Fan replacement help

Push - pull can help by say 2 or 3C temp drop.

You should set up your fans to blow in from the font, bottom and sides and out from the rear and top.

You are misunderstanding. Push/pull means having fans on both sides of a heatsink or radiator. One fan pushes through the heatsink or radiator and the fan on the opposite side is pulling. Generally, this is always done in a direction that expels air from the case. In some installation configurations however the radiator fans may be pulling air into the case.

I don't agree with this configuration but many other experienced guys (And gals) use this configuration quite often. In my opinion, based on years of work with automotive and equipment cooling systems, it's far more efficient to extract the hot air from the radiator in a direction that removes heat from the case rather than bringing it into the case, which is does if you use the radiator in an intake setup. Anyhow, regardless, that's the deal on push pull.

If you have no rear exhaust fan currently, then you need to use that fan there. It's very important to exhaust heat from the rear upper or at least top, section of the case as that is where the heat tends to gravitate. If you already have a rear fan, aside from the one in the power supply (Don't mistake that for a case fan.), then using it as a front intake (If you don't have one there) would be a good idea. If you remove the panel on the other side of the case you will be able to see exactly where all the cabling is routed to and from.

Push pull means you have one fan pushing air into the radiator and one fan on the other side of the radiator pulling air thru

Fan ------ Fan
____________
|____________|
Fan ------ Fan

As for which way to blow, lets do a little lesson in thermodynamics.

Ambient air outside your case = 25C
Air inside your case = 30C
Coolant temperature = 35C

Option 1 = Well heat rises right ? Correct. So I should blow air up and out of the case right ? Incorrect.

q = k A dT / s

where
q = heat transfer (W, J/s, Btu/s)
A = heat transfer area (m2, ft2)
k = thermal conductivity of the material (W/m.K or W/m oC, Btu/(hr oF ft2/ft))
dT = temperature difference across the material (K or oC, oF)[/b]
s = material thickness (m, ft)

Everything in bold is a constant (Let's call it C)as the radiator is the same size, thickness and material. So we can rewrite the formula as:

q = k A dT / s

So looking at those temps again.....

....with air blowing up and out of the case, we have 30C air cooling 35C water of a Delta T of 5C. Our quantity of heat removed therefore might be expressed as q = C * 5C

....with air blowing in from outside the case, we have 25C air cooling 35C water of a Delta T of 10C. Our quantity of heat removed therefore might be expressed as q = C * 10C

It should be obvious now that blowing cooler air from outside the case thru the radiator will remove twice as much heat from your CPU as blowing preheated interior case air thru the radiator and outside the case.

Does it increase case interior temps....yeah ... so what ? You bought a water cooler because you were concerned about the temperatures of your CPU. Does it matter if anything in your case is exposed to 33C air ? Will having 30C air inside the case improve anything ?

Think about a car ..... does than fan blow hot air from inside the ngine compartment out thru the radiator and outside the vehicle ? Or does it suck in cooler air, not preheated by the engine, and suck the much cooler air from outside and blow that thru the radiator ?

I can understand that, but given that other components exist inside your case, isn't it generally considered a bad idea for you to have the radiator dumping heat into the case? Even the cheapest GPUs use about 100W+ of power, and with an open-style heatsink, it will also be releasing heat into the case. Doesn't bode well for the longevity of components, especially if your case is not well ventilated or you have an overclocked CPU that eats up 120W (and maybe even more).

Also, if the radiator is sucking air out of the case as a source, then won't it also take some of the ambient heat within out as well? It'll make the air inside the case cooler.

Aren't cars designed to take in air from the outside because when you're driving (at say 30 or 60MPH), the intake grille on the front of the car allows air to just slam into and through the radiator? It's a lot more efficient, cools accordingly (the faster you drive, the hotter the engine, but also, the faster the speed of the air running through the radiator), and reduces the need for the fan to turn on. The only time it's needed is when the car is idling (and also not generating as much heat as when it is moving fast), or when it's moving at slow speeds.

PCs don't have wheels. They can't take advantage of the relative "wind" you get when you are driving.

