The Turquoise Energy MPMG:
a "Modular" Permanent Magnet Generator

A Low RPM, High Power Density, Axial Flux Electrical Generator (or synchronous motor) that is easily made at home from readily available parts.
by Craig Carmichael,
Victoria BC Canada
December 3 2007

1. Intro to the Unit & Possibilities
2. Make or Buy One?
3. Topic by Topic: Makeup and Making of the MPMG
4. Parts List
5. My Test Results & Some Tentative Specs
6. Original Drawings
7. More Experiments To Try

1. Intro

This generator is based around a car axle hub and two ten inch disk brake rotors, with a twelve inch diameter stator which is cast from polyester resin poured into a melamine "shell". Thus the diameter is just over twelve inches, and the length is about seven inches with one stator (with possibilities for making it shorter yet). Additional rotor-stator pairs ("modules") can theoretically be added, each pair adding only 1-3/4 inches to the length while adding the strength of another whole generator. The machine should be capable of handling perhaps two to three kilowatts per stator, which it should achieve at a few hundred RPM. A machine with three stators each producing 2.5KW is 7.5KW or 10 horsepower, if that's not pushing the limits of a car wheel hub and some threaded rods too far.

It is configured with 12 magnets per rotor and 9 coils (per stator), to give three phase power. A key difference between this unit and axial flux windplant units is the incorporation of iron in the stator to increase magnetic flux and power density. The surprise magic iron ingredient: nail gun nail strips!

The prototype unit produces 12 VAC at about 400 RPM (and 40 Hz). About 45-60 amps per phase is expected. (12V * 45A * 3 = 1620W.) Other figures could be used, eg, 24VAC at 800 RPM. That would be over 6000 watts, but the unit would probably overheat (and hence smell bad) if pushed beyond 2000 or 3000 watts for considerable periods of time.

Extensive testing hasn't been carried out: I have no budget or time or equipment for doing so but wish to share the possibilities now regardless of the power specs being somewhat tentative.

What is it good for? Probably not the best for wind power owing to cogging, but a good generator for small hydro power, wave or tidal power, as a synchronous motor, or with an appropriate power inverter as the drive for an electric car (a two stator unit on each wheel would give something like 32 HP and all wheel drive.)

My own application is an ocean wave power unit, to light up a bank or two of bright lights for a demo... later perhaps for off-grid power at the seashore or on a boat.

And why is it called the "Turquoise Energy" MPMG? Well... here's the cover shell!

- Further Possibilities -

The possibilities go well beyond a fairly small specialty generator. If it's simple enough to make at home, what could an electric motor company do with this axial flux design? It could also be applied in general to induction motors and generators. The essence of the design could be scaled up for machines of tens and hundreds of kilowatts, again perhaps using multiple stator-rotor pairs to attain a desired power at lower RPM and from a smaller diameter machine.

Cost savings could be major. Not only are the die cut laminations eliminated and replaced by simple wire, but the hand insertion of windings into the slots is eliminated. The coils can be wound by machine with the iron included, and dropped into the stator can. Varnishing and oven baking the motor is replaced by forming the stator cans and casting the stator(s) with polyester resin. The number of models required is reduced by the ability to add stator/rotor modules to a more generic shell, and the motor assembly is separate from the stator, so the stator for the customer's desired voltage/current/RPM (with choice of x1 or x3 coils per phase voltages and even more customer voltage/current options with multiple stators) can be installed or configured just prior to shipping, or by the customer. To repair a burned out motor with no other mechanical damage, simple stator replacement is required.

And if this design - with the stators having less iron overall but more effectively utilized than today's radial flux designs - should prove a bit more efficient, there are also valuable energy savings to be had.

I don't plan to do "up-scale" machines myself... unless it's the 250-350 KW machines for the production model ocean wave powered generators, and we find we can save twenty or thirty thousand dollars per unit for dozens or even hundreds of units. Aye, that might be worth doing!

2. Make or Buy One?

One idea behind this generator design is that they should be simple enough to make at home, but it does take a certain amount of time, aptitude and ambition to do so... and with experience, the second attempt is bound to be more refined than the first. Hopefully some of my experience detailed below will be of assistance in achieving an excellent result on the first try. I will of course be glad to hear of construction details and results obtained by others, and will be sure to add any relevant measurements and specs available to any future version of this writeup, with any specially interesting pics you send and of course credit to the author.

It is unlikely I'll spend more of my time making more generators as I now have what I need and I have many other things to do. But... if anyone would prefer to order one of these home-made generators ready made from me, wound for any voltage and RPM, it would give me a chance to try out variations on the theme. (The voltage and RPM are very approximately defined at this point - the RPM may need to be adjusted somewhat to obtain the desired voltage.) Price would be 1000 $Canadian for a single stator machine, with delivery within two months. (Probably sooner but I don't like to promise too much.) A double power machine with two stators (and three rotors): $1500. Just think: you'd get your ready made generator and you'd be funding product research and development at the same time. The design parameters would become better defined and some of the minor variations would be tried and tested. I would of course write up the results on this web site and give credit to the purchaser for the funding. Note that of course there is no UL or CSA approval of the units and that what you buy should be considered a prototype. I guarantee that it will work when it arrives, but there will be no other warranty, eg as to detailed specs, and I will accept no liability for anything that happens to it, with it, or because of it.

