Mars Electric LLC Axial Flux Motor

I'll check it when I get it. I have a decently accurate lathe and dial indicators. I'll chuck up the rotor and measure it.
 
Put some holes in the rotor while you are at it, give it better spool up times :twisted:


So how could the stator be constructed better on this motor? Maybe dig up some of the old material from the other main discussion thread? http://endless-sphere.com/forums/viewtopic.php?f=28&t=13957&hilit=+axial+flux It seems to me that with the tooling and capabilities of Mars they could put a pretty slick motor into production.
 
johnrobholmes said:
Every time a magnet passes a pole there is a little thunk. Add up all the times that happens in one revolution, multiply that by the rpm, and we get a nice howl!
Click to expand...
Then my guess is that this noise (and that of possibly a number of other motors used on ebikes/etc.) is caused by not-so-precise placement of magnets relative to stator poles, in that not every one is placed exactly symmetrically relative to all the others. Thus there is a time delay between every pole's magnet-interaction noise, and this creates even more noise than it would if they all happened at exactly the same time.

If they were synchronized, the noise would be louder, possibly, but only at a single frequency (or harmonics) and not just a general rumbling of many frequencies (and their harmonics) together.

Would be interesting to be able to precisely place all the poles/magnets so they interact simultaneously, and hear/see the results.


BTW, those axial-flux brushed radiator fan motors I used on DayGlo Avenger had a similar rumble; they are ironless, with no laminations, just coils embedded in an epoxy ring, with a single ring magnet on one side of the casing.
 
Hey folks,

Well, we got some recommendations for getting this motor to run better. I spoke with John Fiorenza today,
and he recommends matching up the right controller to smooth it out:

"The problem is that the control is switching a very powerful motor on and off very
quickly. The energy stored in the windings needs to go somewhere, and it is
generating back into the battery/capacitor. Due to the inductance of the motor,
this regenerative energy is not happening quick enough, and it is causing audible
cogging. A sine wave control would not have this problem.

If the control can not be changed, then I need to skew the windings or change the
magnet shape to round magnets, giving up some of the peak torque. This would make
the motor quiet. "

But, as some people have expressed, maybe the natural sound is ok?
Maybe we can think of it as the sound of condensed power?

SO, We are trying out the recommended controller, and have another one coming at us, from, May I Dare Say, Randy Draper!

I have spoken with Randy, and he is interested in getting some of these motors tuned for our needs,
and has a special controller that he recommends...So, I am sending him a unit to try out on his E bikes,
and he is sending me a new controller to wring this motor out with. Can't wait to get it on a machine and running.
Seems after ten years, he is still interested in helping out the E bike community with his recommendations...

Side note, Here is a little article in the local school paper on me http://www.collegian.com/index.php/article/2010/11/113010_its_electric_charging_the_cycling_world

Peace, Josh K.
 
Josh K. said:
SO, We are trying out the recommended controller, and have another one coming at us, from, May I Dare Say, Randy Draper!

I have spoken with Randy, and he is interested in getting some of these motors tuned for our needs,
and has a special controller that he recommends...So, I am sending him a unit to try out on his E bikes,
and he is sending me a new controller to wring this motor out with. Can't wait to get it on a machine and running.
Seems after ten years, he is still interested in helping out the E bike community with his recommendations...
Click to expand...

I doubt the rather rude and offensive Randy Draper has the knowledge or understanding to be able to do anything more than act as a middle man for someone who does know what their doing, but would be happy to be proved wrong. If he comes up with a controller design that works, then more power to his elbow.

As an aside, the noise issue isn't unique to this motor. The bigger brushless Mars motor also makes a fair bit of noise. I tried to help a friend reduce the noise on his Mars brushless installation. He tried different controllers, with differing PWM frequencies, to no avail, but did find that physically changing the mass/stiffness of the rotor made a difference. His conclusion was that the rotor disc was deflecting and ringing at the commutation frequency (not the PWM frequency, as that's effectively damped by the motor inductance).

I'm pretty sure that this is an inevitable consequence of a single sided axial topology motor; a disc rotor is going to be less stiff than a cylindrical one and so more likely to locally deflect with each stator pole energising pulse and act a little like a loudspeaker diaphragm. Sine drive will probably make the sound less harsh, by removing the higher frequency components, but my guess is that it'll still be noisier than a radial design.

