What you want is an Arctic Fan 12. That is the only series of readily available "reverse" fans that I am aware of. They're designed to be reasonably quiet, and the hub would be facing upwards as you desire. My only question is whether it can deliver adequate airflow. An earlier, 80mm generation of the fan didn't do so well in that regard (see the 80mm fan round up). The bearing is FDB, and Arctic Cooling is confident enough to give it a 6 year warranty.

Thanks for all your answers.


Has anyone tested a Sharkoon Silent Eagle 1000 in horizontal position (head down if possible) ? Has anyone compared it to a Noctua 800 in an enclosure where some air pressure needs to be delivered ? The folks at Madshrimps tend to prove that the Sharkoon would be better in that case :
http://www.madshrimps.be/?action=getarticle&articID=516 (see the graph on page 3).

Thanks.

I can see there is still some confusion about bearings and mounting orientations. Here is a link to Cooling-Masters.com where they review the noctua fans. I've also included some of the diagrams from the review.

Cooling-Masters Bearing Diagrams




Sleeve bearing from the Noctua - primarily a sleeve bearing with magnets that supposedly stabilize the shaft within the lubricant filled sleeve. When vertically mounted, the weight of the blade assembly is supported along the length of the shaft by the sleeve. When horizontally mounted, the blade assembly is supported by clips or washers. The axis/bearing contact is illustrated in red.




Typical double ball bearing - the bearing contact points remain the same regardless of orientation.




Maglev bearing - I believe this bearing is similar to the globalwin ncb with a single ceramic bearing supporting the shaft, but it's not clear what is supporting the other end. This would explain why some have noticed a decrease in rpm when orientation is shifted.




S-Flex bearing - This looks like a simple sleeve bearing with some magnets that are claimed to stabilize the rotor inside the sleeve and extend the lifespan of the fan. The spindle or sleeve is probably grooved to prevent the pooling of lubricant at one end or the other. In this sense, it is an FDB or fluid dynamic bearing, but quite different from the FDBs used in hard drives. In a horizontal orientation, it seems to have the same limitations as an ordinary sleeve bearing.

There are several US retailers now. Check the Noctua page. I ordered two 800s from http://www.moddersmart.com/

Apologies if this has been asked before, but what causes the "ticking" that's audible in some of the fans recordings? For example, the Antec Tri-Cool at 5V appears to have some sort of "ticking" in the mp3 recording, but the two I have in my P180b seem to have no such noise at 5V.

Yeah, and they now have the 1200s at Jab-tech.

http://www.jab-tech.com/Noctua-120mm-Fa ... -3629.html

Hello silencers! I´ve been meaking some investments these days because I felt that something was not right with noctua fans.

Well, I think I can afirm that Noctua fans move a good amount of air when they´re alone over a table or sth like that, but not when they´re mounted in a computer, because these fans don´t make almost any pression compared to nexus or similar. This lack of presion makes them move less air than others at the same rpm.

To prove this I´ve made a comparison betwen a nexus at 7v and a noctua at 5v when they´re suposed to move the same amount of air.

this are my conclusions, maybe I´ve done something wrong, I´d like someone else to try the same experiment, maybe devon or someone with expeirience, don´t know, but I think that way of measure cfm is not valid for the use of the fans in real life; use in heatsinks, and cases where air find obstacles that force the fan to push air not only to move it from noe side of a foam to another.

Nexus is running always at 7v (670 rpm and 27 cfm) and noctua at 5v (640 rpm and 30 cfm) noctua should win due to it´s better airflow. I have some pics here.

this is the noctua at 5v in a noctua nh-u12



this one, the nexus at 7v in THE SAME computer everything is the same, both insances of prime were running for 1 hour



((temps from left to right begining in the black one reffer to: HD , GPU ambient , GPU memory , GPU diode , GPU , CPU core1 , CPU core2 , CPU , Motherboard. motherboard is a dfi ultra-d ))

At this point I was thinking that maybe noctua fans don´t make enogh pressure so air can´t go fluid through the heatsink. I believed that was the difference, nexus better for heatsinks and noctua to put air out of the case.