3. Topic by Topic:
Makeup and Making of the MPMG

Overall strategy
First of course one must make decisions about what is to be made: The number of stators, unit length, casing, wire and windings, drive gear or pulley, and so on. The only components that need to change to add more interior rotor-stator pairs are the threaded rods and the body, which need to be about 1-3/4 inches longer per added "module". (I cut the rods longer than required for my one stator, as I had originally intended to have two. I still have the option.)

Then of course, one must get the parts to make it. In particular, the hub, rotors and magnets need to be acquired. Best place for the hub is probably an autowreckers (though I don't know the price for a brand new one), and you might pick up the 10 inch disk brake cylinders while you're there, though they're not pricey to buy new, and you'd get smooth, shiny ones! For the magnets, unless you can find them somewhere in your own town, you'll need to find some on line and order them. In some locales, evidently magnet wire is also hard to come by and has to be ordered on line.

Once you've got that far, you need to build the thing. Here is a guide by subassembly component:

(A) end rotors
(B) the 9 identical stator coils (per stator)
(C) the melamine stator shell (seal it everywhere below the top!)
(D) cast the stator by pouring in polyester resin and clamping the melamine top piece down on it
(E) the end plates (rather than "end bells") and casing
(F) center rotor(s) (for multiple stators)
(G) assembling it all
(H) testing it.
(I) Send me your pics and results! Post about them on the alt-power discussion list!

(A) End Rotors

A nice feature of this sort of machine is there's no electrical parts on the spinning rotors, hence no brushes and commutators to connect them. The spinning magnetic field is produced by powerful permanent magnets, and the stator with the coils doesn't move. The magnets are screwed and epoxy glued to the rotors.

One thing I wanted my rotors to do was to blow air centrifugally for cooling, and since it's a low RPM machine, the more air vanes there are the better. I chose rotors with air vanes, and more importantly, the magnets themselves act as air vanes right next to the stator(s).

Make a template for drilling the magnet holes. I used 9/64" holes for #8 bolts (see #3 below.) The outer ends of the magnets should be at 5" from the center (rotor outer 10"; diameter) and the inner ends at 3" (6" inner diameter). My aluminum template had magnet slots for magnets to be glued-on to 12" rotors, but I ignored these except as alignment guides, and marked it and drilled holes for the 10" rotors with screw-on magnets.

Drill the holes in both rotors. If one of the lug/alignment holes lines up with a magnet (as the closest one does here), it won't matter which way up the template is for the two rotors. Otherwise, you have to invert the template for the top rotor.

Thread the holes. I drilled and threaded #8 - 32. The magnets I got will take #10 flat head machine screws, but #8 gave me some valuable slack for sloppy tolerances if (when) the two holes weren't EXACTLY the right distance apart.

If you're just drilling plain holes through the rotors and putting nuts (use nylock) on the end of the bolts, drill the holes a little bigger to give a bit of tolerance and use the #10's by all means.

Put on the magnets. Easier said than done! First a warning: be VERY CAREFUL with these incredibly powerful magnets. If you lose control, you won't have time to react. The accident will be virtually instantaneous. Having a wrench or screwdriver fly out of your hand working near the rotor is bad enough, but if the loose magnets hit each other, they'll shatter. If there's a part of you like a finger in the way, it can be badly cut, broken or even crushed.

Likewise, keep the two loose rotors way way far away from each other so an accidental or thoughtless move won't bring them within a distance (2 feet?) where they'll fly together. Even if they don't maim or kill you, you may never get them apart. It's quite difficult just to pry any two of the magnets apart.

There are two stacks of 12 magnets in the package. One stack has north on the tops, the other has north on the bottoms. On mine, the polarity was indicated by white dots. (Note the white dot between the screws on every second magnet in the pic above.)

The overall strategy is to glue and screw the 6 north magnets onto every second position, then put 6 south magnets between them. I sweated really hard doing the second six on the first rotor, even with some bits of wood between magnets. The magnets attracted and were closely spaced, and things could go quite wrong with the slightest misstep or nervous twitch.

On the second rotor I wised up and made the little wooden jig shown. The bottom piece is a wedge that goes between two mounted magnets, with the rectangular hole being big enough to maneuver the new magnet a bit. The side boards ensure that the magnet can't flip over and glom onto the magnet on either side. I also chiselled out the ramp to get the magnet really close to the right place before it grabbed on. (I didn't think to take any pics during the assembly process.)

Get out one strip of magnets, and put it well away from anything metal. (I put them on the floor a ways off.) Put the other set out of harm's way, eg, back in the box.

For each magnet, get out the "epoxy steel" epoxy glue, mix the two creams, and spread some on the rotor where the magnet is to go. Using a small philips screwdriver that fits into the mounting hole in the magnet, pry the magnet off the stack and get a piece of wood in between it and the stack. Once the magnets are far enough apart, the one magnet can be separated and taken over to the rotor. Using the piece of wood, get the magnet up to the rotor and lower the far end to the rotor, then remove the wood and lower the near end. Put a thin shaft into one hole and wiggle the magnet around until the magnet hole and rotor hole are in line. Then put the screw (stainless steel, flat head, #8-32, 5/8" long, machine screw) into the hole and screw it in a bit. Line up the other one and get it started too, then tighten them both down.