Jeremy
 
It's no surprise that Draper can source parts. He's found some spendy mil-spec bits before, but the costs prolly defeat the merits of a low-cost AF motor. (Miles may have one of those little gold bricks; I couldn't find an uploaded pic.) EDIT: pic was here

Jeremy's theory seems reasonable... There is likely enough flex in the rotor to sonify cycles. I presume that outrunners can sing a bit louder than inrunners for the same reason (I don't own either, so mebee somebody can confirm).

If I understand correctly, the motor resistance & inductance are so low that the phases are hammering/getting-hammered with each cycle? Any other workarounds, other than sine-control or round mags? (IIRC, adding caps is absurd, yes?)
 
Let the games begin ....Pikes Peak or Maui

I will try to have an entry at both
 
I figured that round magnets or skewing the stator would help, as would sine control. For making this an affordable drive combo, it seems that block commutation makes the most sense unless he has something up his sleeve.
 
Congrats Josh! I didn't realize that you had your own shop, very cool. Even the Optibike guy commented on your article.

John,
Is this thing anywhere near as loud as a medium sized outrunner?
 
Yeah, Jeremy you should stay on ES it's awesome! And Dodjob offered you a free beer in Bayern! That's as almost as good as it get's in my opinion! Anyhow you could have a glass of wine or bottle of wine on me too in sunny New Mexico if it seals the deal.
:shock:
 
TylerDurden said:
Jeremy's theory seems reasonable... There is likely enough flex in the rotor to sonify cycles. I presume that outrunners can sing a bit louder than inrunners for the same reason (I don't own either, so mebee somebody can confirm).

If I understand correctly, the motor resistance & inductance are so low that the phases are hammering/getting-hammered with each cycle? Any other workarounds, other than sine-control or round mags? (IIRC, adding caps is absurd, yes?)
Click to expand...

In my experience outrunners can be pretty noisy too, certainly noisier than inrunners. When I completely filled the spaces between the magnets on one of mine, with high temp epoxy, it made it noticeably quieter. Adding the extra epoxy both stiffened the can and also added a damping medium, which supports the theory that it's the can making the noise. My guess is that the flat rotor disc in an axial motor is going to be tougher to stiffen up, but probably not impossible to do.

Jeremy
 
http://www.kidlogger.com

In the future, this just saves everything you type on the computer, you'll never lose any of it again

Jeremy Harris said:
Well, I'm a bit pissed off.
Goodbye.

Jeremy
Click to expand...
 
Maybe it's not enough to have a spacer between rotor and stator? What if the rotor get's a repelling magnetic impulse and want to move the other way? Is this side of the rotor fixed, too?
There is a low voltage sinewave controller in germany from www.sinusleistungssteller.de
Maybe not the cheapest, but for sure an option.
Adding external coils in series would help too, I guess :?
-Olaf
 
Reply to the test video ....

"If you give me a call, then I will send you a quiet control for your experiments. John Fiorenza, Mars Electric LLC, 262-644-7525 .......

....1 week ago

The sensorless control must be a 6-step BLDC control and I do not think it has sine wave excitation. The motor is a PMSM motor with a sinusoidal back-EMF, not Trapazoidal. It would be much quieter if you used a sensor-type speed or torque control."

http://www.youtube.com/watch?v=ISxy9RpALeI&feature=player_embedded


anyone following up on this ?
 
I haven't heard anything on my end, but I don't have a motor here anyway. All of the controllers I tried were 6 step commutation, and I tried both sensored and sensorless in multiple brands and switching frequencies. I do think the only way to have the motor run quiet without modification would be a sine drive.


But BigMoose has a scope and will show the output when he gets time. I bet it is a pretty blocky looking output, totally shooting from the hip.
 
jmygann said:
Reply to the test video ....

"If you give me a call, then I will send you a quiet control for your experiments. John Fiorenza, Mars Electric LLC, 262-644-7525 .......