It would spin at the fastest speed it can all the time when you plug it in directly to the PSU, thus making it potentially noisy. You can reduce the speed by getting an inline resistor to reduce the voltage being fed to the fan. A fancier option would be to get a fan controller. I've been eyeing this bad boy for some time now:
http://www.nzxt.com/product/detail/144-sentry-3-fan-controller.html

If I'm correct, Noctua fans are so high-end that their boxes come with resistors and a molex adaptor (allowing you to plug it in to the PSU). And not just one resistor, but TWO. One lowers the voltage even more than the other, so choose the speed you're comfortable with, but Noctuas are really quiet anyway.

Lets look at that ....

1. Since the case hasn't exploded, we must assume that every cu.ft. or air that comes in goes out. That's what all those grilles are for. So the heat doesn't "build up" inside the case.

2. You buy a liquid cooler because overclocking brings CPU temps above safe levels. Is anything in the case reaching unsafe levels ?

3. The CPU has a very small die area and .... if it was as big as a GPU, we wouldn't have this problem. The large surface area of the GPU gives it large surface area.... remember the equation above ? A = Area.... with so much area to dispel heat, the GPU stays cooler. For example..... my water cooled GPUS run at 39C under Furmark while putting out 300 watts each. But my CPU is at 74C only putting only 120 watts. The CPU is hampered by the fact the water is 35C but the CPU is 40C higher.... the same water is 35C at the GPU but the GPU temp is only 4C higher.... a 10:1 difference.

4. Back to Delta T .... we doubled the air / water delta T by blowing ambient air in. With the rad fans blowing out in example above, we had case temp of 30C .... switching them to blow in past 35C water might take that 25C air up to 32C maybe ? We have no coolant so it's strictly air temps and heat sink temps.... so what are these heat sinks at ? 50C lets say for the sake of argument .... that changes our delta t from 20C to 18C a far smaller differential.

So yes it does make the case cooler but nothing in that case is getting anywhere near it's thermal threshold.

My build has:

(6) fans on a 420 rads in top of case blowing in
(4) fans on a 280 rads in bot of case blowing in
(2) fans in front of case blowing in
(2) fans in side of case blowing in
(1) fans in side of case blowing out

And 13 fans worth of air blowing out thru the rear grille.

4770k @ 4.6 Ghz w/ 4.3 cache, CPU temps are at 69C
2 x 780 @ 26% OC (21% on memory), GPU temps are at 39C
Interior case temps are 26-27C w/ 23C ambient.

Yes, the motherboard automatically controls and adjusts fans plugged into it's limited number of ports.

The way you set up a fan controller is you usually first plug it into the PSU via it's own molex/SATA power connector. It then has it's own fan headers on the back of itself where you plug in the fans. It's basically a board with variable resistors that you can adjust the resistance of with the twist of a knob (or other equivalence motion).

As for the resistors, I managed to get a picture of exactly what it looks like:
http://cdn.overclock.net/4/42/820x323px-LL-42f4bdb7_FanNoctuaNF-F12LNA81ohm001.jpeg
It should say something like "Low Noise Adapter" on the tab (shown in the picture). All you have to do is plug this into the molex adapter plug or other fan header (with the male end) and then plug the fan's power cable into the female end, hence "in-line" adapter. If you feel the thick central part of the wire after some time in operation, you'll notice the cable is warm. That's the resistor encased in rubber dissipating heat. Resistors are somewhat like lightbulbs, but instead of radiating light when they consume power, they radiate heat. The heat is where all the energy that would've been powering the fan is going.

I never said it would reach it's threshold, but you do know that even a slight difference in temps can alter the length of a component's lifespan right? But hey, if it works, go ahead. I'm just following what makes sense to me.

I take it that with your math, you must have some understanding of engineering/physics. I am very well capable of understanding it, but I'm just saying. The people who are asking questions here are looking for simple understandable answers. You don't need to show every single step or quantify every single variable present. That might just confuse them, as they are often beginners or not accustomed to these concepts. I appreciate that you are putting in effort to educate them along the way though.

