Miles' 90mm inrunner build thread

[Honk]

You still haven't justified tripling the iron losses....

You misunderstand me. :wink:
I didn't tripple the iron loss for fun. I just estimated three time higher no-load current
from experience as iron loss alone is not equal to full no-load loss.

Bearing losses at no load should be practically nothing...

Not true, Viscous bearing losses is RPM dependent and requires squared power input to overcome the drag going up by RPM.
I can tell you about my latest build. It's a small experimental powder core motor and it should really have
ultra low no-load losses due to the almost non existent losses of these materials.
I initially had 0.65Amp in no-load at 8000rpm. This was using a selected high quality low drag 8x19mm bearing.
When I swapped bearings to really small RC flight bearing 3x5mm the no-load dropt to 0.11amps.
This is the importance of low drag bearings.....a lot more than people considers.

 

[Miles]

Oh I understood that. I just thought it was a bit extreme...

50 Watts of stray, bearing and windage losses?

 

[Miles]

Bearing losses at no load should be practically nothing...

Actually, it was bearing what wrote that So, it was bearing's loss... :lol:

 

[Honk]

Sorry, I was in a hurry to answer and got it wrong when copy and pasting.....
I have fixed my error... :lol:

 

[Miles]

No problem..... It gave me an excuse for a pun

 

[Miles]

I can tell you about my latest build. It's a small experimental powder core motor and it should really have
ultra low no-load losses due to the almost non existent losses of these materials.

I remember following your thread about it on RC groups. Really looking forward to hearing about your tests.

 

[Honk]

Hehehe, I'm remembered....

Yes, the project is still going on but becoming a father have taken so much more time than I could ever imagine.
The project was standstill for long periods of time but never forgotten or abandoned.
Meanwhile I was tinkering on new ideas and improvements and put them in text documents so they would come to good use.
I also upgraded my mechanical equipment with everything needed to finish the project.

At the moment I'm about to mount the sinusoidal controller and keep testing the small test motor.
I have pictures and videos but have been to lazy to uppload them.

The big motor have been extended from 6KW to 9KW at 3500 RPM. 8)
Hopefully I can soon get more time to start building it.

 

[bearing]

No problem..... It gave me an excuse for a pun

LOL!

I initially had 0.65Amp in no-load at 8000rpm. This was using a selected high quality low drag 8x19mm bearing.
When I swapped bearings to really small RC flight bearing 3x5mm the no-load dropt to 0.11amps.
This is the importance of low drag bearings.....a lot more than people considers.

Interesting. But for me it sounds like the bigger bearing either was a damaged, or it was a poor choice from the beginning. Do you have the part number?
I've taken a course which covered bearing selection. It's not very hard to do, but if you don't have the knowledge, it's probably possible to make bad choices.

 

[Honk]

I don't remember which type it was right now but I can guarantee you that the bearing was 100% OK (and very low drag).
But the low drag was nothing compared to the ultra small RC bearing that had almost non existent drag.
For comparison, when finger twisting the shaft with the larger bearing it would freely spinn for a pretty long time before stop.
But spinning the motor with the RC bearing it would go on and on before coming to a stop.
That difference resulted in going down from 0.65A to 0.11A No-Load at 8K RPM.

Now it's time to go home. I'll keep up with answers later tonight.

 

[bearing]

Was that at 48V? Thats like 30W. That bearing must have been smoking hot!

30W at 8000 rpm is 35 Nmm, which is the equivalent of a static drag which would hold a 350gram weight on a 10mm lever.

Is it possible that the large bearing was exposed the magnetic field, and had induced eddy currents which created the drag?

 

[Honk]

No, it was at 18-19V, if I remember correctly.
It was an outrunner and the bearing was inside at center of the motor.
The entire motor was getting mildly warm but I never investigated exactly if there was a hotspot.
It was kind of difficult when the bearing was mounted inside.
But all no-load problems was solved by the bearing swap. 8)

 

[spinningmagnets]

Is it possible that the large bearing was exposed the magnetic field, and had induced eddy currents which created the drag?

I had always felt that the lower-friction of ceramic bearings was an un-necessary upgrade for E-bike hub-motors, but now...I wonder if ceramic balls would avoid eddy-currents in a non-hub motor...thoughts?

 

[bearing]

No, it was at 18-19V, if I remember correctly.
It was an outrunner and the bearing was inside at center of the motor.
The entire motor was getting mildly warm but I never investigated exactly if there was a hotspot.
It was kind of difficult when the bearing was mounted inside.
But all no-load problems was solved by the bearing swap. 8)

That's still the equivalent of more than 100grams on a 10mm lever.

Was the bearing seated directly at the inside diameter of the core?
The flux in the core may decide to go through the bearing. And since it's not laminated, it will get hot.

 

[Honk]

The bearing was 100℅ shielded from any magnetic field.
It was sitting at center of a permalloy core with really thick walls.
I have a fluxmeter and measured flux levels at several positions, at center there was zero flux.
What you forget is the force needed to overcome the slightest drag at high rpms.
The only difference was the drag difference between bearings.