Repeat for all the six magnets. Then put away that stack (6 magnets remaining) and get out the other stack of 12. Glue up one position, then with the jig as shown get the magnet into place and screw it down. Do all six, and the rotor is done. Put the rotor in a box with some styrofoam to keep the magnets well inside, being careful not to pass by any metal too closely on the way by. Put the box somewhere safe.

Now there's just the second rotor to do.

(B) Stator Coils

The desired RPM and the desired output voltage determine the number of winds in the stator coils. The number of winds determines how thick a cross section of wire will fit in the available space, and that in turn determines the output current capacity.

Note that, for example: double the winds will double the voltage, but the wire will have to be half the cross sectional size to fit in, and hence will have half the current capacity. Watts = Voltage * Current, so in principle at any given RPM the watts is the same regardless of how the coil is wired.

With double the RPM, double the voltage and double the current are both obtained, so the power is multiplied by four. (In keeping with "squared" in the law e=mv^2) The reason for building a large size generator such as this one is to obtain a desired power at a low RPM. One can attain a kilowatt or so with a hefty car alternator, but at 3000 or 4000 RPM, about ten times our speed for similar power.

Note that the AC voltage can be magnified 1.4 * by rectifying the AC output of all three phases into DC with 6 high current diodes. Also note with 12 magnet rotors the frequency of the AC: Hz = RPM / 10, thus 400 RPM produces 40 Hz power.

The configuration of 12 magnet rotors and 9 coil stators provides three phase power. Suddenly as I write this I ask myself why? Frankly, I just copied the windplant designs. They wanted to charge batteries and needed DC. Best way to get fairly steady DC is to rectify 3 (or more) phase AC. As I want to light light bulbs directly and won't be rectifying, three sets of leads is just an extra complication. But the die (not to mention the stator) is cast: For single phase power, one could wind 12 coils, but there really isn't room. 6 coils would waste space in the stator: you'd only get the power of two of the three phases of the 9 coil version. Seems the shapes just work out for 12 magnets and nine coils. But three phase is also valuable if the unit is to be used as a motor.

For this unit configuration, one may find the following to be useful rough guides:

Available Power per phase (for each stator - rough estimates. Multiply by 3 for total stator power.) At higher speeds this is limited by internal heat regardless of RPM. It can be driven harder but at risk of burning out the stator. At low RPMs it can't get hot enough to burn out.

100 RPM -  37 W
150 RPM -  85 W
200 RPM - 150 W
300 RPM - 350 W
400 RPM - 600 W
500 RPM - 800? W
600 RPM - 800? W
800 RPM - 800? W

Estimated Voltage (per phase, estimated based on my limited test results). Really, you can see how to derive your own figures. The cross sectional area of wire is halved for each increase by 3 gauge numbers. Thus, #15 gauge is 1/2 the cross sectional area of #12, and #18 is half of #15. (If that seems backwards to you, you're not alone.)

30 turns per coil (3 coils per phase, in series, so 90 turns total, ~#11 or #12 wire)

12V at 400 RPM
13.8V at 420 RPM (the nominal voltage of a car that's running.)
24V at 800 RPM

45 turns    (#13 or #14 wire)

12V at 270 RPM
24V at 530 RPM
12V at 810 RPM    (coils in parallel instead of in series)

60 Turns          (#14 or #15 wire)

12V at 200 RPM
24V at 400 RPM
48V at 800 RPM
12V at 600 RPM    (coils in parallel)

90 Turns         (#15 wire)

12V at 130 RPM
24V at 270 RPM
48V at 530 RPM
12V at 400 RPM    (parallel)
24V at 800 RPM

120 Turns         (#17 wire?)

12V at 100 RPM
24V at 200
48V at 400
120V  1000 RPM

160 Turns       (#18 wire?)     120V at 750 RPM/75Hz

200 Turns       (#19?)                  120V at 600 RPM/60Hz

A typical coil, 91 winds of #15 wire - 7 layers with 13 winds per layer.

Coil Details: The leads exit out near the rim through the top cover of the stator, which must have holes or slots cut for them, just outside the spinning magnets on the rotor. Note that there are fewer nail strips towards the inside, to make it a little thinner there, as that's where there is the least room. Since the inner and outer ends are parallel to the rotor motion, there's no use putting iron there - voltage is induced at right angles to the magnet motion. This is well as there's also the least room in the inner end before the stator would hit the five spinning rods of the rotors. Half the nail strips have their heads towards the top and are quite visible. The other half have their points facing up and are hard to see. There are thus 16 strips in the core, 8 on each side of the plastic center. There is a strip, or half a strip, between each layer of wire. While there is much less iron overall, this would seem to be actually more working iron than can be placed in a common radial flux machine, as they have no practical way to put iron laminates between winding layers.

As shown the amount of copper the magnetic field passes by determines the power output. Trying to squeeze as much copper as possible into the smallest space possible is an interesting exercise, which gets more interesting with nail strips wound into the coils. The cross sectional area of a round wire is (pi*D^2)/4. If we consider "D" (diameter) to be 1, that's 3.14159/4=.785 . Fit that into a 1x1 square and there's 21.4% empty space. But if one winds a neat coil, the bottoms of the second row wires can fit into the spaces between wires of the first row except at one point where they need to cross over, so there is actually less theoretical wasted space. That's a conventional copper wire coil.