Ok, first of all, while it's true that you will cool the cpu more efficiently using cool air. Nobody can dispute that. But now you're taking ALL the heat you removed from the cpu and putting it right back in the case, along with the new heat that the cpu generated milliseconds later. So along with the other components in the case that will suffer due to this inefficient configuration, you've now increased the ambient temperature INSIDE the case where the cpu is and therefore increased the baseline amount of heat that now has to be removed from the cpu.

Not you, or anybody else can tell me that a cpu in an 60c environment is going to be as easy to cool as a cpu in a 45c environment. Whatever amount of heat transfer is needed to keep the cpu at 60c in a 60c ambient environment is going to exceed what is needed in a 45c environment. Period.

Now, on to your car example. Unlike a cpu cooler, an automotive cooling system has a thermostat which is there for two reasons. One is to enable it the warm up more quickly, thereby reducing startup emissions and enabling the heating system to begin warming YOU more quickly.

The other is to actually keep the engine warm (Considering a fully operational engine with no anomalous issues such as poor tuning or blown head gasket.) during operation because without the thermostat installed, most internal combustion engine cooling systems far exceed what is necessary to keep the engine at the pre-determined temperature which for most vehicles is 195-205 degrees F.

If you remove the thermostat on most vehicles, on all but the hottest of days, the engine will run at between 170-185F on average for most vehicles. So there is no need for the best possible cooling solution on an internal combustion engine because under normal operation, the cooling system already exceeds what is necessary to provide optimal operating temperature. Find a new example.

In this instance a water cooler was purchased to cool the CPU .... you can buy a Water Block for the MoB, RAM, GFX Cards, Coolers for HDs and such but this market is teeny compared to the market for CPU coolers.... why do ya think that is ?

The reason few water cool the GFX card is that, outside of cards physically modded to deliver voltage beyond specifications, there is no performance advantage to be gained. I have both air and water cooled cards here and there is no difference in the overclocks that I have gotten. I didn't water cool the cards to reduce temps, I did it to reduce noise. GFX cards run just fine at 82C and with 100s of builds out there, only twice had a a component failure due to heat. Both were optical drives in the same box (same unit failed twice) and it was in a case that the user insisted on using against my advice.

As for my background, I am a licensed professional engineer that owns a consulting firm and among the things we are involved with is cooling. We have a lot of preconceived notions that the number show don't pan out. For example:

Lotta peeps have trouble going against the heat rises thing that we all learned in 7th grade .... I assume you either own or have been in a house with a ceiling fan..... do they do anything ? Of course they do....they wouldn't sell gazillions of the things if they didn't.

Lotta peeps have trouble with not having an exhaust fan for every intake fan. Not at all necessary to have any exhaust fans. If you have 10 intake fans and no exhaust fans every cu in of air that gets into the case will find it's way out w/o any help from a fan. Look at attic fans ... when you turn on your attic fan, do you go someplace else in the house and turn on an intake fan ? Does your car have an exhaust fan under the hood ? Does your car's ventilation system have an exhaust fan ? Industrial buildings, commercial kitchens that generate a lot of heat all have exhaust fans.... not an intake fan in sight.

Lotta peeps worry about positive case pressure .... a completely meaningless concept in and of itself. The ONLY reason positive case pressure is an issue is **if** the case is equipped with inlet filters on every fan. If there's no inlet filters, the fans will do quite a fine joib of blowing dust into the case.

As y showing the people simple answers, I agree, kit's easier just to say "Blow air in" and be done with it..However, when an argument is presented using psuedo physics / thermodynamics, that sound good but which are not based in science, it's hard to dispel that with presenting the factual evidence.

As for useful life.... yes, temperature does have an effect on useful life. the question however is it going to impact teh design life. We are talking about a water cooler here.....the reason we are talking about a water cooler is that overclocking is involved, If you are concerned about useful life than don't overclock..... that extra voltage and those higher temps will shorten your CPU life.... instead of 10 years, you might get 7 or 8.

IBM issued a paper back in 1983 showing that a 10C temp increase over the design point decreased HD bearing life by 50%. But if you are below that design point, then you'll never see that reduction. Jumping 3 C from 30C to 33C is not going to have a significant impact on anything .... however if your concerned buy an extra case fan or 2.

