There is a reason why serious water cooling enthusiasts exhaust heat from radiators outside the system case.
No... that's the reason misinformed system builders exhaust air outside their case.
1st law of water cooling - Don't mix metals
2nd law of water cooling - Rad fans always blow in
Speaking from the perspective of having taught college level fluids and thermodynamics, and have been a consulting engineer in the field for 40+ years, simple "arithmetic" is all that is needed here. Let's "Do the math":
Assumed Temps
Ambient Air = 23C
Case Interior Air = 28C
Coolant Temperature = 33C
Cooling Capacity is proportional to Delta T:
Using Ambient Air @ 23C .... Delta T = (33- 23) = 10C
Using Interior Case Air @ 28C .... Delta T = (28- 23) = 5C
Clearly, with those temps, radiator cooling with ambient air is twice as effective as interior air, you simply can't argue with the math. .... which is also why engine cooling in emergency generator sets use ambient air for radiator cooling. Change the numbers to whatever you want, but the interior case air will ALWAYS be hotter than ambient case air.
The argument I get about that is "what about other components" ... well if you are doing custom cooling, your GFX cards are water cooled too. Three questions:
1. What component inside your PC are you concerned about ? I have yet to see a product manual that warns me about the product being harmed or performance impacted by 28C. Looking at my temperature display
Ambient Air = 22.5 / Interior case Air = 23.9
Rad 1 in = 26.1 / rad 1 out = 26.9
Rad 2 in 24.8 / Rad 2 Out 25.4
Interior Case Air (Stress test) has never topped + 2.9 - 3.0 over ambient with a Delta T of 8.4C
You have your CPU and possibly GPU water cooled because a) it would exceed desired thermal limits at your OC or b) to reduce noise. What other PC components are in this situation ?
2. If you are concerned about it still, why aren't you using special methods to cool it ? No MoBo water block ? ... RAM blocks ? ... SSD Block why not ? Answer to rhetorical question ... because if you have the room for a custom loop,
nothing is affected by interior case temps.
3. Why to the people who get paid to design, market and support PC Water cooling components write this in their manuals ?
"For best cooling performance, we recommend mounting the fans as an air intake to your PC."
The corollary of this is (since we only have 2 options here) :
"For worst cooling performance, we recommend mounting the fans as an air exhaust from your PC."
Here's what we use to evaluate cooling performance:
(3) x 140mm x 45mm Top radiator
(2) x 140mm x 60mm Top radiator
(10) Radiator fans (Up to) 0 - 1250 rpm
(6) case fans (Up to) as above
Twin 35x2 Variable Speed 52 watt Pump / w/ heat sink and fan (0 - 4500 rpm)
(4) Thermal Sensors (in and out of each rad) accurate to 0.1C
(2) Thermal Sensors (ambient and inside case air) accurate to 0.1C
(3) fan PCBs to control each set of rad fans and case fans independently
(6) channel temperature display (0.1C accuracy)
(1) Chauvet Fog Machine
(1) EK CPU Waterblock
(2) EK GPU Water Blocks
(1) MoBo Water Block
Another fail to this logic is that few PC cases provide for enough fan mounts to maintain proper air balance. Everyone agrees that intakes are generally front and bottom and exhaust is rear ... why is the rear fan never an intake ? Because that's where our 250 watt GFX card(s) and 750 watt PSU exhaust their hot exhaust and you don't want to suck that back in. Again,
you don't want to suck that back in. So lets look at that:
Assume typical /popular Case Fan Mounts:
Top (2)
Front (2)
Rear (1)
Assumed air inlet restriction from moderately dirty dust filters = 20%
Fan CFM for all fans = 1.00 EF (equivalent fan)
Fan CFM for intake fans = 1 EF - 20% = 0.80 EF (equivalent fan)
Scenario A - Top fans on AIO as exhausts
Top (2) - Exhaust from AIO = 2 x 1.00 EF = 2.00
Front (2) - Intake thru Air Filters = 2 x 0.80 EF = 1.60 EFF
Rear (1) i Exhaust Fan = 1 x 1.00 EF = 1.00 EF
So we have 1.60 EF coming in and 3.00 EF going out leaving a deficit of 1.4 EF. Since the case doesn't implode, air must be getting in somewhere ... and where is the path of least resistance ? ... or where are the largest openings in your case ? They are the rear grille and vented slot covers ... what do we know about the air in back of the case ? ... the place between case and wall that your exhaust card and PSU push the hot air to ... yes, that's where all the hot exhaust from your 750 watt PSU and 250 watt GFX card(s) is going.