But if there is a layer of nails between each two rows, we're back to the 21.4% figure, not to mention the space that the nails take up. The space of the nails isn't wasted though, because it increases the magnetic field going by each wire. Just where the tradeoff is optimum could be the subject of much experimentation. But copper is much more expensive than iron, so a saving of copper without a reduction of performance is good economy. I would note, though, that even if there's iron only between every second layer of windings instead of every layer as I wound it, each layer of the coil would still have iron right next to it on one side or the other. Thus I could possibly have made my coils somewhat more compact without losing much (or perhaps any) performance.

One could take square or rectangular wire and theoretically have no waste space, but square wire is actually quite difficult to keep from twisting. Rectangular wire can work, and flat wire is better even with rounded sides. This is the shape of the heavy surplus "ribbon wire" I used.

Wishing to experiment with optimizing performance to the nth degree, I flattened some round wire with a jewellers' rolling mill. However, I found that with too little flattening it will twist too easily and is almost impossible to put on straight. But too much flattening will "work harden" the wire, and work hardened copper actually has a slightly higher electrical resistance than soft copper, so you lose by reduced conductance what you gained in occupied cross section. It may also risk damaging the insulation, and will slightly stretch and decrease the gauge of the wire. So in the end, if you can't buy factory made rectangular or ribbon wire for around the same price, you might as well just use round and wrap it as neatly as you can. You'll get about 70% actual copper filling of the available space (excluding the iron space), where you might have got 75%+. Some say even that's going overboard and that there's no point keeping it tidy at all - just wrap it up quickly and accept 60%-65% fillage. However, those espousing that philosophy - which isn't without merit - aren't putting iron strips into their coils.

-- Winding the Coils --

The coils and hence the stator are one inch thick as I used strips of 1" long nail gun nails in the coils. The 9 coils I used are each 30 winds of "ribbon" wire that a motor shop had surplus, about "#11.5" gauge. (It's so old the insulation flakes off at the slightest excuse - probably sat there for decades!)

Coil Winding Bits: White side guides, 1" acrylic plastic center spacer, center spacer with nails, something that used to be a ruler to show size.

Coil Winding Jig. I used an existing tool (I made it for reaming recorder bores) with the v-belt pulley as a very nice crank handle. (The pic shows winding of a small coil for something else... seems I took no pics of winding the PMG coils.)

(C) the melamine stator shell

The stator assembly with some miscellaneous coils placed in it (not all shown are the flat wire ones I finally used)

This is a great labour saving technique compared to other methods of stator construction. Here you make a decorative melamine can and fill it with resin to hold the coils solidly in place. Without cooling holes, as is common for windplant generators, it's extremely simple: glue the outer and inner side strips to the bottom. For higher power density units with iron I expect cooling air channels are valuable (without having actually got as far as trying out continuous higher power operation), and they add a certain a amount of complication to making and casting the stator.

As long as the can doesn't leak, the mold - really just a jig - doesn't have to be solid. The outer ring of about 1/2" to 3/4" plywood and the inner plywood circle hold the outer and inner edge strips in place for glueing and while casting. (with a piece of wire or rubber bands around the inner strip at least while glueing.) Any piece of plywood or other flat surface under the can, and a piece over the top with holes or slots to accommodate the coil lead wires, with weights on it, should suffice to hold everything while the resin hardens. Put a piece of polyethylene over the bottom piece and under the top piece to catch any drips of resin.

If the can leaks you'll be sorry. I used much extra resin and spent some aggravating time chipping away resin from the outside after I cast mine, and it isn't quite as pretty as it should have been. It would have been well stuck to the mold without the polyethylene sheets. So I recommend glueing the seams of the can carefully. Luckily the resin didn't really bond to the outer face of the melamine. (It didn't help that my resin took WAY longer to harden than the chart indicated for the temperature. It had hours to leak out, and it was still tacky days later. If I measured the mix wrong, I did it consistently 3 or 4 times.)

Owing to the stator straws, I put an extra strip of melamine around the outside edge, facing inwards. The straws went through holes in the inner piece and butted against the outer.

But the drinking straws for cooling air channels proved to be more tricky than I thought. They necessitated dropping one layer of windings from each coil to fit in, and with the leaks in the can, most of them filled with resin when I cast the stator. Solid polyethylene rods, smooth square strips, or perhaps telephone wire cable with the wires still inside, would have been better and easy to remove. Perhaps tapered cylinders could be of use: stuffed in to fit, they could ensure no leaks around the holes and should pull out easily after casting. They would even be thinner near the inside where there's least room and wider at the outside!

Or maybe make vent holes from top to bottom instead of from outside rim to middle.

An obvious place to put some cooling holes would be top to bottom through the centers of the coils. When I'd finished casting, I wasn't very certain exactly where the centers were. My suggestion is to mark and drill the holes for the nine coil positions in the top cover, and make certain the coils line up with them before casting. Then you can drill where the holes in the cover are.

Better perhaps would be to also remove the coil formers and stuff in cylinders of something that won't bond to the resin (polyethylene rods?) through the top holes. Before casting, ensure they line up with the centers of the coils to keep the resin out. The polyethylene top sheet and the top plywood piece will need holes to accommodate these plugs. After casting, pull them out, then drilling is just through the bottom melamine. On the other hand, if you're only doing one, perhaps it's easier just to drill through the resin afterwards!

Another good place for similar vent holes is between the coils near the outer rim, though three of these nine positions are needed for the mounting rods. I did in fact drill holes in those spots, which intersect the holes from center to rim.