A typical case has less than 2 Cu.Ft of space inside ....let's say your typical 80 cfm rated fan only actually outs out 40 cfm due to the filter and other restrictions..... still that fan changes that air out and puts fresh air in every 3 seconds.

No need for a new example because you made my point better than I did and you highlighted exactly why the example is is perfect. Restated:

So there is no need for the best possible case cooling solution inside a PC case because under normal and even overclocked operation, the cooling system already exceeds what is necessary to provide operating temperatures, well within desired ranges as stipulated by component manufacturers.

The operational temperatures are orders of magnitude below what the manufacturer's consider significant and that does not in any way change the fact that in both PCs and internal combustion engines it is the most efficient method.

Ok, first of all, while it's true that you will cool the cpu more efficiently using cool air. Nobody can dispute that. But now you're taking ALL the heat you removed from the cpu and putting it right back in the case, along with the new heat that the cpu generated milliseconds later. So along with the other components in the case that will suffer due to this inefficient configuration, you've now increased the ambient temperature INSIDE the case where the cpu is and therefore increased the baseline amount of heat that now has to be removed from the cpu.

Click to expand...

How do you draw this conclusion that the heat is somehow getting trapped in the case ? A typical case fan throws 80 cfrm (let's say 40 after restrictions from filters and grilles) or enough to to a complete changeover in a large case every 3 seconds.....so we have 2 fans blowing air thru a rad and let's say 3 case fans which would be low for a case using water cooling. So what we have now is all the air in the case being exhausted 100 times per minute (5 x 40cfm / 2 cu.ft). You don't think that's enough to adequately remove that heat ?

But theory is just that until proven via experiment so let's just do that and use some real numbers . I have 6 temperature sensor in my box which display temperatures on the front of my case to the 10th of a degree.



Looking now at the display ....ambient is 26.0C, the highest of the 4 coolant temp sensors is 27.1C, interior case temps are fluttering between 26.2 and 26.3. Under stress testing at 4.6 Ghz, at 23C ambient my max coolant temp was 25.3 .

Ran Furmark while I stopped typing to take a phone call just now and recorded the following after 20 minutes:

GPU Temp = 44C (Both)
CPU Core Temps = 48-51C
Ambient Air = 26.1
Case Interior Temp = 29.9
GPU Rad In = 36.2
GPU Rad In = 34.9
CPU Rad In = 35.3
CPU Rad In = 34.7

I used only the push fans for this test .... pull fans are out at the moment. So this should exaggerate the results

So with 5 fans blowing air in thru radiators (3 on top, 2 on bottom in push / pull) with coolant temps of around 35C, I have an interior case temp of 29.9 . Where's the problem ? With ambient air at 26.1, just how much ya think things would improve with those 3 top fans blowing out ? Let's assume that reversing the fans reduced the delta T from ambient to case air by 50% ? Just how much extra life do you figure I will get with interior case temps at 28.2 instead of 29.9 ?

In summer times, ambient can climb 10C above the other 3 seasons .... so what are we to do ? If you are concerned about interior case air being 3.8 degrees above ambient at 29.9, should we not use our computers in summer months because it's 86F outside and that's hotter than the 29.9C case interior temp ? ?

Not you, or anybody else can tell me that a cpu in an 60c environment is going to be as easy to cool as a cpu in a 45c environment. Whatever amount of heat transfer is needed to keep the cpu at 60c in a 60c ambient environment is going to exceed what is needed in a 45c environment. Period.

Click to expand...

Where is this CPU that it sees ambient air temps of 113 - 140F ? Again, you make the case for me

1. The CPU is not in that environment..... the CPU is in no way impacted by "the environment" because being water cooled and therefore is in now way directly affected by interior case air temp.
2. The CPU temp, being water cooled is affected by one thing and only one thing and that is the coolant temperature. And as you have already confirmed, the water will be cooler with air blowing in.

Now, on to your car example. Unlike a cpu cooler, an automotive cooling system has a thermostat which is there for two reasons.

Click to expand...

