How to make a noisy fan quiet

[phconnell]

As I wrote somewhere else on this site a lot of fan noise comes not so much
from the tips (perhaps at very high speed) but from the four ribs attaching
the motor to the frame.
The problem is that the ribs are too close to the trailing edge of each
blade which has an eddy swiping over the ribs, each blade creating
a hum with 4x the frequency of rotation (but phase shifted).

I have just bought a Silenx 120x120x38 mm fan with 60 cfm at only 14dB
and like most things from Silenx the 14dB is a lie and the actual hum
a very real and loud disappointment.
When I cut off the ribs and run it (just motor and fan) it is then 14 dB or even
less, mostly air woosh (and any motor rattle or whistle if present).
If you then bring up a bar under the fan the hum begins slightly at 10 mm
distance from the trailing edge and louder as you get closer.

Now you know why the fan is so loud.

The solution is to find a way to remount the motor with ribs (thin steel wire would
be better) 10 mm from the trailing blade edges.
A wooden disc glued on the hub or some simple clasp system (if you want to remove it)
and you have the quietest fan you ever can expect to have.

For most PC applications the frame is not necessary - just a circular wooden
or plastic disc to attach the motor hub, and a cross with attachment holes at
the ends for ribs. It would be perfect. On some quieter CPU coolers they have
adopted this approach to minimize noise.

Don't be afraid of cutting up what you think are gorgeous bits of technology.
If you just take the parts you need and put them together intelligently you can
get the best of all worlds. You only have to think about it, then do it.
One of those small dremel machines is also a big help.

I am thinking of devising a simple remounting kit so that anyone can take
his typical noisy fan, cut it off the ribs and attach it to a new mount and
all the fuss is over.

The problem is with the standardized 25 or 38 mm fan depth which forces
any manufacturer to juggle fan blade depth (and volume flow) against the
inevitable eddy interaction with the ribs. If they would just think about it
there would be no forum like this but they are either fossils or afraid the
increased cost, or deviation from a standard, would make it a lemon.

[amjedm]

Perhaps create a duct and somehow stick the modded fan/motor in there?

[Chris Beard]

I'm so trying this when I get home with one of my £1 fan spares from ebuyer.

If you lower the RPM does the eddy effect become less and/or interact at a smaller distance?

[johan851]

What if you were to cut out two of the ribs? Does it lower the frequency of the noise? That might be an easy way to reduce the intrusiveness of the hum without remounting the entire thing.

[didi]

What if you were to cut out two of the ribs? Does it lower the frequency of the noise? That might be an easy way to reduce the intrusiveness of the hum without remounting the entire thing.

You would probably loose stability for the fan. A fan that runs off center/vibrating is an even worse noisemaker.

[Gojira-X]

Hey phconnell, some pictures of the process you were following to get your reasoning would help. That way, we can evaluate it against fluid dynamic theory.

[phconnell]

Hi everyone,

Interesting question about cutting off two of the ribs.

If the fan humm is due to a trailing eddy of the fan blade then each blade would
create its own humm.
Every time the eddy sweeps over a rib then it creates a pressure distortion
blip so the frequency of the humm should be N*RPM/60 Hz if there are N ribs.
So leaving two should halve the pitch, though it is more complicated as
each fan blade puts out the same pressure blip pattern but shifted by
60/(RPM*B) seconds in phase if there are B blades.
If there were 8 blades and two ribs I suppose that would put the humm at
4*RPM/60 Hz. With seven blades its a bit irregular so I will have to write a
program to get a graph of it.

Apart from theoretical ideas there are emperical ideas. Just cut the ribs off and
run it while the motor and blades are dangling in mid-air and the fan (especially
if its a Silenx fan) is now less loud than it says on the packet. I have the parts
of a new computer connected on a small table next to me with a Thermalright
heatsink and a Silenx 120 mm dumped on top but with the motor hub (no ribs)
glued to the centre of one of those metal fan protection grills, and its the
quietest little mouse you ever heard.

I have cut up several fans like this and the results are mostly the same.

The test is to somehow hold the motor with some small grip (or your finger nails)
and listen to the noise as you bring a bar up (like a rib) close to the fan blades.
It is then clear what makes a significant part of the fan noise.

Once you do that the theory is a bit irrelevant and the question is how to remount
it all to look like some wonderful piece of hi-tech and not the ruins of that georgous
fan you paid so much for.