One should be careful that a good honeycomb of holes doesn't seriously compromise the structural strength. The two round ("drinking straw") holes from rim to hub could have been one rectangular hole right from top to bottom for best cooling, but that would have made the stator quite susceptible to breaking into nine pie shaped pieces, especially where the mounting rods stress it.

Another view of the stator assembly

(D) cast the stator by pouring in polyester resin and putting the melamine top piece on it

Probably not much more needs to be said about this except do a "dry run" of putting everything together and putting the cover and weights on first, and to follow the resin instructions. Mix the resin. Perhaps paint some on the inside of the top cover to ensure it gets a good coating. Put the top cover on, trying to leave no big bubbles. Put on top of that the polyethylene sheet, and on that the plywood cover piece with slots for the wires, then some weights on top to be sure the melamine is flattened out.

(E) the end plate(s) (rather than "end bells") and the casing

These may seem like simple components... and they are, but they do need to be made, or at least one end plate does. It is an advantage that the generator assembles and mounts easily onto a flat plate.

Consider making at least the lower plate square or rectangular rather than round for easily attaching the machine to a flat surface. And if there are drive gears or other axles, perhaps a turbine, to be mounted nearby, why not mount them all on one piece of steel?

I used a round 12" diameter piece of 1/4" steel plate that I already had, but I had to make an "A" frame out of angle iron to hold the unit steady for testing. I may swap for a big rectangular plate when I start assembling shafts and gears. 3/16" or even 1/8" might be sufficient thickness. If it bends in use, though, the gears will skip and it'll be disappointing.

The generator itself needs just 8 holes in the plate: a large round hole in the middle for the GM Hub, four bolt holes nearby for the hub bolts (get the layout and size for these from the hub), and three 1/2 inch holes at about radius 5.5 inches from the center, 120 degrees apart, for the stator holding rods.

While the hub is still loose, pound out the five wheel lug bolts - they're a friction fit. IIRC the threaded rods don't quite fit through the holes, so the holes have to be drilled out a bit bigger, to 1/2".

The basic point to having an outer casing and a top end plate is to keep things out of the generator, including dust, grit, magnetic things, and fingers. Safety first! If the 3 stator rods are longer than anything else, things should be simple. Cut the casing from a piece of 12" turquoise plastic culvert pipe (I'm not sure it'll work if it's not turquoise!), shorter than the distance from the bottom plate to the ends of the rods by enough to put on the top end plate and nuts to hold it.

Somewhere near the hub there'll need to be a big slot or hole cut through the side of the casing for the drive belt, chain or gear. However you work it, make it safe. A hand dragged through a v-belt pulley or a bicycle chain gear is really ugly!

The top cover can be a flat piece of just about anything that can be painted turquoise - metal, plastic, plywood...

(F) Center Rotor(s)

For running more than one stator, we want one spinning magnet width between each two stators. The iron in each stator acts as the iron backing to the magnet for its neighbor stator, so the rotor needs only magnets.

Since I didn't build the second stator, I haven't made one of these, and you're charting new territory. The basic plan is:

(a) cut two 10" melamine circles and a 1/2" x 31.4159" (ten inches times pi) strip for the skin of the rotor, which will be cast in polyester resin. I used the thinnest melamine I could get for the stator, but I suggest quite heavy, sturdy stuff for the rotor - having a magnet break loose would be bad, very bad if it did it while the generator was running.

It may be best to cut center holes and the five stud bolt holes in the melamine in advance of casting as well. In that case, you'll want another half inch wide strip around the center hole and some 1/2" diameter round cylinder pieces 1/2 inch long to fill the bolt holes (made of something that doesn't stick to resin. Perhaps plastic hose or polyethylene rod?). Or you could drill the holes later in the finished casting. The only point to having a center hole(s) is for ventilation air passage. There are specially shaped drill bits for drilling holes in plastic that reduce the chance regular twist drills have of breaking it.

(b) You'll need to make a ten inch magnet holder similar to the original twelve inch one above that I turned into a hole drilling guide.

(c) Glue the melamine bottom to the inside and outside strip to form a "can" to cast the resin in. (You may or may not need to cut a 10 inch round hole in a piece of plywood, and another one to ring the inner strip, to hold things in place while the glue sets.) Did I mention that if the can leaks anywhere, you'll have a mess?

(d) Put the magnet holder in the can. Put in some non-magnetic washers or something first to hold it up off the bottom and away from the glue. Use some more of that "steel epox" glue to glue the six "north up" magnets into place in every second slot. Be sure the position of the slots matches the other rotors. You can of course flip this rotor over if the magnet polarities are wrong, but if nothing lines up you'll have to live with skewed rotors. I suggest on all rotors lining up one magnet with one of the five stud holes.

(e) Use some kind of safety jig so the magnets can't get at each other in case of a slip, and glue in the six "south up" magnets between the others.

(f) Be sure the magnets are well glued and carefully remove the magnet holder and the spacer bits.

(g) Put a sheet of polyethylene on a flat level surface (one you don't care about a mess on) and set the can on it. Whenever you move it, REMEMBER THE MAGNETS!!! Don't pass across or set it near metals or - especially - the other rotors. When you're not actually working with the magnets per se is an easy time to forget about them and have an accident. Mix and pour in the resin. Perhaps paint some on the inside of the top cover to ensure it gets a good coating. Put the top cover on, trying to leave no bubbles. Put on top a polyethylene sheet, and on that a flat cover piece with heavy weights on top.

