PDA

View Full Version : Exploring magnetic gears, with video



Evan
07-31-2008, 08:26 AM
I posted a thread a while ago about a new type of magnetic transmission. Since then I have been intrigued by the idea of non contact magnetic gears. I decided to make a test stand that I can use to try out some of these configurations.

http://vts.bc.ca/pics4/magear1.jpg


http://vts.bc.ca/pics4/magear2.jpg

In the first photo the pinion to driven gear spacing is set large to better show the parts. Normally it would run with the minimum possible gap for maximum torque. Several thing about this have surprised me. There is very little cogging action in the drive. The torque transmission capability is considerable. Below the slip torque the drive is positive, no slip and quite stiff.

The ratio is set by the number of poles, not the relative diameters of the pinion and driven gear. In this demo there are 2 poles (N-S) on the pinion and 14 on the driven gear. This gives a ratio of 7 to 1. The pinion makes one complete revolution as the driven gear moves the distance between magnets on it's perimeter.

While the best name for this sort of drive seems to be "magnetic gears" there is little real similarity to purely mechanical gears with meshing teeth other than the obvious visual aspect. Ratio is independent of diameter. Torque is independent of ratio.

This is a first effort and I am surprised at how well it works. There does seem to be something analogous to a pitch circle and it seems to be somewhere in space past the OD of each gear. This pair of gears operates much more smoothly with a large gap than it does with a very small gap. It may be important what the spacing is of the magnets on the driven gear, more experiment is called for. Also, these "gears" have no pole pieces to concentrate the fields. One thing is already obvious, the driven gear may be placed in any planar alignment up to and including a right angle drive. More experiments...

I used Lexan for the test stand to minimize unwanted magnetic interaction and because it makes it much easier to see what is going on.

To really appreciate how this works you need to watch the short video. It shows the gears operating at speed and at a very slow crawl so you can see the relationship between the pinion and driven gear rotation.

1.7 megabyte
http://www3.telus.net/metalshopborealis/magear.wmv

hornluv
07-31-2008, 10:11 AM
That's pretty cool, Evan. I noticed a slight speeding up and slowing down of the gears when run at slow speeds, presumably because of the magnetic fields interacting. Is that obvious at high speeds as well or does momentum cover it up. Also, aren't you at all concerned with having powerful magnets that close to your computer? :D

Lew Hartswick
07-31-2008, 11:07 AM
Does the "timing" get out when a sudden stop occurs? Due to the
inertia of the driven one. That is the only possible "gotcha" I can see.
...lew...

dp
07-31-2008, 11:23 AM
That particular design looks like it would have some noticeable cogging because of the variable gap between the driver's single magnet surfaces and the driven wheel. This would grow worse as the load increases and finally it would begin to skip. But - the possibilities for this open coupled power transmission are interesting. Unlike gears, there's no reason the wheels cannot have a good deal of overlap which would result in interactions at two nodes. There's also no particular reason there need be a single ring of magnets on a wheel which would introduce a third node, and a forth if each wheel had two rings each.

Evan
07-31-2008, 11:49 AM
The most surprising thing is the lack of cogging. If you turn it very slowly by hand the large gear smoothly follows the rotation of the pinion. There is a noticeable difference in the torque which produces a torque ripple but not visible cogging. I suspect that when I make another gear today using more magnets and a different spacing that will also go away. There are so many possible configurations I don't know where to start. For instance, can I have a 2 to one reduction from similar size wheels if one has twice as many poles? I guess probably.

Lew, the timing does get scrambled if the drive slips.


Also, aren't you at all concerned with having powerful magnets that close to your computer? No. I don't use magnetic media except the hard drives and they have a pair of super magnets inside. The screens are TFT LCD so they are not affected.

Evan
07-31-2008, 03:39 PM
Here is a new configuration which has much greater torque. This is now more than a toy as this configuration can do useful work. It has a holding force of about 20 to 24 inch ounces and the torque value is a bit under that.

This uses a smaller driven wheel with 8 magnets. The magnets are arranged with alternating pole directions instead of all the same. The ratio has become half what it was and is now 4 to 1. I still haven't tried it with pole pieces.

http://vts.bc.ca/pics4/magear3.jpg

Here is another short clip showing it lifting an object on the output shaft that extends from the back side of the jig.

http://www3.telus.net/metalshopborealis/magear2.wmv

Doc Nickel
07-31-2008, 06:17 PM
Now, I'm certainly no expert with magnets, but I thought those disc-type NIB "super" magnets had their poles in the flat faces. If that's the case, wouldn't a "gear" train like this be stronger if the discs were mounted face-out?

Or do you need the "thinner" part of the field to make this work?

Next up, how about bar-type elements? I recall seeing some NIB types 3/8" square and 2" long- one presumes that with more... field "area" for want of a better term, the transmittable torque would be just that much greater.

