Sunday, January 24, 2010

Epic Axial Motor - IAProgress

It's been quite a while since I've done a post on the axial motor. In the last post, things were just starting to get built and the big question was how to validate the performance of the motor without actually building the whole thing. "Performance" meaning...? The predicted torque/speed curve, the thermal characteristics, the efficiency, or even just does it shred itself into tiny little pieces? I'll get back to that. First, build updates:

"Master of the EZ Trak" Mike N. put together two beautiful rotor halves worth of aluminum and steel. The magnets, as scary as they look, were actually very easy to put in. Only problem is that in the process of gluing them, the magnet O.D. changed enough that the wonderfully-machined inner surface of the magnet retaining ring didn't slip on anymore. I came up with a quick fix, but I don't think Mike will like it:

Sorry Mike. :(

I am fairly certain that this piece will need to be replaced in the next version anyway, for reasons I will get to in a bit. This first build is really meant to be a learning experience, with the possibility also of collecting some data on the back EMF. By learning the voltage generated by a single stator segment and winding, the torque/speed curve of the whole motor can be extrapolated (to a degree). Which is good, because I only have enough material and time to build one stator segment.

One problem I avoided for as long as possible but finally had to deal with is getting the conical bearings to be properly loaded, even with a single stator segment. The axial forces in a motor like this are tremendous and the imbalance of having just one steel segment makes it very difficult to line up the rotor halves and load the bearings instead of just having one half slam down on the face of the stator segment. The "solution" I pursued was one of undersized axial spacers that, when they go into tension, pull on the rotor halves with more force than the magnetic attraction of the single segment. Like this:

The problem was that we didn't design the rotor back-iron to screw into the spacers from the side. And I had already put the magnets in... Time for some very careful drilling.

On the plus side, it is self-fixturing and has built-in chip collection.

Not wanting to cut too much on the first try, I left them a little long. As expected, it was impossible to get an air gap on both sides of the stator segment. One or the other rotor half would just slam down:

In the end, I wound up making the spacers a good 0.020" undersized. This sounds huge, and I expect some of it is actually being taken up by the rotor back iron flexing. But whatever, it got the job done:

Air gap on both sides!

I'm not thrilled with this solution. Mostly because I don't think it's stiff enough to last. I am fairly convinced that what's really needed is a no-nonsense 1/2"-thick solid aluminum can that goes around the whole thing. It would do the job of both the magnet retaining rings and the spacers. I will look into this option. But for now, I need a steel can. Here's why:

(Warning: Science Content)
I learned in freshman physics (the hardest class I took at MIT?) that if something looks symmetric, it probably is. The full axial motor certainly looks symmetric. But this version with one tooth...certainly not. Well, how to trick it into thinking it's symmetric? Mostly, I care about the magnetic circuit formed between the rotor disks, stator segments, and permanent magnets. In the case of symmetry, there's no telling which rotor half is which. They should therefore be at the same magnetic potential.

With just one stator segment, though, there is no clear return path for magnetic flux and the rotor halves will rapidly shift from "N" to "S" as the motor rotates, shooting off field lines into space for lack of a better place to put them. Not good. Solution: connect them with a giant steel can. This will force them to be at the same magnetic potential, just as a wire forces two nodes to have the same voltage. This is probably a close approximation to the symmetric case with all stator segments in place.

So, I need to make a giant steel can, preferably not out of one solid piece of steel since it would be very hard to make. On radial motors, this might be called a flux jacket. Cue the giant role of sheet stock:

Don't worry, it's not a solid roll. It's only about 3/16" thick.

The dotted line is where I will be attacking it with a cutoff disk. The enormous hose clamps are to keep it rolled up. (It's under an incredible amount of tension.) They might also be useful for squeezing it down to the correct diameter. I'm sort-of winging this part, since it isn't part of the real motor, so don't expect much...

I also need a way to spin the motor so that I can measure the back EMF of the single stator winding. Chain drive seems like a good option, for reasons that will become clear momentarily. I made some quick disconnect hubs...

...which I lost for an hour in MITERS. Turns out after cutting them off I, without noticing, put them on the lathe tool holder post and then put the regular tool holder back on top of them:

No wonder I couldn't find them. Anyway, they provide a quick mounting solution for sprockets of various sizes and tooth counts. Altogether, the motor (minus its forthcoming flux jacket) looks like this now:

So all that's left is to find a suitable test rig. Normally when I test a motor, I just stick it in a drill and hook up the leads to a scope. But this is...bigger. The test rig will need enough torque to overcome the large cogging force of the single, unbalanced stator segment. And it will need to get up to a high enough speed that the rotor inertia damps out the speed ripple created by this cogging. I only have access to one machine with enough power to pull this off:

You might be surprised, but this is only the second most dangerous thing ever to be driven by the kart motor. (The kart itself is the third and last.) The first...well I don't talk about that anymore.

Actually, I'm not that worried. The biggest problem will be vibrations. It's going to vibrate like crazy with just one stator segment. I predict I will only get about 10-30 seconds of useful data collection time per run before something breaks. My money is on the spacers. When your experiment is almost guaranteed to end in catastrophic failure, you can't be disappointed. And you take appropriate safety precautions such as not being near it when it does. Everything will be recorded on high-speed video, including a scope channel with the back EMF, while I stand...somewhere else.

As for the future of this project...well, if the back EMF test fails miserably that's easy. If it actually works, well then it may or may not continue. I don't really have a use for this motor, nor does anyone else around here, really. It's meant to be cheap, easy to build, and easy to test. But for me, with my resources, it is none of those. I'm pretty sure somebody else with more skill and access to better equipment could build this motor or a better one in much less time. It's still fun, though, and a good learning experience. So, I'd like to finish it, but no guarantees.

For now, insane, one-shot, near-certain-destruction testing awaits!

1 comment:

  1. Just curious about your results.....
    Also working on axial flux 3-phase DC.