News Arctic uses a fan to cool another fan — S12038-4K 120mm fan adds an extra fan in its center to cool its central shaft

[Admin]

Arctic's new S12038-4K/8K server fans have an extra set of fan blades for the true cooling enjoyers. Really, the extra blades are to cool its own center shaft, which may be necessary based on the fan's class-leading performance.

Arctic uses a fan to cool another fan — S12038-4K 120mm fan adds an extra fan in its center to cool its central shaft : Read more

 

[Kamen Rider Blade]

I think the next step in "Case Fan" tech evolution is to go back to look at
TMD (Tip Magnetic Drive) and implement it in the best possible form.

Benefits:
1) You narrow the center Dead zone down to a 5mm diameter which is "Good Enough".
2) With Magnetic Bearings, the hub of the fan can literally float on a cushion of air/oil
3) You move the EM Coil Windings to the 4x outter edges, ergo spreading the heat load.
4) Your fan design will always have a closed outter wheel edge, so you get rid of tip vortices inherently by closing the fan blade tip edges.
5) You have a Larger Surface Area for air to move through with "Less Obstruction"
6) AlphaCool has proven that you can "De-Couple" the fan components from the frame components of the fans and get better noise characteristics, the next step is to open things up and use the metal housing as "HeatSinks" for the EM Coil Windings at the edges. to help with cooling.

 

[Zenthar]

For the curious, about how loud those are, Tom had a news about the 8K variant with some videos.

 

[Phaaze88]

I clicked on the article expecting to see a certain meme, and was not disappointed.

 

[80251]

Nice fan for a cheap price. Longevity is an important fan performance metric in a data center. Power is too and this fan seems to have a too good power draw metric: 3.96 watts for a 4000RPM, 120x38mm fan.
It's weird how Arctic have distanced themselves from their own fan design.
Here's a review from Amazon:
'It also didn't help that Arctic refused to answer questions on their reddit sub about these fans.'
Bizarre.

 

[thisisaname]

Next step would be a fan for the fans fan.

 

[Notton]

I think the next step in "Case Fan" tech evolution is to go back to look at
TMD (Tip Magnetic Drive) and implement it in the best possible form.

No, it ain't.
This is basic math, like Gr.7 levels

10.5cm fan blade w/ 1cm hub (diameter)
area of fan blades: 86.59cm² - 3.14cm² = 83.45cm²

11.5cm fan blade w/ 2cm hub (diameter)
area of fan blades: 103.87cm² - 12.57cm² = 91.3cm²

And that doesn't even get into the highly technical aspects of fan blade design. The gist of it is, bigger and slower is better than smaller and faster, even with a slight loss of static pressure. The only reason to go small and fast is because of size constraints.
It's better to change the fan thickness and geometry.

 

[yakidd]

They should've named it Fan²

 

[80251]

@Kamen Rider Blade
How efficient were those TMD fans? Did they produce similar or better CFM/static pressure than traditional radial axial fans of the same size?

 

[Kamen Rider Blade]

And that doesn't even get into the highly technical aspects of fan blade design. The gist of it is, bigger and slower is better than smaller and faster, even with a slight loss of static pressure. The only reason to go small and fast is because of size constraints.

I don't disagree with you on those points, "Bigger & Slower" = Less Noise.

But I'm applying the TMD design principle to fans of ALL sizes.

Ranging from 4 cm to 20 cm in Case Fan sizes.

It's better to change the fan thickness and geometry.

I concur, but 25 mm remains the most popular thickness, I to want "THICC" fans for certain specific applications.

@Kamen Rider Blade
How efficient were those TMD fans? Did they produce similar or better CFM/static pressure than traditional radial axial fans of the same size?

They made 1x model, it was 70 mm (Why, 70?, I have no idea as to why that was the first design size to be chosen).

