The Halbach Disc Motor Ebike Project

[safe]

Compliance

It's really hard to know sometimes if you are really producing good results or if you are going down a blind alley. What gives me more confidence is when two separate systems come up with the same results. The FEMM Simulation program can produce a number for the total force that a configuration can produce for a given current supply. My spreadsheet can take this "stall torque" value and compare it to what I expect to occur. If these two numbers match then there's a stronger possiblity that I got the math right.

So I tried it for two separate configurations (so far) and was able to get compliance within about 1% which is good.

More fiddling is required before I'm really done and I'm actually now going through the testing phase of many different designs to see what will give me what I want. Lately I'm starting to look back on the "Standard Halbach" configuration again because using "enhanced" ideas runs the risk of getting results that are unpredictable. In some cases (like with adding extra iron in the cores) you add back in things like cogging which is a waste of energy.

I'm sort of going in circles, but at least it's starting to all make sense... and that's good. (spring is a month or two away now)

[safe]

This post had crosslinked files that were unreliable, so I removed the link. In it's place I'm going to load an avatar file (to be used elsewhere) thus repeating the crosslinking but in the other direction.

avatar 004.jpg (22.07 KB) Viewed 493 times

[safe]

Maximum Power Theorem

This might apply to everyone...

Seems that in an electric circuit (like one with a battery and a motor) there is a situation such that if all the loads are balanced and have the same resistance that this creates the optimal power transfer through that circuit.

Balance is good.

How this might effect an ebike is that if the battery (or controller) is significantly out of balance with the motor resistance that you would not get optimal performance.

Now in the "real world" we live in with ebikes the efficiency is typically not very good with the average peak efficiency being less than 90% and the more realistic "real world" efficiency when in use being less than 70%. So within this mix of losses there is this issue of the Maximum Power Theorem:

http://en.wikipedia.org/wiki/Maximum_power_theorem

Load Resistance vs Motor Resistance.jpg (108.54 KB) Viewed 499 times

...what got me interested in this is when I looked at the Halbach Power Formula and in it the efficiency becomes related to the following formulas:

Halbach Formula.jpg (130.17 KB) Viewed 495 times

...if you look at their conclusion about efficiency it boils down to the same issues as the Maximum Power Theorem.

This explains why in the CSIRO motor they actually add back in some Inductance because they are toying around with this formula to get those last few percentage points of efficiency.

We can conclude that:

The standard heat equation (I^2*R) represents the "optimal" power transfer possible for a motor. (load resistance being balanced with motor resistance)

It's probably "okay" to use the I^2*R formula because the results are "good enough" for estimating winding options, but to take things to the next level you would need to model the entire system including the battery and controller.

Anyone who has tried to ride with an SLA battery when it is cold will quickly realize that it's increased internal resistance translates to lower power transfer. What's interesting is that as the motor resistance is reduced you have to be careful of things going bad on the "other" side of the curve. (the motor needs to match the battery) I don't think I've ever personally experienced this because my motors have had very high resistance relatively speaking, but the formulas predict that there is another side to all this.

[safe]

More About Cores

Cores really increase the yield of the magnets. It's nice to want to use the "Standard Halbach" design without cores because it does make the math easier and it guarantees results, but it also doesn't seem to work very well with the disc concept I'm working on. The problem is that the "Standard Halbach" works better with coils going around the entire circumference of the motor and that's not good for a stator design that is supposed to behave like a brake caliper. The brake caliper concept requires that the power is delivered in a very concentrated area... ideally 25% of the circumference.

Simulations show that if you use the Silicon Steel (Electrical Steel) that the losses are minimal. If you try to use regular carbon based steel you will burn the stator up in a hurry because the heat can be massive. (I'm seeing 800 watts of wasted heat energy in the simulations with regular steel)

So the "bottom line" is that I need to find some small little steel pieces to fit in between the coils in the stator. What I'm thinking of doing is cannibalizing one of my broken Unite motors and use the Silicon Steel it has. I've worked with those plates before and they are extremely hard to grind, so this might be an exercise in frustration, but it's worth a try.

Take a peak at the difference between the cores and not having cores: (this was discussed before)

Cogging 001.gif (25.3 KB) Viewed 630 times

Cogging 002.gif (3.57 KB) Viewed 627 times

...also, now I see why they create the "I-beam" shape in the steel. It's done like that to reduce cogging. If you try to use a straight piece of steel you will get cogging that is so bad that the motor can barely run.

[safe]

Feeling Optimistic

I took apart an old Unite motor and separated out the flat Silicon Steel plates. With a sharp cutting tool (shown) it was easy to cut the plates up and most of the shape is already what I wanted.

The concept is looking more and more viable:

Silicon Steel.jpg (266.09 KB) Viewed 933 times

Quick Update:

Silicon Steel 2.jpg (82.34 KB) Viewed 619 times

...this is enough for my purposes. (took about an hour)

[safe]

Push / Pull

Push Pull.jpg (67.53 KB) Viewed 492 times

With Three Phase power the "inner" part of a coil produces a magnetic force, but the "outer" side is not pared with a complimentary (in phase) coil, so it's external field is not really used effectively.

With Single Phase power each alternation of the coils from positive to negative produces an equal north and south directed magnetic field. (assume green is positive and blank negative)

What this means is that you get a doubling of the magnetic strength for the same compaction of coils with Single Phase. This can be done with Three Phase if you layer your coils so that they overlap each other and have a 120 degree shift, but the increased interference with the other phases reduces the effect somewhat. The bottom line with Single Phase is that you can throw in an extra Silicon Steel core in between and that doubles the effective magnetic field.