Anyway I tried again comparing noctua and nexus putting air out of the case. the only fan trowing that air out was one, first noctua then nexus. I blocked top vent and begin again.





noctua at 5v

GPU:
gpu 52
memory 47
ambient 47

CPU:
core 1 56
core 2 56

Motherboard: 61



now nexus at 7v temps here raise 0.5º in my room nothing important anyway

GPU:
gpu 49
memory 44
ambient 44

CPU:
core 1 54
core 2 54

Motherboard 59



Thats all, i´ll keep on investigating, I´m going to buy or try to do sth to know exactly how fast my fans are spining just in case my nexus fan is somehow faster than usual...

I give you the link of the forum where i usually post, it´s in spanish, but you can translate it with babelfish if you need to.

http://foro.noticias3d.com/vbulletin/sh ... p?t=176215

also this afternoon I´ll try with antec tricool at low position (30 cfm)

What do you think about all this? And sorry fot my english, I don´t usually use this language

Hi Paulesko, nice to see a colleague from Spain in these forums

I had a similar experience with Noctua fans. I could compare them with my papst fans, and I got similar results. At the supposedly same CFM from noctua and papst I got higher temps with noctuas.

I could observe that there was not only a problem with pressure, I could sense that there was a problem with air aiming. The subjective impression was that the noctua disperses more the airflow.

I couldn't make many scientifical type comparisons because I had the noctuas only for a couple days before I shipped them back, so mine are mainly subjective impressions. The differences in temperatures were real though.

I´ve also notice that. Noctuas move air very fast in the external area of the fan, and that air spreads widely, another thing important is that in the center of the fan there is a big area where air just don´t move at all...

Take a look at www.noticias3d.com in refrigeracion/silentpc subforum, there are at least two people that have the same problem..

In that case, maybe we should examine more closely the Sharkoon Silent Eagle 1000 (please see my post at the top of this page).

It would be interesting to test them. Besides they are very common in Europe. I've seen the review at madshrimps and it seems that this model could be a good candidate.

However I don't think that the measures in that review in dbA at only 5cm are too much representative of the noise. I would rather like to see the starndard SPCR measures at 1m to compare with other fans.

Then we just have to hope SPCR reviews the Sharkoon soon !!

This concerns me. I think we will have to do our own "in-case" testing and take a close look at how we test airflow. Keep up the testing, and we'll see what can be done.

I'm looking forward to those results.
I really think that those tests will change the ranking of top fans.

Devon, that's great news ! I also think that in-case testing would definitely improve SPCR's already great methodology, and that it could result in a new fan hierarchy.

May I add that it would also be a good point if you added (for the top fans at least) some noise/vibration testing in horizontal position, head up and down ? Just to see if there are always significant differences with the vertical position, or if some fans are more subjected to noise/vibration than others (probably depending on the bearing types, but maybe not only).

Thanks.

In frankgehry's post, note that the first diagram is of an ordinary sleeve bearing fan, not the Noctua, which has an additional magnet.

That magnet seems to have the same function as the magnet in the S-FDB bearing: to apply an axial force to the rotor, to preload it in one direction. This seems strange at first glance, since sleeve bearings don't do particularly well against axial loads, so why would one want to add an additional axial load? My guess is that the idea is that one can arrange for oil to be pumped by the bearing in a certain direction, by adding helical grooves along the shaft or housing. If you know which end of the bearing needs the oil (because you preloaded it, using a magnet), then you can pump the oil in that direction, and thus get an oil film which, while the fan is moving, completely lifts the fan rotor off its axial bearing plate (aka thrust bearing plate), avoiding all metal-to-metal contact. (In the radial direction, I think even ordinary sleeve bearings avoid metal-to-metal contact once the fan is started up: with good design, paying attention to oil type and bearing gap, one can naturally get a film of oil which separates the surfaces while they are in motion.)

(Another possibility would be that the axial force from the additional magnet is there to counterbalance the aerodynamic forces acting on the fan. But it'd be hard to counterbalance those forces exactly, since one doesn't know exactly how fast the fan will be rotating, nor against what pressure difference it will be pushing, both of which affect the aerodynamic forces. Besides, counterbalancing the aerodynamic forces doesn't do anything about gravitational forces, which are of similar magnitude if the fan is mounted blowing up or down.)