....1 week ago

The sensorless control must be a 6-step BLDC control and I do not think it has sine wave excitation. The motor is a PMSM motor with a sinusoidal back-EMF, not Trapazoidal. It would be much quieter if you used a sensor-type speed or torque control."

http://www.youtube.com/watch?v=ISxy9RpALeI&feature=player_embedded


anyone following up on this ?
Click to expand...

Pretty much all the brushless motors we typically use have a roughly sinusoidal BEMF and the vast majority of controllers use 6 step commutation; whether they use Hall sensors or BEMF sensing makes no difference to the commutation waveform. I've looked at the BEMF waveform from a few BLDC motors and it always seems to be near enough sinusoidal, so this motor is just like others in this regard.

A sine controller will reduce or remove the higher frequency harmonics from the drive current, but may not massively reduce the overall noise if it comes from commutation frequency excitation of the rotor as I believe that it does. Sine drive usually needs better rotor position information than you can get from just having three sensors on the motor, too, as the controller needs to know the precise angular position of the rotor in order to ensure that the right current drive is being provided to each winding. This is often done by using an encoder fitted to the motor shaft, but it can be done by BEMF sensing with a sensorless controller. This will use the relative amplitude of the BEMF on the non-powered phase to derive the angular position data needed to drive the two driven phases at the correct relative current level. I believe that it should be possible to crudely sense angular position by interpolating between Hall sensor outputs to derive position, but would guess that this suffers from inaccuracies when the angular rate changes quickly, such as when accelerating or decelerating, which are just the times when you want accurate commutation for good torque and low torque ripple.

Jeremy
 
Guys, I took all the magnetic and runout data at the end of last week. I was in the process of setting up for the back emf measurement when I went up to Church to assist in the installation of a new sound system. Let's say that it has not gone as planned... I've been living inside a couple of equipment racks for last 3 days trying to sort out and reorder reams and reams of audio cabling and trying to get the new digital links talking to each other. It's been a bit of a daunting task for the ol' guy.

If things go as planned, I will be free Wednesday to finish the back EMF measurements and prepare the posts on what I have measured to date. My apologies for the delay in posting.

I now question the wisdom of our group decision to upgrade right before Christmas... We barely functioned this past Sunday, and next Sunday is the Children's Christmas Program... We cannot let them down and need full functionality back by then. I think I got all the house and monitor wiring sorted out last night, now "all" I have to do is get the digital links talking to each other for recording and playback...
 
olaf-lampe said:
There are two different hall sensors: latched and analogue. I guess the latched version we mostly use isn't useable for interpolation, Jeremy?
-Olaf
Click to expand...

It shouldn't make any difference, as you'd want to work of the edges of adjacent sensor pulses. I haven't tried it, but my guess would be that you'd measure the time between adjacent sensor edges and then convert this from a linear to a sine function and use it to drive the PWM signal. It'd take a fair bit of processing time to do accurately and would always be one electrical revolution behind the actual motor I think, because I think you'd need to divide the time between the previous pair of sensor edges up to derive the sine values for the next electrical revolution.

There are probably better ways of doing this, but off the top of my head this is pretty much the only way I think you can do it with Hall sensors and still retain an ability to deal with rapidly changing motor speed. Fitting an encoder with better resolution is the foolproof way to do it, and most probably the way that most sine controllers work, I think.

Jeremy
 
Well, lets try to get some of the data on this motor published, so to speak.

First runout.

Axial runout 0.0014 inches

View attachment 4

Radial runout 0.0058 inches

View attachment 3

12 magnets and 18 poles. 33 laminations across 0.694 inches, so approximate lamination thickness 0.021 inches per lamination and insulation. Trying to figure out how the stator laminations are made is perplexing to me. I laid a straight edge across the OD of the lams and they are flat and not curved as an arc. But perhaps they spring back when the winding groove is cut?

A close up of the screw thread in the backside shows the laminations in the screw thread as in the pix below:

View attachment 2

The two pix below are the two sides of one lamination pie slice. Note that the bulge on one side translates to the divot on the other side of the pie. This feature does not ripple throughout the stator however.