And that makes me wonder if you could reduce the "cogging" even further by having two "gears" on common shafts, offset one spacing.

And interesting project. Looking forward to more.

Doc.

Evan
07-31-2008, 06:39 PM
They do have the poles on the faces. For this sort of application both poles of each magnet must be brought into play at once and to do that placing them as shown accomplishes that. Rectangular magnets would certainly be better for some configurations. I don't have any on hand though.

[edit]

I will be trying them on the flat on larger wheels with more poles. That will require the use of pole pieces.

wierdscience
07-31-2008, 09:23 PM
Interesting,any idea what the diffrence is in mechanical loss between it and a geared transmission of the same ratio and output?


Looks promising,if you scale it up and use some of these you could replace the tranny in the Rover:D

http://www.emovendo.net/magnet/6-x-4-x-2-block.html

darryl
08-14-2008, 03:57 PM
finally, a decent fridge magnet. :)

Magnetic gears, very interesting. It occurs to me that since magnetic materials can be magnetized and de-magnetized, that one could create his own pattern of magnetism in a ring or disc, and tailor the number of 'teeth' to suit the application. I have de-magnetized nibs by heating, then remagnetized them both with other magnets and with electromagnets. Since many materials have a preferred direction of magnetization you have to stay with that, but the polarity of magnetization can be reversed. This means that an ordinary disc magnet can be modified to have a multiple of poles spaced around it's surface. Something for suitably certified experimenters to play with.

I'm reminded of the high-ratio gearing setup, I think it's called harmonic gearing. Such a setup has a lot of teeth in contact at one time, and extending this concept to magnetic gearing would mean that a high ratio could be achieved with a very strong torque transfer capability.

There's also the application where a container of sorts has to be completely isolated from another space, and where a set of mixer blades has to be spun, or similar. I've seen magnetic coupling used for this, and with todays strong magnets it might broaden the range of applications for isolated actuation. Where you previously had to use a shaft with seals, you could get away with not needing either. Could be useful in flow control valves, for instance.

There are a large number of shapes available in very small sizes with these neodymium magnets, so some very compact mechanisms could be made up. I would bet that a nifty planetary gearing setup would be useful.

Just throwing in some ideas.

Evan
08-14-2008, 05:07 PM
I already thought of planetary gears, done right a lot of magnets would be interacting at once. I made a new multi pole wheel for the demonstrator and it has greatly increased torque capability. It will lift 4 lbs on the output shaft, a non trivial amount.

http://vts.bc.ca/pics4/maggearbig1.jpg


http://vts.bc.ca/pics4/maggearbig2.jpg


http://vts.bc.ca/pics4/maggearbig3.jpg

I'm going to have to order some more magnets as I have run out of stock.

davidh
08-14-2008, 06:14 PM
"It will lift 4 lbs on the output shaft, a non trivial amount."

what does that figure in inch lbs. of torque ?

Evan
08-14-2008, 06:28 PM
I haven't measured it yet exactly but based on the OD of the shaft with the twine wound on it I would say at least 2 inch pounds of torque. The holding torque is quite a bit higher, that is the torque required to make the gear slip a "tooth" when stopped.

Lew Hartswick
08-14-2008, 06:29 PM
There's also the application where a container of sorts has to be completely isolated from another space, and where a set of mixer blades has to be spun, or similar. I've seen magnetic coupling used for this, and with todays strong magnets it might broaden the range of applications for isolated actuation. Where you previously had to use a shaft with seals, you could get away with not needing either. Could be useful in flow control valves, for instance..

They have been using magnetic couplings through the walls of vacuum
chambers for years in chem labs. course they were all 1 : ! ratio.
Now for another thing for Evan to try is reverse the operation.
Try increasing the speed of the output.
...lew...

Evan
08-14-2008, 06:44 PM
Now for another thing for Evan to try is reverse the operation.
Try increasing the speed of the output.


The ratio on that latest gear is 9 to 1. With that weight on it when I remove the power it will back drive the motor no problem. It works just as well either direction. I'm considering using something like this for a gear train in the vertical wind turbine I will be building. It is silent, low loss and zero maintenance. As well, it has built in overload protection if the wind speed is too high.

dp
08-14-2008, 09:25 PM
Evan - how is the linearity with that latest rig? I'd guess shaft orientation is unimportant within reason, so you have a coupler that can be used at very unusual shaft offsets and orientations, and obviously in different enclosed spaces.

Evan
08-14-2008, 10:01 PM
You will note that I added an articulated arm. That is to test how well it can drive at varying angles. As it turns out it will drive at nearly any angle from perpendicular to parallel, side by side, side by face or whatever you want. The degree of coupling varies depending on the arrangement but it is in effect a very wide angle universal joint with no issues about binding or locking (greater than 180 around Z and about 90 around X in relation to a normal spur gear orientation) .