I can't tell how the Sunon "maglev" bearing could work. The explanation given in French seems to be that the magnetic forces keep the rotor centered. But maglev systems which operate by attraction, as this one is depicted as doing, generally require active electronic feedback control, since the system is otherwise unstable: the closer the attracted magnets get to each other, the more force between them, so instead of maintaining the gap at a constant level, it tends to reduce it. And this system uses permanent magnets, so there is no possibility of electronic feedback control; besides, that would require sensors to detect the gap -- and given the high degree of precision required, I can't see them fitting that stuff into a fan meant for consumer electronics; it would cost far too much. More likely "maglev" is a misnomer, and what they've really done is to use magnets to preload the bearing in an axial direction, forcing the rotor against the single bearing ball at its end which forms a contact point. The absence of any metal-to-metal contact while the unit is operating (which the cooling-masters website shows a graph of) is normal for a sleeve bearing in the radial direction; and in the axial direction, the lack of metal-to-metal contact is explained by the use of a nonmetallic ball.

So my guess is that of these three high-zoot bearings, two are describable as fluid dynamic bearings, and the third as a sleeve bearing with a single-ball axial thrust bearing -- with all three bearings using magnets to provide an axial preload.

Axial preload would explain why it's not unusual for fans to prefer, in horizontal mounting, to either blow up well or down well, but not both. Gravity would either assist the preload (good) or overcome it (bad).

It would be very useful if there were some kind of list of fans that would, in horizontal mounting, blow up or down well.

In fact that would also be a great improvement to SPCR's testing : to tell if a given fan can be used horizontally, to blow up or to blow down. For instance, I still don't know if the Sharkoon can do it or not !

If the preload is done right, it shouldn't matter which direction the fan is blowing in. The idea would be to preload the fan in a direction that it's set up to handle well (due to using a better-than-ordinary bearing to take the thrust forces in that direction), and to have enough preload force to overcome any aerodynamic and gravity forces that might arise in use.

It would be rather difficult for SPCR to assess how long fans will last blowing in different directions: it would involve testing to destruction, which takes quite a while.

Norman, as far as I know a head-down position can also cause (for some fans at least) an oil leakage, thus shortening the life of the fan. I've also read something about additional noise (SPCR's primary concern of course) and vibrations...

Sounds like the Noctua would be fine sending air out of a system then, but would be very suboptimal for blowing through heatsinks or other dense cooling devices. I'd definitely take a noctua for a back case fan then, but not for something internal.

I take worse temperatures with noctua sending air out of the sistem than with nexus at the same rpm. Think about it, in a heatsink, the fan has to blow through the fins but in a case it has to suck it through dust filters, wires and so on...

Did you remove the fan grill for the test? If the grill is still there then the situation is similar to a heatsink, many grills have a high airflow impedance (the one in my antec sonata for example...).

I have an antec p180 and all exhaust grills have been removed, I also removed dust filter, but this didin´t improved temps drstically, (as I was expecting) just 1º

I'm doing some benches lately because of my thoughts about the best configuration.

I've got at hand:
- Scythe DFS122512L (comes with a Ninja Plus Rev. B)
- Papst 4412FGL (my fan is very smooth. It's about as smooth as the recordings from the Scythe-fans from SPCR)
- Noctua NF-S12-1200 (from very close a twinking sounds emits, but fades completely at close distances (~15 cm/0.5ft)

The Papst is momentally unused.

My system:
- Antec's P180, 2 fans top&back(variable). Vents are not restricted, cut open. The spoiler was over the top at all times.
- AMD 4400+ rated high, like 120W TDP, core temp shows 65W. I'm guessing from the temps that it is the 89W version.
- AsRock Dual939SATA2. The Ninja sitting on top is fanless and aligned with the backfan. The topfan covers only half of the Ninja-fins.
- XFX 6800GT /w NV Silencer @5v.
- Seasonic S12-430 cooling my Hitachi Deskstar T7K250 @ AAM, does never ramp up.

Test Methodology:

- Leaving everything the same, occupy/switch the top- and backfan @ 5v=inaudible.
- Core Temp and Speedfan and the nVidia-console were used to gain temperature stats.
- Orthos was used to burn my CPU with 2 threads.
- Atitool was firing the 6800GT up.
- Give the hardware a feeling how hell is for 20 minutes.
- Ambient temp didn't vary during tests ~22 degrees C.

Testresults:

First with Scythe @ top and Noctua @ back and burning the:

CPU only CPU+GPU

CPU 62/57 61/59
GPU 53 84
Amb 31 40
NBri 47 45

Next with Noctua @top and Scythe @ back and burning the:

CPU only CPU+GPU

CPU 60/56 59/57
GPU 53 82
Amb 31 38
NBri 45 45

I've ducted the Ninja and Mainboard also - give you some results of that in the future, maybe.
Didn't do tests with the Papst, I'm curious myself.

