Fans, cooling, and air movement - an opinion

[Felger Carbon]

Air moves because of a pressure difference. No pressure difference, no air movement. The velocity of the air is proportional to the square of the pressure because energy is conserved, and the kinetic energy of air molecules is proportional to the square of the molecules' velocity.

CFM is proportional to the velocity of the air times the cross-sectional area of the moving air.

A given fan's CFM is linearly proportional to RPM. It follows that the fan's pressure is proportional to the square of its RPM, and that is in fact the case.

The design of present-day 7-blade 25.4mm-thick fans (80mm to 140mm) is pretty much standardized. There is little difference between brands and models with respect to the fan casing and the fan blades, including the fan pitch. However, some differences sometimes exist. The Noctua 120mm fan famously has a much higher pitch than usual. Less commonly known is that Yate Loon D12SL-12 fans have a noticably lower pitch than, say, the Global Win NCB fan. Also, the SPCR favorite 80mm fan, the Nexus, has a much lower than average pitch - it takes 870RPM to achieve 10CFM, while most of the others only require 750RPM.

Because of this standardization, the common way to compare fan noise between two fans of the same size is to test them at the same RPM. At the same RPM, the lowest-pitch fan will almost always push less air, producing less "whoosh" noise - and win! This has a great deal to do with the favorable "common knowledge" about the Nexus 80mm and the YL D12SL-12.

In SPCR's review of 80mm fans, the 12V Nexus and Papst measure 22 and 23 CFM, respectively, and 20 and 21dBA, also respectively. Guess what? 60*LOG10(23/22) = 1.16dBA. Of course the Papst is noisier! There's more "whoosh" noise! (The Papst also has lower-than-average pitch, but not as low as the Nexus.)
------------------------------------

By happenstance, a great learning experience

Not too long ago, I found myself with two very different 220mm x 30mm fans. One had 7 nicely curved blades, the other 13 utilitarian paddle-blades. A casual glance revealed that the 13-blade fan had much lower pitch than the 7-blade. I decided to do a noise/cooling comparison.

First, the hand-held fan test: adjusting them with a Sunbeam Rheostat fan controller, I set the fans for equal noise (by ear) at a given distance. Then I allowed the fans to blow into my face. Clearly, the 7-blade was pushing more air at the same noise. A lot more. I don't claim superiority for the biological instrumentation used (me); I'm saying the difference was so large the result was unmistakable.

Next, I mounted the fans on the side panels (two of them) of a big-fan case/computer, with no other changes to the computer. I could easily and quickly swap side panels. If I adjusted for constant noise, the exhaust air of the 7-blade fan was hotter WRT the ambient (a higher delta). If I adjusted for constant exhaust temperature, the 7-blade fan was louder. Again, the difference was so large the result was unmistakable.

Conclusion: in open air, big win for the high-pitch fan. In a real-world computer case, big win for the low-pitch fan. This is based on noise/cooling.

Has anybody noticed that the Noctua high-pitch fan has had widely varying reports on its performance? Do you suppose it performs better in a less-restricted environment, and falls down where air movement is not so free? I'm just asking the question, I don't know the answer.
--------------------------------------------

Fans come in different sizes. What's the best size for a cooling fan? Simple. The fan size should ideally match the size of the object to be cooled. Conversely, the size of the object to be cooled should ideally match the size of the fan. This is why the best HSFs have evolved to use 120mm fans.

Most of you know that I've been doing a lot of experimenting with 220mm fans. They do a fabulously great job of quietly cooling the case and the PSU - the PSUs in my big-fan cases are all fanless. I believed they would also do a good job cooling heat sinks passively. I was wrong. 220mm fans do a low-average to poor job of passively cooling heatsinks because, while they push a lot of CFM very quietly, the air velocity is too low to do a good job of cooling the heatsinks. The computer I'm typing this on is a big-fan case. The heatsink is an XP-90 being cooled by a very low-RPM 92mm fan. It works very well and very quietly. I don't need better CPU cooling because I never stress the CPU (except deliberately for testing via BurnK7).
-----------------------------------

Fan makers test their fans in an open-air environment (an acoustic testing facility is open-air as far as the fan is concerned). The standard 7-blade 25.4mm-thick configuration has evolved to provide optimum performance in this open-air environment.