Thats a serious problem, but there are no vibration problems whatever way the
fan is mounted (as someone suggested) - that only happens if you want to cut up
a 120 mm fan to make a 100 mm fan.
The problem is that if you want to remount it in the frame but pushed forward by
about 5 mm then you have to mount it exactly in the centre to make sure the blades
don't hit the frame walls giving you a clicking sound.

Thats actually not that easy. For most fan configurations (case and CPU) I don't
think you actually need the round duct (or the difference is not that great) so the
simplest thing you could do is just glue it to a fan grill which then enables you to
screw it where you like. Or glue it to a simple wooden or plastic cross with holes
at the ends of the cross.

It doesn't look great but the "zen of motorcycle maintenance" would tell you that
you are in functionality heaven.

There is a more elegant way but you need a small round circle of wood 5 mm
thick (same diameter as the motor hub), some 8 staples, a rubber band, 4 small
springs, and 4 lengths of 3mm diameter aluminium or plastic rods.
With this you can remount it exactly in the centre of the old frame and it will look OK,
but its a bit of DIY drilling 8x1mm and 4x3.05mm radial holes around the wood
circumference.

Its actually a good intelligence test - how do I cut out this little motor and remount
it right in the centre of its frame again. If only the manufacturers would make the
fans 30 mm thick, instead of 25 mm, there wouldn't be so much wailing.

For the real theorists the problem seems not in the usual ideas of fluid dynamics
but in eddy theory where the two air flows from the top and bottom of a fan
blade meet at the trailing edge with different velocities. But you can imagine such
a turbulent trail passing over a near object (like a frame rib), giving a pressure
distortion when it meets the object and then returns to normal when it leaves the
object behind.
This must give some kind of pressure "blip" and I presume it is the frequency of
those blips that makes the humming noise.
Just first thoughts - please correct me if I am wrong.

[Felger Carbon]

This must give some kind of pressure "blip" and I presume it is the frequency of those blips that makes the humming noise.

There are a lot of products with fans that don't have the ribs in question. They're extremely common: for instance, the Arctic Cooling Freezer 64 Pro and its many cousins. Is this your ideal noiseless fan?

[jaganath]

I have one such ribless fan taken from a TT Mini Typhoon, pushes a decent amount of air at top speed (2.2krpm) but not that quiet. The sound the OP is talking about is the blade-passing frequency, this is just one of the many types of sound fans generate:

http://www.jmcproducts.com/cooling_info/noise.shtml

of course in an ideal world there would be no ribs and we would have infinitely close tip clearances, but even without that we still manage to find some decent fans like Nexus, Yate Loon and Globalwin. IMO the fan blade design and bearing design also have a very strong influence on noise emission.

[phconnell]

I don't know how quiet the Arctic Cooling Freezer 64 Pro and its cousins are
and do not have any sort of ideal fan out there - only in my head.

My observations come from experience in that I had trouble with a lot of fans
being far too noisy for the compact PC case I am building - far noisier than the
outright lies on the packet. For many of these just cutting off the motor hub
got rid of the problem, but not for high speed flow fans where there are other
sources of noise and whoosh.

I have some NEXUS 92 mm fans and they are very quiet, but I think at the
cost of a lower air flow. I can well believe some manufacturers have done a
lot to make their fans as quiet as possible.

All I am saying is that if you have spent all that money and the fan is a hair
dryer then there could well be a lot to be gained in remounting it, but in high
flow cases perhaps not completely silent.

I have found that to get a really quiet case (I mean no noise at all) and good
cooling power (a thermal resistance of 0.1 - 0.2 C/Watt) I have to take off the
ribs of the fans I have found, run them as slow as possible, and design my
own large heatsink-heatpipe coolers as most of the CPU and VGA coolers
you buy for this today are highly inefficient and only suited to high air flows,
making noise unavoidable.

So the problem is not just the fans. Cooling solutions on sale today seem to
depend on "looks" and not efficient, functional design that will make them
whisper quiet at low air flows (< 1 m/sec), even for getting rid of 175 watts
from a VGA.
A good example is in some new motherboards with all sorts of fancy heatpipes
snaking around the CPU and MOSFETS but the cooling area (and thickness)
of the heatsink fins is almost laughable. A lot of heat is taken straight off the
heatpipe surface. I have talked to engineers involved in serious attempts to
make efficient, quiet heat transfer systems and they also wonder just where
some these heatsink manufacturers have their heads.

So the point is that you really have to redesign all these things yourself if you
want a really silent and powerful computer. Just find the bits you need and
reshape them.


Regards

phconnell