(h) Once it's made, put it in a box with styrofoam or whatever both above and below it, and put it away until you're ready to install it.

(G) Generator Assembly

This is the "mechano set" part.

The best part is that the hub and axle unit is already done for us by GM! A five bolt hub that cames out of the back wheel of a front wheel drive GM car at the auto wrecker is the axle with bearings. (Evidently there's more than one model that will work. Was it a "Firebird"?)

Bolt the hub on the end plate. (I know, you already did.) Lay it down and we'll call the hub end the "bottom" of the assembly.

Screw a nut to around 1-1/2 inches from the end of each of the five 1/2" threaded rods. Put the short end through the bottom rotor holes from the magnet side.

If there is a drive gear or v-belt pulley (that you've drilled five matching holes in or otherwise fit), put it on behind the rotor. It should go at the hub end since there's no proper shaft and bearings going to the other end of the rotors to take the sideways stresses. Since the only thing to attach it to is the five bolts, five holes may have to be drilled in a solid body pulley. The bicycle pedal gear I used happened to have a five point star pattern and with a little grinding and some washers for spacing, it fit right on. Its form kept it centered on the hub. At the moment, I've taken it off in favour of a v-belt pulley.

Fit the rotor (& pulley or gear) on the hub. There should be at least the thickness of the nuts sticking out. Screw on the nuts and tighten them very solidly with two wrenches. Nothing else on the rod should be tightened quite as much as these, though if they start crushing the pulley you'll have to stop tightening. (Mind the magnets with the upper wrench.) If these are the tightest, the rods won't turn when you're trying to tighten other nuts on them.

This is a good time to slip in the desired v-belt or drive chain, as the stator and its mountings will block access later. A link belt may be a practical alternative to a regular v-belt, as it can be unlinked for removal or installation, and anyway is said to have less friction. (Having neglected to put on a v-belt before the stator, I borrowed a link belt from my bandsaw in order to run the tests.)

Screw 3 stainless steel (or brass) nuts a little ways onto the 3 stainless steel 1/2" threaded rods for the stator. (Stainless steel is virtually nonmagnetic. Brass is nonmagnetic. The magnets spin nearby these rods.) Put them through the end plate holes and screw 3 more nuts onto the ends. As with the rotor rods, tighten them very solidly with the two wrenches. Screw 3 more s/s nuts onto these rods down to about 3 inches above the magnets on the rotor, then slip on 3 s/s washers. Line up the stator mounting holes and slide it down the threaded rods to the nuts. DO NOT PUT THE STATOR ON WITHOUT THE NUTS AND WASHERS PLACED AS STATED TO HOLD THE STATOR AWAY FROM THE ROTOR! When it gets close enough to the rotor it will suddenly clamp down, even lifting the entire machine, and can crush your fingers! Turn the nuts to gradually and fairly evenly lower the stator down near the magnets of the rotor, leaving an even air gap of somewhere between .02 and .2 inches as desired. (You'll probably know better after you've done some testing what gap seems best. Distance reduces power at a given RPM but also reduces cogging.) You can pry up the edge of the stator a bit until it gets close so the nuts turn freely by hand. When you get close you'll need a wrench (and those annoying magnets will keep grabbing at it).

Put on 3 more s/s washers and screw on three more s/s nuts. Tighten them against the stator enough to hold it securely in place. Recheck the gap if it's a close fit.

Screw 5 more nuts on the rotor rods to about 3 inches above the stator. Line up the magnets and mounting holes and slip on the top rotor. Gradually lower it by turning the nuts one at a time. DO NOT PUT THE TOP ROTOR ON WITHOUT THE NUTS IN PLACE TO HOLD IT AWAY FROM THE STATOR! Again, without them it will suddenly clamp down and could crush your fingers or otherwise injure you! One must pry it up on one side to free the nut being turned as there's no way to get a wrench in there. Leave an even air gap similar to that of the lower rotor.

Make sure all five of the nuts are connecting against the upper rotor with your fingers. Screw on five more nuts and tighten them securely against the upper rotor. If the rod starts to turn you may need to use a second wrench on the very bottom nut, or even screw two more nuts on the top, tightly against each other, and use one wrench on them to hold the rod.

Check the gaps again. Make sure the rotors turn freely (other than cogging). Be sure you're not going to hurt your fingers in the threaded rods if there's a lot of cogging and it turns suddenly. Perhaps put two nuts on one of the five rotor rods, tighten them firmly against each other with two wrenches, and use a ratcheting socket wrench to turn the rotor.

(H) Testing It

Let's see... You need at least one meter that can measure AC frequency, voltage and current. With 12 magnet rotors the frequency is 1/10th of the RPM, thus giving the RPM. If the generator will put out over about 10 amps, you need a clamp-on amp meter. Luckily there are digital meters these days that measure all these things. I got mine at an electronics parts store for about $80. I used a separate multimeter for the voltage because I already had it, but it didn't have the clamp on ammeter and it didn't measure frequency. A third meter would have been even more convenient to get all 3 readings without having to switch meter functions while things were running.