JRouche
08-14-2008, 10:15 PM
Great experimenting Evan... I was trying to figure out a possible use. Like why would you use it over standard coupling like gears or better yet belting??

Now you have given a use. I still dont see the advantage, and Im sure there is one otherwise you wouldnt be doing it :)

I see the silent aspect. Low loss? I see friction with belts or gears. But Ol some one's laws of physics, dont know their name LOL They kinda stated you dont get something for nuthin, or whatever LOL

I see there isnt any friction loss. But coupling loss has to be accounted for. So if we were to take two similar drive systems and calculate the frictional losses from belts. Then take the a similar mag drive and calculate the loss through de-coupling that would be interesting..

I imagine it has a break off point for the mag coupling. When you try to overdrive it yer loss increases dramatically. If the load is kept low, within its peak performance envelope the loss should be minimal, maybe less than a conventional drive system??

Either way, I love it all!! JR






I'm considering using something like this for a gear train in the vertical wind turbine I will be building. It is silent, low loss and zero maintenance. As well, it has built in overload protection if the wind speed is too high.

Evan
08-14-2008, 10:35 PM
There are losses in a magnetic system of course. They can be extremely low compared to regular frictional losses. The only losses aside from the ball bearings and windage are eddy current losses in the magnets as the fields impinge upon each other. This is proportional to the change in internal field in the magnet and inversely proportional to the degree the magnet body is a conductor.

The change in internal field is proportional to the magnetic permeability of the magnet itself. In a superconductor the permeability is zero so there are no eddy current losses although another mechanism comes into play known as "flux pinning". In a NIB the permeability is non-zero although I don't know what it is. I also don't know what the electrical resistance is of the NIB compound but I suspect it is high. That will determine the level of eddy current losses as will the rate of change of the fields.

In all, I am sure that the losses are extremely low up to the point that the magnets "slip". Then the losses increase but not that much. Drive ceases but energy put in is given back by the magnetic "spring" action as the poles pass each other.

rdfeil
08-15-2008, 12:35 AM
JRouche,

One use for the magnetic coupling that I have seen in use is on a refrigerant pump. The liquid transfer pumps on one Freon refrigeration system are 1/2 Hp and use magnetic coupling through a non-magnetic metal housing to drive the gear pump. The advantage is no seal that can leak. Neat little pumps.

Robin

2ManyHobbies
08-15-2008, 11:02 AM
I was reading about SRMs the other day and thinking in the direction of electric motorcycles. A single or even compound magnetic gear coupling could save lots of wear and tear on a drive from a 10k+ rpm SRM. I'd imagine that you could then go with an SRM that has built in magnetic bearings and runs totally sealed. As an added benefit, regenerative braking and possibly something akin to boost could be added to the magnetic gearing. Unless of course I am mistaken and regenerative can be done via SRM...

It would have the nifty factor of a clip-on black box with a power coupling that when removed causes the whole drive-train to free-wheel...

Evan
08-15-2008, 11:16 AM
Simple magnetic couplings are trivial to make and capable of high torque loads. I was experimenting with a pair of plain ceramic disc magnets about 3" in diameter from the pancake motors of a pair of identical 5 1/4" floppy drives. When spaced only a few thousandths apart is is nearly impossible to make them "skip" poles by hand. I haven't gone any further with that but will probably make them into a proper coupling. A coupling that can slip indefinitely at a certain breakaway torque without wear or damage can be very useful. The torque can be adjusted by changing the spacing. I was thinking I might put it on the electric leadscrew drive on my SB9. Presently I rely on the belt to jump teeth if it is stalled (hits carriage stop) which is a less than optimal solution.

SRM?? Solid Rocket Motor? :D

2ManyHobbies
08-15-2008, 11:35 AM
Switched Reluctance Motor
I think they are driving them as high as the 50k-100k rpm range. The easy version is using coils to switch the magnetic field around the stator with a rotor built in such a manner that it attempts to short the magnetic fields of the stator. Almost like a stepper motor, but SRMs don't usually have a magnet in them, just transformer like plates cut and stacked to make the rotor.

Evan
08-15-2008, 01:51 PM
I am familiar with switched reluctance motors and have worked with them. I'm not accustomed to that abbreviation though. They are also known as variable reluctance steppers. I have several that I am considering using as servo motors. I was thinking of putting one on the Z axis of my mill. It would really be nice to be able to run the head up and down at maybe 500 ipm. :D I used a couple of them in about 1985 to build a flat bed plotter from scratch. I was getting maximum plot speeds of 2000 steps per second with a 1 (one) megahertz 6502 controller. At 12 steps per rev that's 10,000 rpm. I had to write special multi stage acceleration software to make it go that fast. It would shake the table when it was plotting.

2ManyHobbies
08-15-2008, 03:06 PM
Lol, I bet that tended to keep the curious at a safe distance. Did it sound like an angry stump grinder?