You can test this, as we do, with a very small investment ... pick up a fog machine cheap from Amazon for $35 and point the exhaust at the back of your case with this set up ... the case interior is now full of fog. Look for the logic here .... "In order to cool our 150 watt 9900K with an AIO, one has to use air preheated by a 300 watt MSI 2080 Ti and a 750 watt PSU ... that just doesn't make sense. It's like trying to cool you restaurant seating area with exhaust from your kitchen.
Scenario B - Top fans on AIO as intakes
Top (2) - Intake from AIO = 2 x 1.00 EF = 2.00
Front (2) - Intake thru Air Filters = 2 x 0.80 EF = 1.60 EFF
Rear (1) i Exhaust Fan = 1 x 1.00 EF = 1.00 EF
Here we have 3.60 EF coming in and 1.00 EF going out .... that's 2.6 worth of EF going out thru the rear case grill and vented slot covers. At that rate, the case is turning air over > twice per second ... no hot exhaust being recycled. There is a mindset in PC cooling that there is an actual need to even have both intake and exhaust fans. Where does this mindset come from ? ... you travel in a car ... fans blow air into the interior ... how many exhaust fans ya got ? You have an fan in engine compartment it pulls air across the radiator ... how many exhaust fans ? Kitchen Exhaust fan ... got any intakes for that ? Attic fan ... how many for that ? In a bedroom if ya wanna cool it down and no AC, at night putting a fan in one window will cool the room quickly with a 2nd window open .... why don't ya need a 2nd fan ? Why doesn't it matter whether then fan is intake or exhaust ?
Ya can only roll y eyes at the youtubers who have tested this and come to false conclusion because they test with the side panel off and / or have the case in it's typical position on or under desk against a wall. ... in which case they are not using interior case air but ambient air.
No other enclosure cooling is designed this way ... power plants, power panels, gen sets ... When we design a restaurant kitchen we collect the hot air and smoke with a hood and exhaust the air ... there are no intake fans ... we do have intake vents , often with heat exchangers ... but the negative pressure created by the fan sucks fresh air in thru the intake vent. Ever walk into a deli or pizzeria and the door is hard to open but once you "break the seal", it's fine ... thats a poorly designed air system ... exhast fan and no intake vent. Those holes in your rear case grille are an important part of your case's cooling design ... don't let them go to waste. Yes, we were all taught in 8th grade that hot air rises ... but not when you sitting in the kitchen or living room below a ceiling fan.
And that's the problem w/ Linus' review.... as usual, I don't see that Linus has made his point ...
1. Performance is gimped by having no access to cooler air source...outside ambient air.
2. They are using Delta fans ... are we talking 3,000 rpm, 4,000 ... 7,160 rpm fans ?
3. The air coolers he compared them to are using 1200 - 1500 rpm fans / Phoenix is up to 2,200 .... test is a fail, not apples and apples.
4. Unit 1/3 the size w/ Copper 5 - 10 degrees ... that's "Duhlusional"
As to apples and apples testing ... yes the Noctua does match or beat 360mm rads .... And the Scyth Fuma beats the NocPut them at the same fan rpm / noise level and.... (23:45 mark)
https: //www.youtube.com/watch?v=PYKdKVxbnp8
Noctua NH-D15 - 70C OC Load / 31C OC idle / 33 dbA
Thermaltake BigWater 3 - 70C OC Load / 33C OC idle / 40 dbA
The Noc ties under load, 2C cooler at idle and the AIO is 1.62 times as loud.
Fuma 2 @ cools better than (2) 360,, AIOs, ties another and loses only Swiftech all copper 360 by 2C while the later is 1.62 times louder and the EK by 5C which is 2.15 times louder