So the logical sequence goes:

Start with 0.4 Tesla magnets.

Combine them so that they magnify themselves to increase the magnetic strength up to 0.8 Teslas.

Add a Silicon Steel core that magnifies the magnetic field up to about 1.1 Teslas.

Position the coils into a Single Phase configuration and get the advantage of Push / Pull forces and the "effective" yield of the magnets go up to (double) about 2.2 Teslas.

...it's starting to look better.

(but before I celebrate too much I need to do another cogging analysis to be sure I haven't cogged this motor to death)

[safe]

What is the True Force?

Cogging is like noise... it confuses the way you look at the motor force and makes it hard to know what you are really getting. One of the great things about the "Standard Halbach" design is that there is no cogging so you can just look at the forces produced and take the RMS (root mean squared) value as the true force. With cogging you literally need to "filter out" the effects of cogging to find what real power you get. It's easy to assume things that aren't true until you see this. (I know I did)

RMS Force.jpg (96.91 KB) Viewed 494 times

...you can see that the blue is the true force that is created after you filter out the effects of cogging. Compared to having no cores the RMS Force is overall higher by adding the cores, but not by as much as I had at first thought.

Cogging is noise...

[safe]

I'm going to load these files first, then I'll talk about them...

[safe]

More Files...

[safe]

Halbach Disc Motor Project (Complete Package)

Starting with the zip file:

download/file.php?id=574

...you begin by loading the FEMM files into the simulation software and look at the design.

Next you open the spreadsheet (3.0 Works format) and notice how the FEMM files match the spreadsheet. There is an interactive process of running simulations in FEMM and then placing those results into the spreadsheet.

The 1K Racing power chart looks like:

The 2K Racing power chart looks like:

The CC Racing power chart looks like:

A comparison of these three controller options is made and the power required aerodynamically is also presented so that top speed can be estimated: (for flat land)

In order to estimate the "True Force" that the motor produces you need to filter out the effects of cogging and so you go through this process to determine the force using a RMS (root mean square) calculation:


...though I can't guarantee that this is going to work (haven't built it yet) it's what I'm working on.

[safe]

Freewheeling Issues

Ideally I'd like the Halbach Disc to freewheel.

I have a Shimano Deore cassette hub that I could build the Halbach Disc off of, but the problem would be in adding the freewheel for the chain. There needs to be some sort of adaptor that could either replace the standard Shimano Deore cassette (the part of which that contains the cluster) OR the other idea would be to have an adaptor that threaded into the end cap of the cassette and then provided another set of threads to connect the freewheel.

The reason for this (in the long run) is to create something that can pass the 20 mph Federal Ebike Law. In that law the motor cannot add speed beyond 20 mph. The law does not restrict freewheeling downhill, but with a motor on your wheel (hub or otherwise) you are going to get cogging and backemf when you are going faster than the motor is allowed to operate. Many people fiddle with regenerative braking which is fine for people determined to go slow and save battery power, but is against the spirit of a racer. There should be no reason that as a downhill racer going down a steep hill that a Federal Law Legal ebike could not reach 50+ mph because ordinary Gravity Bikes do that. (they do that in the Tour De France on skinny tires)

So in the long run I need a Double Freewheel:

One freewheel for the Motor

Another freewheel for the Pedals

...I can envision the adaptor possibilities, but I doubt that I could create one very easily without some more sophisticated shop tools.

[safe]

Here are a few more things that I've added to the spreadsheet.

Heat.jpg (125.15 KB) Viewed 496 times

Efficiency.jpg (135.84 KB) Viewed 487 times

Hill Climb.jpg (109.08 KB) Viewed 485 times

...I'll add the revised spreadsheet soon.

[safe]

This is a FEMM image from the actual project as it is now.

FEMM.jpg (226.16 KB) Viewed 479 times

[safe]

Next Task...

I'm pretty sure the math on the motor is done. At this point my next challenge is to figure out the controller. I've decided (at present) to go Single Phase with a sensor and use this chip:

ZXBM1015 Circuit.jpg (58.96 KB) Viewed 464 times

...which means I need to learn about optimal MOSFET configurations. Time for another spreadsheet. As it turns out Fairchild Semiconductor has a spreadsheet that deals with some of this and I'll include that into my research. (might even incorporate the MOSFETS into the Halbach spreadsheet in the end to make a single grand unified calculator)

I want to post the present (solid) Halbach spreadsheet at the top of the next page as "Version 1.0", so when I get there I'll do that.

[safe]

Some Thoughts...

I was browsing through some CSIRO documentation and noticed a couple of interesting points:

"Tritium" - Apparently the controller they recommend for the CSIRO motor comes from a company that named itself Tritium. Since Halbach technology came out of Lawrence Livermore Lab and was initially involved with particle beam research (and was probably looked at by others doing nuclear weapons work) it makes sense that one way or the other the "super cool" concept of nuclear weapons (using Tritium to increase weapon efficiency) would find it's way over to the most "super cool" and efficient electric motor. It's always fun to discover these interesting connections between things.

"Weight" - The CSIRO weighs 5.8 kg which is 13 lbs for those of us still still addicted to American measurements. 13 lbs is not all that light weight. The Halbach Disc Motor that I have in mind should only be a few pounds so I'm doing pretty well on paper so far.