They cranked it up to 5800 RPM as the upper limit.

http://www.dansdata.com/tmdfan.htm

So while it was "Noisy", it wasn't as bad as other fans in that RPM range.

 

[80251]

@Kamen Rider Blade
That article you cited points out some of the downsides associated w/TMD fans. For one thing, the outer impeller ring has to be supported while it spins. On a 70mm fan thin struts might work, but on larger fans this might not work (i.e. because the fan blades are larger and heavier) so you would need some sort of bearing race to support the outer impeller ring, which adds more friction -- especially considering the linear velocity with which this outer impeller ring bearing race would have to rotate -- and the fact you would have to more than two ball bearings to support it. Maybe a Sunon maglev type solution could be adapted for this though.
I hope some industrial fan manufacturer somewhere is working on next-gen TMD fans or maybe even some drone manufacturers.

 

[Kamen Rider Blade]

That article you cited points out some of the downsides associated w/TMD fans. For one thing, the outer impeller ring has to be supported while it spins. On a 70mm fan thin struts might work, but on larger fans this might not work (i.e. because the fan blades are larger and heavier) so you would need some sort of bearing race to support the outer impeller ring, which adds more friction -- especially considering the linear velocity with which this outer impeller ring bearing race would have to rotate -- and the fact you would have to more than two ball bearings to support it. Maybe a Sunon maglev type solution could be adapted for this though.
I hope some industrial fan manufacturer somewhere is working on next-gen TMD fans or maybe even some drone manufacturers.

1) It's only truly viable via "MagLev Bearings", but "MagLev Bearings" are my favorite type. They last the longest.
2) We have new LCP (Liquid Crystal Polymer) to enhance material strength for the base polymer material for the fan blades & to hold the magnets along the rim.
3) Nothing that I want is using any truly "New Tech", it's just a packaging / assembly issue & solving for the "Improved AeroDynamics" that would come out of this due to the improved smaller hub. That just requires the Fan Team's resident AeroDynamicists to redo the computation until he finds a fan blade design to Min/Max the new Case Fan structure / Volume to get the most CFM / db emitted and to make sure that the pitch / frequency of the new blade config is stellar like the AlphaCool Metal Fans are.

They don't get enough credit for making the tone of the Metal Fans REALLY good for the dB rating it has.

Obviously it's not the most efficient Fan, or the one that pushes the most CFM.

But very few companies are "Noctua Level" of Efficiency for CFM/dB.

 

[80251]

The only noctua fans I really like are their NF-A14 iPPC3000s. The rest of their fans just don't produce enough CFM and/or static pressure to interest me. I usually use 38mm or 50mm thick PWM fans in my rig wherever I can because if I need more CFM and/or static pressure I can always get it by running them faster. SanAce, Delta, silverstone (for their 180mm fans) and Nidec are my current favorite fan manufacturers.

 

[RoLleRKoaSTeR]

Wow, "phrasing" in the headline, much?

NTTAWWT

 

[Kamen Rider Blade]

The only noctua fans I really like are their NF-A14 iPPC3000s. The rest of their fans just don't produce enough CFM and/or static pressure to interest me. I usually use 38mm or 50mm thick PWM fans in my rig wherever I can because if I need more CFM and/or static pressure I can always get it by running them faster. SanAce, Delta, silverstone (for their 180mm fans) and Nidec are my current favorite fan manufacturers.

Noctua tunes for "Maximum Efficiency" CFM/dB.

They rarely ever target Maximum CFM or Static Pressure.

Other designs might go for that niche, but Noctua doesn't care.

They want their fans to move "Enough Air" while being "Quiet Enough" in nearly all situations.

They managed to find a market for that.

 

[purposelycryptic]

Well, you got me, I picked up a set of these and the 80mm version to try out in my 4U and 2U server, respectively. They are in a pretty well sounds isolated rack, so I guess I'll see what happens. At $35 for three, my curiosity just got the better off me.

 

[purposelycryptic]

Noctua tunes for "Maximum Efficiency" CFM/dB.