Again, I don't know how you work your boxes but my PC, well every PC does in fact have a thermostat. It's that "thermostat" which makes your fans spin up and spin down in response to temperature.

most internal combustion engine cooling systems far exceed what is necessary to keep the engine at the pre-determined temperature which for most vehicles is 195-205 degrees F.

Click to expand...

Similarly, most case cooling systems, far exceed what is necessary to keep the case interior and components at less than the pre-determined temperature which for most cases with water cooling will be in the neighborhood of 5 above ambient ....I can't imagine someone putting together a build so poorly that it exceeds 15C above ambient

To go your way, we must assume the manufacturer's of the very coolers we are talking about don't know anything about cooling as the installation manuals are quite clear on the topic.

http://www.corsair.com/~/media/Corsair/download-files/manuals/49-001175_rev_AA_H110_QSG.pdf

Attach the 280mm radiator and fans as shown. For the best cooling performance we recommend
mounting the fans as an air intake.

Click to expand...

I never said anything was "trapped". I simply can't imagine anybody actually believing that a system designed to remove heat can function MORE efficiently by adding unnecessary additional heat back into an equation already designed to only be adequate for the task.

As I said in the first place, will it work, yes, do people do it, yes, is it the most efficient way, no. All your fancy calculations don't alter the fact that the cooler you keep the inside of the case, the better the cpu cooler will perform, and adding heat back into the case does not help to keep it cooler. Besides, unlike an automobile, a cpu does not function better when warm. Heh.

Well if the heat ain't trapped, then it's not in there and so there's nothing to worry about....the entire case volume is flushed 100 times a minute ... that's just over once every half second.

I just ran another CPU stress test .... moved the case air temp sensor to 1/2" from the radiator so as to pick up the air temperature exiting the Radiator.

Ambient Air = 29.6
Coolant Temperature = 31.2 C
Air exiting radiator = 30.6
CPU Max Core (4.5 Ghz) Temps = 65

Now we ran Furmark again....this time w/o side fans near storage devices to exaggerate the results

Idle / Max
Fan Speed = Off / 850 rpm
Ambient Air = 29.6 / 29.6
Max Coolant Temperature = 30.8 / 38.0 C
Max Air Temp exiting radiator = None / 35.2
Interior case temperature = 33.1 /
GPU Max Core Temps = 28.9 / 45.0
SSD1 Temp = 35 / 37
SSD2 Temp = 38 / 40
HD1 Temp = 40 / 43
HD2 Temp = 41 / 46
MoBo Temp = 34 / 43

Recommended Temp Limits:
HD = 60C
SSD = 0 - 70C operating / -55 - 95C non operating

So obviosly, the manufacturers of those devices have no concerns.

Believe what you want, as I said above the proof is in the pudding and the pudding is the test results. Argue the theory all ya want if you don't understand all the "fancy calculations" but "the laws of physics" are called "laws" for a reason. But, setting that aside for a moment, what's your explanation for the actual test results ? Let's summarize:

1. All the people who actually design the things ..... you think they all got it completely wrong, since they all instruct the user to "install their units blowing in" ?

2. Swiftech, a major player in the custom water cooling field for 20 years has their units blowing in, you can't install the 220-x any other way....they got it wrong too.

http://tpucdn.com/reviews/Swiftech/H220-X/images/contents3_small.jpg

3. The test results confirm the "fancy calculations", tho they're actually rather elementary. You haven't explained those away, just ignored them ... is the data from the "fancy" thermal sensors somehow invalid ? .... perhaps because, to borrow from the movie "My Cousin Vinny" I didn't have each of the sensors calibrated that afternoon by the National Bureau of Standards.

Speaking of My Cousin Vinny.....

https://www.youtube.com/watch?v=ZZkbtP-t_D8

The laws of physics don't cease to exist.....unless perhaps, the PC is in that guy's kitchen,

Yes a PC doesn't function any better when it's warm, doesn't function any worse either.... ....as long as you keep all temps within manufacturer's recommended ranges, all will be fine.

But your CPU can function at a higher OC when the CPU has twice as much heat removed. This will be determined by the coolant temperature which you agree will be lower with air blowing in.

As to your other components you have not yet explained how their performance or life will be negatively impacted by a 2C temperature rise ..... orders of magnitude below any published thermal limits.

