Improving fan efficiency by reducing the tip gap.

[Lynxman]

Another one: http://noctua.at/media/catalog/product/cache/2/image/9df78eab33525d08d6e5fb8d27136e95/n/o/noctua_nf_a9x14_pwm_1.jpg

[Lynxman]

PLA isn't a problem It's in a cold area. It's been installed for weeks. 0,1 mm isn't necessary, or feasible. Sub 1 mm should be possible on quality fans.

[Lynxman]

I did the same modification to the intake case fan on my Nuke build and it lowered the GPU temperature from 66°C to 61°C (Ethereum mining). That's a 5°C reduction. There are real gains to be made by this, people. The efficiency improvement is obvious.

The Sickleflow fan had a 2mm gap originally. I printed a 1mm thick ring, with holes for the LEDs.
 

[WoodenMarker]

On 12/9/2017 at 4:50 PM, KaminKevCrew said:

As for the blades warping over time, I highly doubt that ever happens. PC case fans do not turn fast enough to have much centrifugal force on them. EDFs, as used in remote control jets have blades that are made of the same stuff in many cases, and those don't warp enough to rub on even tighter tolerances, despite the fact that they run at 20,000+ rpm easily, even when small debris can get sucked up and hit the blades. It's really not an issue here.

On 12/9/2017 at 4:48 PM, Lynxman said:

I have never seen a fan blade stretch over time. I've used PC's since the 80's and many of my current fans have been running almost 24/7 since 2010. They just don't spin fast enough. A 120 mm fan spinning 2000 rpm will have roughly 223 g on each blade which makes a 1 g blade weigh 223 g. Insignificant unless it's heated to 105°C, which is about the glass transition temperature for ABS.

On 12/9/2017 at 4:57 PM, manikyath said:

i'm sure that noctua could easily get their fans an extra edge in silence and efficiency, but i'm also sure that they dont want to degrade reliability.

Fan widening does happen. The reason why fan blades don't scrape the frame is because the growth is accounted for with a wider gap.

A noctua representative describes it as impeller creep here:

This is one of the reasons their NF-A20 was delayed for years and why the NF-A12 is still under development. 

[Lynxman]

Not sure why there's so much resistance to this idea here. He even says the growth is only up to 0,1-0,2 mm, which isn't anywhere near enough to touch the ring with my upgrades. The mod I'm doing is to reduce the 2mm+ gap of most fans to less than 1 mm. 

 

Here's a ring for those who want to make their own:
https://www.thingiverse.com/thing:2730566

[SkilledRebuilds]

So who.... like me is going to just grab a ghetto spare fan from the drawer and just add electrical tape around it to "buff" the edges closer and see what happens?

YOLO

[WoodenMarker]

3 hours ago, Lynxman said:

Not sure why there's so much resistance to this idea here. He even says the growth is only up to 0,1-0,2 mm, which isn't anywhere near enough to touch the ring with my upgrades. The mod I'm doing is to reduce the 2mm+ gap of most fans to less than 1 mm. 

Here's a ring for those who want to make their own:
https://www.thingiverse.com/thing:2730566

The ring mod is really neat and I'm not against it. The wording used so far seems to imply that it doesn't ever happen to other fans though and I disagree.

1 hour ago, SkilledRebuilds said:

So who.... like me is going to just grab a ghetto spare fan from the drawer and just add electrical tape around it to "buff" the edges closer and see what happens?

YOLO

Let us know how it goes!

[Lynxman]

Why even worry about the insignificant theoretical blade stretch though? If it starts rubbing after six years just make a thinner ring or remove it. If anything it's probably only going to close the gap even better than a new fan, considering how small the stretch will be, if any at all. It's not a problem, neither is PLA glass transition temperature, or any of the other concerns brought up.

I'd like to try a bell mouth with an integrated ring next. Maybe I'll do it to the Noctua server fan I have on the CPU of my workstation and run tests.

[WoodenMarker]

12 hours ago, Lynxman said:

Why even worry about the insignificant theoretical blade stretch though? 

It means less reliability and lower lifespan. It's for the same reason that intel uses a thermal paste under their consumer cpu's as opposed to solder. 

[Lynxman]

Not making sense.

[WoodenMarker]

On 12/26/2017 at 10:04 AM, Lynxman said:

Not making sense.

There's always a margin for error. There are a number of fans that fail due to the blades hitting the frame. The closer the blades are to the frame to begin with, the higher the number of failures occur from this. 

Similarly, there are also a number of intel cpus that fail due to the expansion and contraction caused by heat. This is a bigger issue for smaller dies which is somewhat remedied by using a thermal paste as opposed to solder. 

Higher failure rates can result is lowering a manufacturer's reputation and increase the number of product returns.

 

You need to quote a post or tag a member or else they won't get a notification you replied to them.

 

[Lynxman]

11 minutes ago, WoodenMarker said:

There's always a margin for error. There are a number of fans that fail due to the blades hitting the frame. The closer the blades are to the frame to begin with, the higher the number of failures occur from this. 

Similarly, there are also a number of intel cpus that fail due to the expansion and contraction caused by heat. This is a bigger issue for smaller dies which is somewhat remedied by using a thermal paste as opposed to solder. 

Higher failure rates can result is lowering a manufacturer's reputation and increase the number of product returns.

 

You need to quote a post or tag a member or else they won't get a notification you replied to them.

 

But look at the premise for the whole thread. I am talking about fans with exessive gaps, not fans that already have well optimized gaps. And if the fan starts touching in six years then make a thinner ring. 

One of the things I want to try is a ring that is secured to the fan and rotates with the fan. The concerns of blade stretch are exaggerated. Believe me.

[WoodenMarker]

30 minutes ago, Lynxman said:

But look at the premise for the whole thread. I am talking about fans with exessive gaps, not fans that already have well optimized gaps. And if the fan starts touching in six years then make a thinner ring. 

One of the things I want to try is a ring that is secured to the fan and rotates with the fan. The concerns of blade stretch are exaggerated. Believe me.

What's excessive and how are you determining that the blade stretch concerns are exaggerated? 

 

There's a certain amount of cost that goes into R&D to make sure that the gap isn't too small. This isn't an issue for fans that charge a premium but you don't usually see these gaps in those fans--usually cheaper ones for that reason. 

I'm not referring to your ring as problematic but instead that there's good reason to have a safer than sorry approach to retail fans. Most people prefer reliability over a bit more performance. 

 

[Lynxman]

Like I have shown before. PC fans don't spin fast enough to stretch. The forces are incredibly small unless you're into 50W Delta server fan territory, of which I have several (who also don't stretch after many years). I've been using PCs since the 80's, and still have fans from twenty years ago. Ive been using plastic fans that spin tens of thousand rpm in my hobby. The stresses that computer fans operate under are just insignificant even to ABS. The best fans even have glass reinforced plastic, and if you think they will expand significantly under pc fan speeds then you are just wrong.

Do you even realize how little 0,2 mm is, as mentioned by the Noctua guy in the video? He's probably talking about a worst case scenario where the fan is operating at temperatures near the temperature limit of the material. No 120 mm computer fan will deform plastically from cold running at 2000 rpm unless it's very unbalanced or cracked. It's just absurd.

Again. Why on earth worry about that anyway? If the fan should stretch the excessive 0,2 mm, which is just 0,008" if that makes it easer to understand, it won't even touch the printed rings. You don't need a 0,1 mm gap to improve the fan. Achieving a 0,5-1mm  gap is a great improvement over the 2-3 mm gap many fans have. A gap of 0,1 would be perfect if it can be maintained, but it's not necessary.

I wish I had kept my findings to myself at this point.