And you need a motor to turn the generator. I found a 1/4 HP 1800 rpm motor wasn't strong enough to start it turning owing to the cogging. (Although I had used it successfully a year ago when there was just one rotor mounted.) Even my 2 HP radial arm saw had a hard time and I was unable to do proper load tests. That was with a 2" pulley on the 3600 RPM saw, driving the generator (no load) to 830 RPM. If there was such a thing as a 1" pulley I could have driven it 420 RPM and had an easier time. (Maybe I should get a new 1.5" pulley and run some more tests at ~620 RPM?)

(I) Send me your pics and results! Post about them on the alt-power discussion board!

Every finding and observation at this point adds a little value to the store of knowledge for the next person who decides to build one.

4. Parts List

Note that there may be some minor 'nuts and bolts' omissions here and there. And, the list only includes parts used in the finished assembly, not parts and pieces for jigs, special tools, etc.

-- Central Assembly --

The chief components of the machine: mounting plate ("end bell"), main hub and axle (car wheel hub/axle unit), lower magnet rotor (car disk brake rotor) with NIB supermagnets mounted, and the threaded "ready rods" that tie the pieces together. Beside the unit is the top rotor before mounting its magnets, and in the background is the uncast stator held together by its mold, with the loose top cover piece sitting on it.

01. "End bell" - a flat plate, probably 3/16" or 1/4" sheet steel.  Metal Shop.
   I used a 12" diameter piece of 1/4" steel. If there are a bunch of gears or other parts that are part of the unit, consider mounting them all on one big rigid plate. I may rework mine and do that.

02. Axle-Hub            5 stud car rear axle (GM)       Auto Wrecker.
   (Save the 4 nuts and bolts that attached it to the car.) This gives you the main spinning hub and "axle" of the generator with 2 heavy duty bearing races.

03. Drive Gear          Sprocket gear, gear, or v-belt pulley.  Bike shop?
   This depends on the application. For a wind turbine, nothing is required as the propeller mounts directly on the rotor studs and turns the unit.

04. 2 of: 10 inch rotors  5 stud Disk Brake Rotors  Wrecker or auto parts store. I used two with built in cooling fins. That should help with the cooling for heavy loads. Rotors without those should make for a somewhat more compact unit.

05. Five of: 1/2" "all thread"; AKA "ready rod"  Hardware, metal, fasteners shop.
   These take the place of an "axle" through the generator. They connect the second and any additional rotors to the first rotor, which is mounted on the hub.
   For a single stator, two rotor unit, five 7.1" (or a bit shorter) rods can be cut from a 36" piece as they are commonly sold. For each additional rotor-stator pair "module", another ~1-3/4" length is required, so I have five 9" rods to accommodate an originally intended second stator and third rotor that were to have provided more RPM and voltage options, eg, it could have cut the RPM for 12 volts from 400 to 200, allowed 600 RPM/12V (at 60 Hz), or 24V at 400 RPM.

06. 20 of: 1/2" nuts, 20 lock washers           Hardware, metal, fasteners shop
   For clamping the rotors onto the ready rods and to the turning wheel bearing plate.

07. 3 of 1/2" x 10" "all thread" AKA "ready rod" Hardware, metal, fasteners shop
   These ones hold the stator(s) to the "end bell". I used 3/8", but they seem pretty light duty given that the rotors want to magnetically drag the stator along with them.

08. 12 of 1/2" nuts, 6 lock washers, 6 flat washers.            Hardware Store
   For clamping the above to the 'end bell' and clamping the stator in place.

-- End Rotors Stuff --

11. set of 24 NIB magnets, 1/2" x 1" x 2", with mounting screw holes.   Web, eBay
   These super powerful magnets make everything work. 12 per rotor. Observe all cautions! These are probably the only parts that most people will have to order on line rather than buy locally.

12. "Epoxy Steel" epoxy glue to glue magnets onto rotors.       Hardware Store.
    This is only part of it. Use the screws, too, for sure!

13. 48 of #10 x 5/8" stainless steel flat head screws.          Hardware Store.
   These are to hold the magnets onto the rotors, along with the glue. Stainless steel is used because it's non-magnetic. You drill and tap holes into the rotors for these screws. I used #8 screws and then realized the magnets were sized for #10, but it gave me more room for error and I was glad I had. For rotors without fins, one could just drill holes and put bolts right through the rotor with "nyloc" nuts on the other end to hold them.

-- Stator Stuff --

21. 2' x 4' melamine AKA arborite sheet for skin. Choose any color and pattern except black. (I must point out that it's hard to see burn marks if everything is black already!) A solid light color might be well, though it's not what I picked. I used .025" thick melamine (very thin), but I should mention that I ended up backing the rotors off a ways and the gaps are around .15", so I could have used any thickness of melamine available in the store and it wouldn't have mattered.

22. About 6 pounds of magnet wire. (I forget. guessing.) The number of turns, and hence the wire gauge chosen depends on the desired voltage and RPM. I used 30 turns of heavy flat ribbon wire equivalent to about "11-1/2 gauge" that I got free. As it happened, this provides about 12 VAC (RMS) at 400 RPM (and 40 Hz). I first tried winding some with an extra layer, 36 turns, but things started looking pretty crowded inside the stator. Space is certainly a limiting factor - the number of turns you want determines the wire gauge that will fit, or vise versa! (Consider the thicker stator idea, below under "Additional Experiments")

23. About 3 pounds of nail gun nails. These come in boxes, within which are strips of nails much like strips of staples. Take an ohm meter to a strip. With some brands, the nails seem to be electrically well connected; with others, the connections are poor. We want POOR connections to avoid eddy currents in the iron - that's the whole point to using strips of "wires" instead of solid iron. All the ones I checked wouldn't magnetize when I rubbed a magnet along them. That's also the correct characteristic for motor laminate iron. But check yours and make sure. If they magnetize, you'll get heat and internal losses. Don't file or grind the nail strips, or the filings will electrically connect all the nails. (As I discovered when I tried to file off the heads and the points to get a flat, uniform strip - evidently it's impractical.)