So, according to this statement, there really shouldn't be any problem using the radiator in an exhaust configuration because the air flowing through the case should be fairly cool, since it's been flushed just over once every half second.

Looking at it another way ....

Heat Summary (stock settings .... overclock only exaggerates the effect):
CPU = 84 watts
GPU = 254 x 2 = 508 watts
MoBo = 40 watts
Water Pump = 23 watts
RAM = 10 watts
HDs = 10 watts
SSDs = 6 watts

Total 681 watts ..... produces a case interior temp of 33.8C .... taking the CPU heat (84 watts) out of the case reduces that to 597. So we could reduce the heat to 88% of that (597 / 681) by blowing Radiator out of the case.

Existing Delta T (Intake) = 4.7C (33.8 - 29.1)

88% of 4.7 = 4.1 C

Net decrease in case temp with rad fans blowing out = 4.7C - 4.1C = 0.6C
Interior case Temp w/ Exhaust = 29.1 + 4.1 = 33.2

Now let's look at the effect of cooling on the CPU.

Coolant = 38C
Interior Case Air = 33.2C
Ambient Air = 29.1

Water to Air Delta T Intake = 8.9C
Water to Air Delta T Exhaust = 4.1C

Intake provides 2.6 (8.9 / 4.1) times the heat withdrawal from the CPU and ya traded that for an increase in Case Temps of just 0.6C

Think about it ... do you really wanna take 597 watts of heat and push most of it thru a radiator that's designed to provide maybe 180 watts of cooling ?

Nevertheless, ....Looks like we are now in complete agreement:

1. "there really shouldn't be any problem using the radiator in an exhaust configuration because the air flowing through the case should be fairly cool, since it's been flushed just over once every half second." I absolutely agree, all of the air cooled components will perfectly fine as they are operating at temperature levels far below manufacturer's recommended limits in either case. Case Air temps won't vary by more than 1 - 2 C either way....and this difference will have no measurable effect on any air cooled component.

2. "it's true that you will cool the cpu more efficiently using cool air. Nobody can dispute that." Again, I absolutely agree. The coolant temperature is the only thing that affects CPU cooling and, using today's numbers .... 29.1C, (with a delta T of 8.9) will provide more than twice cooling for the 38.0C coolant than the 33.2C interior case (4.1 Delta T) .

I think the best solution is a divorced radiator, therefore, attention to all case manufacturers, you need to now start including a separate compartment behind/under the motherboard but separated by a divider panel from the rest of the case internals with it's own source of cool airflow from outside the case. This will provide more room for cable management, will allow for no case heat to alter the efficiency and best of all, it will allow the use of the top case vents to be used for exhausting case heat only, as it was intended to be used by original design. Now get on it.

They have been around quite a while......Cosmos II for example, does that but it is H U G E (28" High) and some of the case labs also

http://www.caselabs-store.com/magnum-sth10/

But that's almost 3 feet high compared to the 20" most peeps prefer. Few are willing to spend $500 on a case and if they are ..... 32+" gets us in trouble with wifie

Most gamers won''t spend the $100 for a 4790k over a 4790k despite gaining 0.5 Ghz..... I think spending an extra $100 to drop case temps 1C will be an even harder sell.

Oh here we go again. I didn't say to put it above the motherboard at the top of the case. Another compartment maybe 4 or 5 additional inches in width could be added to the right side of the case (For conventional designed cases.) with the fans blowing out through the panel, mounted flat against the panel, with it's own intake vents on the front or back of the case, or both. So it would be four or five inches wider but it wouldn't have to be any taller. Cooler manufacturers would need to start making longer lines optional though or use a custom loop. Anybody that worried about increased performance and liquid coolers shouldn't cry about the case being a bit bigger.

I'm going to build it, and test it, and see what happens. I'm interested now.

Ya mean like this



http://www.caselabs-store.com/magnum-th10a/

Yep, but that's overkill. It could be done in a cost effective manner that shouldn't increase case price by more than maybe an extra 20-50 dollars. Actually, a kit that could be bolted on to the panel on the backside of any case could be done as well. And by backside, I mean the right side. Heh.