'm pretty sure the best 120mm fans are more efficient at the same RPM. It's just that you need more of them for the same airflow, and that's more expensive, and takes up more space (because of wasted space in the frame and fan hub). So if you can afford to stick in 2x480 rads with 8x120 fans, that's certainly a substitute for 2x200mm fans. (if you can also somehow solve the proximity with radiator problem)

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×