Short Conclusion:

The Noctua pushes much more air, noticable by hand (even when mounted), comparing it to the Scythe fan.
The temp-drops of the CPU with the Noctua on top, are explainable by the amount of air the Noctua moves, compared to the Scythe-fan. The top-fan is very close to the Ninja heatsink even thought it's not perfectly aligned.
The Ninja is fairly unrestrictive, so the weak pressure/suction created by the Noctua is'nt that big deal. So the noctua wins here definetly.

I'm sorry for the f****d up aligning of the temperatures.

The space devides the table, it should have looked like this:


Testresults:

First with Scythe @ top and Noctua @ back and burning the:

CPU only--------CPU+GPU

CPU 62/57------61/59
GPU 53----------84
Amb 31----------40
NBri 47----------45

Next with Noctua @top and Scythe @ back and burning the:

CPU only--------CPU+GPU

CPU 60/56------59/57
GPU 53----------82
Amb 31----------38
NBri 45----------45

Greetings,

After reading the Thermaltake Big Typhoon review, and looking at the side-by-side photos of the black&white Nexus and the Orange Nexus -- I've spotted a third variation!

The black & white Nexus has three round struts/spokes -- while the orange Nexus has rectangular (nay, trapezoidal!) struts.

The other tiny difference is the slot where the three wires pass through to the back of the fourth strut is perpendicular to the strut on the B&W model; while it is an arc on the orange model. Both Nexus fans have the tab that helps hold down the wires, too.



And the black & orange Tt/Hong Sheng model has the three trapezoidal struts -- and it definitely has a slightly wider fourth strut/spoke -- there is room enough for a fourth or even a fifth wire there, while the other two have just enough room for the three wires. And it is missing the wire tab...




Hmmm, the plot thickens.

This are small differences -- the round struts are probably the most meaningful. But they are differences; where we did not know there were any, save the color.

We also have a sample in the lab that is completely orange with the rounded struts. I believe that the trapezoidal struts are an older revision of the Nexus.

Wow, I've been away too long. Thank you very much for the 120mm comparisons & followup discussion of how backpressure can compromise the winner's real life performance. Thanks again!

Craig in NJ

I just joined this forum & making my first 'post here right now...

Got here through the link provided in the 120mm fan review. That's
where I read the reviewers comments on the Slow Version: S-Flex SFF21D (800 RPM) and the "ticking" noise the reviewed sample made... well... I think that really was a "bad sample" question as mine doesn't make any "ticking" noises, whatsoever.... and I'm running it at 12v! (fitted "decoupled")

Let me elaborate on that... I first learned of SPCR when "googling" for the Antec NSK2400 and found this;
http://www.silentpcreview.com/article591-page1.html


That article made me buy that very desktop case plus the included PSU.

Yup... I was quite impressed! I already built a PVR/HTPC system (AMD Sempron 64 3100+ plus Scythe Samurai Z + SilenX 92mm 14dba fan running at 5v, AS Rock K8NF4G-SATA2 754 MB, 2x 1Gb Kingston Value DDR400, ATI 1600 Pro 512mb PCI-e card with Zalman VF900 (fan running at 5v).... ATI 550 Theatre Pro card....And a Samsung 250GB, S-ATA II hdd...It was "fitted" in a "no name" desktop housing....

That "housing" plus the included PSU never made my system quite.. well.. quiet... quite the opposite..sadly...

So... after I read the article I ordered the Antec NSK2400, together with the S-Flex SFF21D (to act as a single 120mm exhaust fan) plus a Yen Sun
aka Y.S. Tech FD1281251B-2A (Green Label) 80mm fan (for the PSU).
Ordered a few of these "rubber" fan "decouplers" as well...

All in all I'm VERY happy with my current "setup".... can't hear the PSU
fan, even under "load"... Same goes for the 120mm exhaust fan, the one in question, and even the SilenX 92mm fan on the CPU heatsink is inaudible... at 5v, that is.... oddly... it has a "ticking/clacking" sound
over 5v!...

Pics & other data are available if needed/wanted...

This has made my PVR/HTPC expierience a LOT more enjoyable!
My wife agrees... so SPCR owes 'kudos for the article mentioned...

Thanks!!!

TDR