A simpel comparison, take a fast spinning motor with low no-load, then brake it slighty
by your fingers and you'll see the load current rapidly increase, the same with bearing drag.
But you knew that already, of course.

But enough of this now....Miles should have his thread focused on his own project.

"Sorry for any bad spelling and sentences, my kid is fully disturbing me as I reply" :lol:

 

[speedmd]

It most likely not bearing size, but grease, seals and badly fitting ball separator that are adding bearing drag. You only need the external side shield/seal and a light oil to check. Also, if the stack dimensions are causing side loads on the bearing races, they will drag badly. Also some are built with too much preload. Balls need to run in the center of the deep v race grove to run free/ lowest drag.

 

[Honk]

In this case I don't agree as I know first hand the importance of reducing drag.
It was 100% the bearing drag that caused the higher no-load. And that was still a really god bearing.

Enough now, we disturb Miles thread. No more replies on this matter....please.

 

[Miles]

Enough now, we disturb Miles thread.

No problem. I don't mind the occasional detour

 

[Miles]

One thing that I experimented with in Emetor was changing the "magnet angle" (magnet coverage). Even quite small changes in magnet angle can make a significant difference to the amount of torque ripple. Work at a low torque output and iterate down to the lowest value.

 

[Honk]

Enough now, we disturb Miles thread.

No problem. I don't mind the occasional detour

Thanks Miles.

Btw, yesterday evening I was thinking about the bearings of my build. Why it had such impact!
Well, it's simply that powder cores have virtually no "iron losses" at the low commutation speeds of a motor.
Therefore the no-load can be seen as all other drag factors within the motor, like bearings, windage and similar.
The bearing had the utmost impact as the other parameters was small in comparison.

An iron laminated motor have a much higher level of hysteresis and eddy current loss.
In your case the part of no-load current from bearing drag is problably masked to some degree by this.

 

[Miles]

Well, it's simply that powder cores have virtually no "iron losses" at the low commutation speeds of a motor.

That's great. How does the permeability compare to silicon steel, though? I can take the pain if the gain is worth it

Most SMC materials don't seem to be worth using on conventional motor designs.

 

[Honk]

Hi. It's considerably lower than laminated steel. The stuff I use (Sendust or MPP) has 125u permeability.
For a regular teethed motor there is not really any gain using powder cores.

My design was a slotless airgap motor where I focused on high performance by strong fields and heavy duty litz wire with high fillrate.
A teethed motor has less winding space and needs physically stronger and better materials due to higher peak fields and avoiding teeth corners from chipping.
In a slotless motor the field is uniform (no peaks) and is easily contained by the flux return ring.
The ring core is only there for return of the flux through the glued airgap windings. There is no need for physically strong materials.
I measured the strength of the return flux using different materials, like pure iron and 125u cores up to 550uH cores.
The difference in return flux was minor due to the comparatively large airgap of 5mm. But the iron loss was almost non existent using good powder cores.
The worst powder core I tested was a regular cheap iron powder core type T26. But any Sendust, MPP or Hi-flux was great. No losses to speak of.

 

[bearing]

I played with the bearing loss calculator at the SKF website. Their 619/8 deep groove ball bearing, greased with thick SKF grease, will at 8000 rpm have a loss of 1,13W with 10N (1kg) axial and radial load, or 3,5W with 100N (10kg) axial and radial load. So, spinning without load in a light motor, there shouldn't be 10W losses in two bearings, probably not even 1W, unless the bearing is unsuited for the task, or has very high axial load / balance problems / other problems.

 

[Honk]

Hi, I took some time double checking my old notes on the bearings. It was actually two sets of 6000-ZZ I used.
The 6000-ZZ has dimensions 10x26x8mm. Sorry for not checking my data before posting info.
But my point was only the importance of bearing drag, at least in my project where all losses count.

The 6000-ZZ bearings was replaced by 1x MR85ZZ 5X8X2.5 Metal Shielded and 1x MR63ZZ 3X6X2.5 Metal Shielded ABEC 3.
Really small and tiny with extremely low drag. The difference was obvious ones the new bearings was fitted.

 

[bearing]

The problem may have been that you did not reach minimum load with the larger bearings. Minimum load is estimated to about 100N for an SKF 6000-2Z at 8000RPM, with the standard grease at 25°C (kr = 0.025, v = 180 mm2/s). This can make the balls slide instead of roll.
http://www.skf.com/group/products/bearings-units-housings/ball-bearings/deep-groove-ball-bearings/single-row-deep-groove-ball-bearings/minimum-load/index.html

Smaller bearings with thinner lubricants will have lower minimum load. Since the size is squared in the formula, a small change will have big bearing. (lol)

The reason why some skirt bearings on outrunners runs hot may be that they fall below minimum load.

 

[Honk]

That could be one of the reasons for the improved performance by smaller bearings.
Hard for me to tell exactly, except finding a great decrease in drag resistance.
As long as I can improve results I'm a happy man......as my slogan says: Efficiency is king