And yet, in quiet-fan circles (meaning SPCR) the fans that are highly reqarded often turn out to be the ones with low pitch...
-----------------------------------

About using one fan to cool everything: since everything is not the same size, this is not a good idea. I've done a lot of experimenting using non-gaming, non-overclocked, non-Prescott computers. I'm using a 220mm fan to cool the case, mobo, and PSU; a 100mm fan to cool the 101.6mm-wide "3.5" HDD, and a 92mm fan to cool the XP-90/CPU. If I wuz a magician and could specify a single fan (X dia, Y RPM, Z pitch) I could not come up with a single design to cool all parts of my computer both well and quietly.

In closing, I'd like to call your attention to the word "opinion" in the title of this posting. What's your opinion?

[Aris]

my opinion?

.....

whats the overall point of the post? Are these just ramblings, or was there an overall statement or question?

my opinion is that the wall of text was too high for me to scale.

[jaganath]

Are these just ramblings, or was there an overall statement

there were several statements or conclusions, which when added together give a better insight into what size and type of fans will yield the best results when cooling a PC:

Conclusion: in open air, big win for the high-pitch fan. In a real-world computer case, big win for the low-pitch fan.

What's the best size for a cooling fan? Simple. The fan size should ideally match the size of the object to be cooled.

About using one fan to cool everything: since everything is not the same size, this is not a good idea.

I've found much the same thing; even though I've read pretty much everything that's been written on SPCR since I joined, my present PC has 3 fans (inlcuding PSU fan), even though the CPU is undervolted and I only use onboard graphics.

I do wonder if better results can be obtained, say, with a 120mm fan and a 120-to-80mm adapter (I realise it would have to be a fan with good pressure characteristic to deal with the backpressure of the adapter).

Felger, another reason for the 13-blade fan's better pressure is the increased blade count; this makes the fandisc more "solid" and hence less recirculation.

[skiddy]

Quote: Has anybody noticed that the Noctua high-pitch fan has had widely varying reports on its performance? Do you suppose it performs better in a less-restricted environment, and falls down where air movement is not so free? I'm just asking the question, I don't know the answer.
--------------------------------------------


I found that Noctua 800 rpm fans (2) mounted on a high-fin-density Scythe Infinity worked quite nicely.

I found that Panaflo 120 x 38mm, 1700 rpm fans (3) mounted on a TR IFX-14 work better. If I could not rpm-control the Pannys, I'd be using the Noctuas, and reduce my overclock to control temps.

From the fans I've experimented with, I don't think there's enough difference in actual cooling capacity from any one brand, to any other brand, to matter a whit, unless you are overclocking to the maximum possible. Since I -am- pushing my OC quite hard, I went to the high-flow fans. But in actual terms, that only allowed me to go from 3.555 GHz to 3.690 GHz. Does any -sane- person really care that much?

[Fayd]

heatsinks rely on pressure, not CFM. it's no suprise that a huge fan wont passively cool a CPU very well.

exhaust and intake fans work on CFM, not air pressure.

imo, from what i've seen, a fan like a papst will work better on the hsf, where a fan like the nexus would work better on exhaust/intake duty.

but imo, the difference is so small that it's down to personal preference.

and i like every fan in a system to look the same.

[jessekopelman]

heatsinks rely on pressure, not CFM.

Not true. It's always the CFM that does the cooling (as long as the air isn't warmer than the thing to be cooled). It's just that a low pressure fan may not make its rated CFM when obstructed by a heatsink. The reason heatsinks with good spacing work so well with low pressure fans is that they obstruct the fan less and it can create close to its rated CFM. An interesting point and one MikeC often brings up is that there is a definite point of diminishing returns for CFM vs. cooling. If you have a heatsink with good spacing (like Ninja) even a very low pressure fan has no trouble generating sufficient CFM at relatively low RPM.

[seemingly.random]

Stimulating.

What's the best size for a cooling fan? Simple. The fan size should ideally match the size of the object to be cooled.

If I recall correctly, the fans on chillers for big buildings are the size of the chiller.

