Axial Flux PMDC design motor noisy

[Highspeed]

Hi all,
I have designed a PMDC axial flux motor for an air compressor based on a toroidal stator with 3 (or 6) windings.
It's a tiny device that runs up to 40 000 rpm with a Back-emf switching mechanism.
All the performances are met : Speed, torque, power accelerations ..... But one isn't: noise !!!
and there is something strange happening:
When hooked to a very sensitive vibration analyser the unbalanced level is measured to be anywhere from 0 to 360°
when i say anywhere it is something like :
@ 20000 rpm no load : unbalanced at 52°
@ 20000 rpm half load : unbalanced at 115°
@ 20000 rpm full load ; unbalanced at 65°
@ 10000 rpm no load : unbalanced somewhere else
The only good thing is that all those unbalanced levels/phases are 100% stable for the same device and this from day one.
All the componants bearings, stator, shaft, wheel and magnet are centered within 20µm.
I tried :
Bearings design xchange
Stator design xchange
Magnets, shaft, driving mode
all that with no luck
Before giving up with this motor topology, and come back to a more classical brushless design I'd like to understand what happened and here, I still don't have any answer.
I have currently one guess but that sounds a bit crazy:
This device has a "virtual unbalance" that is turning in the opposite direction .....
>:-<

 

[jbartos]

Suggestion: Please, would you post a numerical value that would characterize the noise level? The 40,000 RPM tend to be on a high side for the &quot;quiet&quot; operation or rotation.

 

[overmod]

This may be simplistic, but what happens when you VARY the load and adjust motor for constant speed?

If you don't have access to a brake dyno of appropriate size and characteristics, you might want to gin something up that can be continuously adjusted in small, measurable increments. Then develop points on a load vs. measured angle that will give you a 'smooth curve' of the (presumably stable) angles you measure.

At the very least, this would tell you whether there is a 'maximum' angle at some load, or whether the angle is progressing around the motor continuously.

I would also like to know the power spectrum of the noise, and whether some components can be recognized as harmonics.

I'm also waiting for the answers to the first reply. Something else I'd like to see is whether magnetic (or air) bearings in this device would reduce the measured noise... and, if so, whether there are measurable periodic deflections observed which correspond to some of the frequencies observed in the noise power spectrum...

RME

 

[jbartos]

Suggestion: Visit

http://www.wfw.wtb.tue.nl/em/pdf/poster02/p02tillema.pdf

for Silent Running Electric Motor by Viscoelastic Bearing Support

http://www.dt.navy.mil/acquisition/acq/sol/N6554003R0081attach2.pdf

etc. for more info
Usually, the Navy submarines have very strict requirements on the equipment noise.
Magnetic (air) bearing will reduce the motor noise by an elimination of regular bearing noise.

 

[Highspeed]

Many thanks for those tips,

I'll go a little more in details but after the first post i did find a few things and fixed the issue.

The setup is as follow from top to bottom
Moving parts : 2 bearings - impeller (35mm diameter)- 2 poles permanent magnet. all attached to a shaft .
Stator : Toroidal 3 phases (6 sectors)
The reason of the 2 bearings on top is for them to be far from the magnetic parts avoiding eddy currents in balls
The bearings are low torque, low noise specially tuned (play & grease) for that application.
The bearings are preloaded at 3.5 N
After force fitting, the radial play left is 0<play<2.5µm.
Bearing housing is H5 (measured to be OD+4µm)
After the assembly of the moving parts, a balancing is performed :
This balancing is done at 15000 rpm / half load using a monitran acceleromter and a labview home made software
The mass that is removed at the last step is below 10µg
-------------------------------------------------------

What I mentioned in the first post is in fact true and verified : there is something that turns in the opposite direction due to the shaft that is bending
Without drawings, it is a bit hard to explain, but let me try:

The stator is a circle as well as the magnet
while rotating at high speeds , the magnet circumference is kind of &quot;glued&quot; to the stator circle that is a little bigger (air gap)
the result beeing that the center of the shaft moves in the opposite direction and that is the noise source.

So, the main issue is due to the mounting of the bearings leaving to the shaft freedom to bend.

The fix that we have right now is using a stiffer shaft but drives the price high.

A new drawing is on the way to move one bearing but the trade off will be to avoid magnetic lines to go in the bearings and that is a challenge because the induction in the system goes up to 2.5 Tesla and there is not alot of room for magnetic barriers/iron thickness.


---------------------------------------------
Magnetic bearings note:

It is something we considered early in the design and we have a prototype running up to 8000 rpm
Without the constraints size/COG, it is the ultimate solution for noise free devices.

---------------------------------
Final device performance :

With the stiffer shaft, and every thing assembled together :
Motor/impeller inside a housing suspended with viscoelatic silicon inside a box with acoustic foams/circuits :
Normal shaft : 33 dBA at 1 meter at max speed/max load
Stiffer shaft : 27.5 dBA &quot;&quot; &quot;&quot; &quot;&quot; &quot;&quot; &quot;&quot;
all that measured in an acoustic chamber
27.5 dBA : hard to say if the device is &quot;on&quot; !!!!