View attachment 1



There is an irregularity in the stator pie wedges as follows:
Lamination Number OD of Wedge ID of Wedge

1 0.588 0.376
2 0.563 0.370
3 0.583 0.381
4 0.588 0.387
5 0.583 0.376
6 0.578 0.377
7 0.586 0.382
8 0.591 0.378
9 0.585 0.379
10 0.586 0.384
11 0.587 0.384
12 0.586 0.383
13 0.584 0.390
14 0.582 0.384
15 0.594 0.384
16 0.603 0.418
17 0.580 0.364
18 0.596 0.395

Outlier are in bold.
 
Phase leads have two strands of 0.0445 inch diameter magnet wire. This is likely 2 strands of #18 gauge wire. The motor is wired as a Y or star, with the splice below:

View attachment 3

All magnetic readings were taken with a Bell 9640 Gaussmeter.

Magnet surface of rotor was 4300 to 4200 gauss. Back iron on the rotor was ONLY 2 gauss or less. So there is no saturation of the rotor back iron. I was a bit surprised at the surface reading being that high for this motor.

View attachment 2

I energized the phases with a 1 amp constant current source and measured the magnetic field on the stator poles. This may give some idea of the winding pattern: field strength in Gauss.



Stator Tooth U V W
1 14 14 0
2 0 -14 -14
3 -14 0 14
4 14 14 0
5 0 -14 14(might be -14)
6 -14 0 14
7 14 14 0
8 0 -14 -14
9 -14 0 14
10 14 14 0
11 0 -14 -14
12-14 0 14
13 14 14 0
14 0 -14 -14
15 -14 0 14
16 14 14 0
17 0 -14 -14
18 -14 0 14

Reading the field on the back of the lamination stack, away from the screw holes showed about 2 Gauss.

I can't tell for sure, but is seems like there are 6, 7 or 8 turns per winding. I didn't want to pry them apart to count.

View attachment 1
 
With the rotor removed I measured the phase resistances and inductance with an HP 4328

Data as follows:

U V W
100 Hz mOhms 73.5 73.8 72.4
100 Hz mOhms 73.4 74.1 72.6

100 Hz uHenries 146.2 159.6 153.3

1KHz uHenries 131.3 144.9 138.0

10 KHz u Henries 118.2 130.5 122.5

100 KHz u Henries 85.8 91.3 87.8

With the rotor installed the following measurements were made with the rotor in "detent"

100 Hz mOhms 99.5 71.7 98.7

1KHz mOhms 114 118 128

1 KHz uHenries 124.7 141.8 135.6

With the rotor installed and the rotor stabilized (locked) between the detents:

1Khz u Henries 123 148 129

For comparison I measured the phase inductance and resistance on the big Turnigy C80100-130:

View attachment 1

U V W
1KHz uHenries 32.5 39.1 46.9
100 Hz mOhms 17.3 20.6 25.6

Now for the Turnigy TC6374-200



UVW
1 KHz uHenries 39.4 34.9 29.5

100 Hz mOhms 39.0 35.3 31.1


I will motor these three motors tomorrow and post the scope traces of the EMF waveforms.

Hope this helps us all some.
 
I don't think I mentioned the hall sensors. There are three molded into a block screwed to the housing cover.

View attachment 3

A closeup of the molded sensors in a block:

View attachment 2

Now the halls have their own magnetized ring magnet to trip them as shown below:

View attachment 1



In the pix above, I measured the N-S transitions. You will note that there are 12 poles, just like it should be to match the rotor. Pie transitions were as follows: Made with a protractor and measured by eye, may not equal 360 degrees.

Pie number included angle (note there is going to be some timing drift because of unequal segments.)

1 29
2 33
3 28
4 29
5 32
6 28
7 30
8 32
9 29
10 33
11 29
12 28

I'll let you guys have a hand at analyzing the data. :D
 
You must log in or register to reply here.

Similar threads

E
    • Like
Eldridge Axial Flux Motor/Generator
Replies
19
Views
4,672
Logic11
L
E
    • Like
Homemade Axial flux PM motor
2
Replies
27
Views
12,447
Erik952
E
T
Multistage axial flux motor design and prototyping
2
Replies
25
Views
6,769
larsb
L
spinningmagnets
Axial Flux Reluctance motor, pondering DIY layout
Replies
7
Views
3,323
everythingisawave
everythingisawave
P
Designing an axial flux motor
Replies
5
Views
2,482
Kingfish
Kingfish
Back
Top