My opinion? never thought about it from this angle but I'm glad you and others have. I appreciate those who dabble in esoterica - eg. melting sand and running current through it. Thanks for sharing. I'm assuming there are other gems on this forum I've yet to read.

I do wonder if better results can be obtained, say, with a 120mm fan and a 120-to-80mm adapter

I've wondered about the same thing since first seeing these adapters at jab-tech but never pursued it thinking it would create more noise. I assume these adapters are intended for putting a smaller fan on a larger opening - just the opposite of what makes sense.

[Bluefront]

IMHO.....and I've tried these adapters several times. It comes down to CFM. A 120mm fan can draw sufficient air through a 120/80 adapter without making more noise, but only if the rpms/CFM is kept low. The adapter itself allows a small space between the fan and the opening, which helps the process.

The availability of numerous quiet 120mm fans makes these adapters useful. A better setup is a 120/92 adapter. Most 80mm case openings can be enlarged to 92mm. Using a 120/92 adapter on the enlarged opening can give you the ability to run a 120mm fan at the max rpm. A Nexus 120 for example, can run at 12v attached to a 120/92 adapter without much/any loss of airflow. Try it yourself....those adapters are rather cheap for experimentation.

Oh.....if you turn the fan around and try to blow through the smaller opening, you will hear more noise, and the CFM will be reduced, no matter the voltage. With this setup, back-pressure is the killer. IMHO.

[seemingly.random]

It comes down to CFM.

I'm feeling dense in that I don't understand how CFM can be decoupled from pressure - must have missed something in school. When I think of a fan blowing on something I think of the pressure of the air at the something. Are there multiple points at which pressure is considered?

Maybe a different scenario will help. I used to live next to a house that had a powerful "whole house" fan. It was located in a hallway ceiling blowing/sucking straight up into the attic. It was a little intimidating being right underneath it but their a/c was on less than other's. Things could change dramatically when windows in different rooms were opened/closed. How would pressure be considered/described in this situation?

And, since I'm in a confessional mood, I've seen the phrase 'static pressure' used. Is there another kind of pressure? 'dynamic pressure'?

Try it yourself....

ok.

[seemingly.random]

As far as all this applies to quiet pc cooling, it seems that there are two camps as far as cases go - one with and the other without vents in the sides. At first blush it would seem that the multitude of vents is going to offer better cooling but ultimately it seems like it would create a confused and possibly contradictory airflow. All these vents also allow any fan noise inside to travel outside easier - one reason I've discounted the new, interesting looking Silverstone uATX. I've always felt that the best coolers are those that have a 'path' of airflow - a distinct inflow and outflow. This is why I think my little nsk3300 is a better cooler than some of the LianLi's, CMs, et.al. This also explains why some components run hotter outside of a case.

[jhhoffma]

I love Felger's so-called "ramblings". I think most of the time they are things that we've all thought about on our own, but never articulated here. They make me and everyone else here think about it, and more insight is never a bad idea.

He has an opinion or hypothesis and he's looking for info to support/refute it. That's how things work in the scientific community. And yes, sports fans, silence (acoustics) is a science.

Then again, I sometimes think, "Wow, this is guy who doesn't have enough to do at home!"

[klingenfelter]

If Aris finds big walls of text a problem he should not try to climb them. I think Carbon’s Posts have been very helpful to me. I use two 250 mm fans to cool my HT rig. His and other comments on these fans turned me onto them.

viewtopic.php?t=42580

I have a thought on the optimal fan size for an application. I hope I've added to this discussion.

Fans come in different sizes. What's the best size for a cooling fan? Simple. The fan size should ideally match the size of the object to be cooled. Conversely, the size of the object to be cooled should ideally match the size of the fan. This is why the best HSFs have evolved to use 120mm fans.

I would think that the optimal case fan size would be the biggest, and slowest you can use considering aesthetics and physical limitations. I would concur with the low pitch higher blade comment. The fans I use are 13 blade low pitch units. They move 105 CFM in free air (210 CFM for the pair). There are several very restrictive choke points in the equipment cabinet they are ventilating. Two of them are ¼ the fan cross sectional area. Yet they still move 145 CFM (low pitch, high number of blades).