They rarely ever target Maximum CFM or Static Pressure.

Other designs might go for that niche, but Noctua doesn't care.

They want their fans to move "Enough Air" while being "Quiet Enough" in nearly all situations.

They managed to find a market for that.

Their "heavy duty" line is a little different - I currently use NF-F12 iPPC 3000's in one of my servers, and, while they are definitely quiet compared to the big Deltas they replaced, so is a 747 on takeoff.

To quote their Amazon listing:

Running twice as fast as the 1500rpm retail version, the NF-F12 industrialPPC-3000 PWM provides extreme airflow and pressure performance. Combined with the industrialPPC series' robust polyamide construction and advanced protection features, this makes the fan ideal for challenging industrial applications.

The industrialPPC models are high-speed fans for industrial applications. As such, they are not as quiet as other Noctua fans and less suited for low-noise PC builds.

 

[purposelycryptic]

The only noctua fans I really like are their NF-A14 iPPC3000s. The rest of their fans just don't produce enough CFM and/or static pressure to interest me. I usually use 38mm or 50mm thick PWM fans in my rig wherever I can because if I need more CFM and/or static pressure I can always get it by running them faster. SanAce, Delta, silverstone (for their 180mm fans) and Nidec are my current favorite fan manufacturers.

Well, Noctua's mainline fans are designed to be as quiet as possible while still giving decent performance, which is basically the opposite of your priorities.

The 'i' in their iPPC line stands for "industrial", and is entirely different from their regular fans, built for static pressure and moving air in harsh conditions, hence the fiber-reinforced polyamide construction, three-phase motor and IP52 water/dust resistance rating.

They still try to keep them as quiet as possible, but, unlike their normal fans, that is priority #3, after CFM/static pressure.

I have the NF-F12 iPPC 3000 PWMs in my media server right now (biggest size that will fit), and they do their job, although I still needed to replace the original Supermicro passive CPU coolers with active ones as a result, since they just can't compete in sheer performance with the hurricane force of the fatboy Deltas that they replaced, which were just painfully loud, even in a silenced server cabinet. The Noctua ones are hardly audible by comparison, and since my servers are in my home office...

I'm curious to see how these Arctic ones compare - you can get a pack of three for about as much as one of the Noctuas, so I'm not holding my breath, but I can always find a spot for them in my workshop if they end up being too loud, or simply don't offer any additional performance over the Noctuas. I just like novel things, and a fan-cooled fan is pretty damned novel.

My desktop boxes all have either Noctua or Scythe fans in them, though, since noise is not my friend. I haven't heard anything about Scythe in ages, but I bought so damned many of their fans back in the day, and their radiator fans continue to hold their own.

 

[purposelycryptic]

Nice fan for a cheap price. Longevity is an important fan performance metric in a data center. Power is too and this fan seems to have a too good power draw metric: 3.96 watts for a 4000RPM, 120x38mm fan.
It's weird how Arctic have distanced themselves from their own fan design.
Here's a review from Amazon:
'It also didn't help that Arctic refused to answer questions on their reddit sub about these fans.'
Bizarre.

They probably want to avoid desktop/enthusiast users buying them and then voicing their complaints online after realizing they are way louder than they expected. Arctic has a public image they worked hard to create, after all, so I don't blame them for wanting to maintain it (although refusing to even answer questions is perhaps going a bit far).

Those that actually want/need this type of fan will likely hear about them regardless, just as we did, and they could be marketing them directly to server manufacturers as well.

¯\_(ツ)_/¯

 

[80251]

I hope everyone here realizes the claimed 3.96 Watt power draw of this 4000 RPM fan is impossibly hard to believe. My 4000 RPM Nidec B35502-35 120x38mm fan draws 16.8W and my delta 120x38mm AFB1212HHE draws 9W even though it spins 1000RPM slower than this Arctic fan.