As to high CFM not cooling heat sinks, it’s true that heat sink cooling is improved by high velocity air movement between the heat sink fins. However, no fan over several inches away from the heat sink can contribute to much velocity between the fins. If the fins are restrictive the air can easily pass around the heat sink. The fan, CPU fan, impinging directly on the fins can only do this. So the case fans can only contribute CFM, that is quickly replacing heated air in the case with room air, so the bigger the better. The exception to this would be if the case is baffled to force case fan air through the CPU heat sink, but this would be a significant choke point and would need to traded against case air replacement.

I tried forcing the air driven by the two big 250 mm cabinet fans through the CPU heat sink, with cardboard baffles put in the HTPC case. The result was the big fans nearly stalled (> less than 50 CFM). Tried opening up the baffle and found that the ventilation system works best with no baffling. The whole rig got hotter with the baffles. But the CPU temp was closer to the case air temperature.

If I stress test the CPU, the CPU fan comes on albeit slowly. Playing music though does not turn on the CPU fan, so all is well.

These 250 mm fans running at full 12 VDC can be barely heard with the ear a few feet from the exhaust, inaudible in the listening locations, serendipity.

[jaganath]

I've seen the phrase 'static pressure' used. Is there another kind of pressure? 'dynamic pressure'?

http://www.aerospaceweb.org/question/ae ... 0025.shtml

the quantity ½rV2 is important enough that it has been defined as its own quantity called the dynamic pressure, or q. It is easiest to think of this term as the pressure measured due to the physical movement of the air flow as it speeds past a body in motion. For example, think of sticking your hand out of a car window while you are driving down the highway. The faster you go, the more pressure you feel pushing against your hand. The keen observer will note the similarity of q to the definition for kinetic energy (KE). In fact, q is simply the kinetic energy of a unit mass of air.

so it is the force resulting from the kinetic energy of the air, hence 'dynamic'.

for reference, here is the definition for static pressure:
http://en.wikipedia.org/wiki/Static_pressure

For fluids in motion the term static pressure is still applicable (in particular with regard to external flows), and refers strictly to the pressure in the fluid far upstream of any object immersed into it, i.e. the pressure of the undisturbed flow, or free-stream pressure... On the other hand, when the fluid comes in proximity to a body, its pressure deviates from the free-stream value and should no longer be referred to as static pressure, which is a common mistake. Such quantity should be called simply pressure.

so it seems like maybe we are using the term "static pressure" incorrectly.

Are there multiple points at which pressure is considered?

yes, there is the free-stream pressure as mentioned above, stagnation pressure which is what the pressure would be if the fluid was brought to a complete stop, and lots of others I'm sure.

also, every fan has a characteristic curve called a PQ curve which describes the flowrate the fan is capable of at various pressures. The flow will be at a maximum when pressure is zero, and the pressure will be at a maximum when the flow is zero.
http://www.comairrotron.com/airflow_note.shtml

[seemingly.random]

en.wikipedia.org/wiki/Static_pressure

Thanks for starting my homework for me... It'll take a while before I grok it. First, I'll have to get used to thinking of air as a fluid.

The flow will be at a maximum when pressure is zero, and the pressure will be at a maximum when the flow is zero.

So they seem to be coupled inversely.

With this and Felger's observations above, I now get what the turmoil was/is with the Noctua. Seems like fans with high pitched blades (and quiet motors) would be the ticket for pc cases with many vents.

-------

Hopefully not overdoing this but I think everyone should be encouraged to climb the "big walls of text" out there. With effort, understanding can be most gratifying even if it comes late. Many times when concentrating on one thing, illumination has come about something else apparently unrelated.

[seemingly.random]

Couldn't resist...

so it seems like maybe we are using the term "static pressure" incorrectly.

Looks like the folks (at least marketing) at Nexus needs some [re]education.

http://www.nexustek.nl/axm-8200_am2-rea ... cooler.htm

"... The RPM has been set to a maximum of quiet 1500 RPM speed, yet with enough CFM and static pressure to give your CPU all the necessary cooling it requires. ..."

The product looks interesting though - competing directly with the AC Freezer. I wonder about effectiveness and price.