24. Thin acetate sheets.        "Overhead transparency sheets"  Stationary Store
   I cut these into 1" wide strips and put them between the nail strips and the copper windings, fearing the wire insulation might gradually wear through and short to the nails, holding them in place for the winding with transparent tape.

THIS LIST HAS NOT BEEN CHECKED FOR COMPLETENESS. It could be missing whole areas, but I'm posting this essay now!

5. My Test Results & Some Tentative Specs

Well, here's a pic of my high-tech test setup.

First I should note that the cogging was so fierce I backed off the rotors to perhaps .15 inch air gaps to the stator. Then it only took two hands on the five bolts to get it to turn. Powerful as the specs show the MPMG to be, the power could have been increased with smaller gaps, at the cost of greater vibration. As it was, several items shook off the saw table and at least three light bulbs (a car headlight and two 120 V bulbs) were vibrated to death.

Carrying the monster out to the shop, I stopped at the bathroom scales, and with everything including the A-frame holding frame (except the turquoise cover), it weighed in at 57 pounds. I'd never given the weight much thought, but each part added its not insubstantial little bit. Luckily, it's not going on a spacecraft.

I wanted to turn the MPMG with a 1/4HP, 1800 RPM motor first and see what speeds, voltages and currents would be obtained. When I flipped the switch, it was obvious that the motor wasn't going to overcome the cogging and start the generator turning. That left me with pretty much one option: my 2HP, 3600RPM radial arm saw. Even this didn't quite want to start it. It seemed I would have to back the rotors off even farther, but I found that some cog positions were weaker than others, and starting at those it could be coaxed to start.

With the 9" pulley on the MPMG and nothing smaller than a 2" pulley for the saw, I was stuck with 830RPM (as measured - theoretically just under 800) as the generator speed, which is about twice the speed I actually want to run it. Owing to the saw's lack of power, though, I did get some slow speed results with heavy loads.

When I shorted a generator output, I couldn't get the thing to turn. With very heavy loads, consisting of aligator several clip leads and lengths of #22 hookup wire, the saw would start but wouldn't get up much torque or run up to speed.

The whole test, then, demonstrated the characteristics and limitations of the radial arm saw motor as much as the generator capacity.

The test results show the different voltages and currents I managed to obtain. With the 12 pole machine, RPM is obtained by measuring the frequency of the output (Don't you love modern digital meters?) and multiplying by 10. Unfortunately, I only had two meters and a poor saw motor that was starting to smell funny (or was that the hookup wire "loads"?), and for some resistances I didn't get the RPM. The 20A breaker to the garage blew twice while I was trying to keep things on long enough to get stable readings and take the RPM.

My coils were 30 turns of very heavy wire (total 90 turns - 3 coils per phase), so the voltage is low and the amperage high. I want to get 12 volts on each phase (actually 13.8 as per automotive standard) and, as the results show (26.5 V at 830 RPM), that'll require about 400-450 RPM.

- or -
Why is this 2HP radial arm so gutless trying to turn my generator?

Voltage Current Power   Speed
VAC RMS AmpsAC  Watts   RPM
2.55    11      28.1   120? (These were all pretty slow, but faster as I added
2.8     12      33.6   ?      wire and thus resistance. They show that the
3.5     12.8    44.8   ?      saw motor just didn't have the oomph to bring
5.3     16.4    87     ?      the gen up to speed with much of a load on it.)
6.7     15.2    102     ?
8.3     10.6    88     ?
20      19      380     710 (Finally the saw gradually spun up to - or at
21.5    17      366     720     least towards - synchronous speed.
22.7    16.5    375     740
23.2    14.1    327     750
26.0    3.3    86     ?  (light bulb loads. RPM probably ~810)
~26     2.7    70     ?  (820 RPM?)
26.4      .7    18.5   830
26.5    0        0     830

Interpretation of the test results: reading between the lines

Nowhere could the driving motor (theoretically 1500 watts) impart 400 watts or more to the generator output. I might have done better with loads I could switch on once the RPM was up, but the refinement of a switch I didn't have set up.

What is the unit capable of with sufficient drive? If one multiplies the 2.55 V of the first reading by 10 the current would be 110 amps and the power (on just one phase) 2800 Watts. It also appears that it would be over 1000 RPM, which is faster than I want to turn the unit. The actual tests showed 26.5 volts no load at 830 RPM.

I want to run car headlights for a demo and tests, which implies running at the automotive nominal standard of 13.8 volts, which in turn implies around 400-450 RPM (depending on load). I'm expecting to be able to put out at least 50 amps per phase (~= 720 Watts, totalling 2160 Watts). At about 3 amps per bulb, that would be, um, er, 50 car headlights!

Basically then, the generator has enough capacity to take whatever I care to extract from it, if the power is there to drive it. Calculations are telling me I'm only going to extract enough power from the waves with my little test/demo setup to light about ten, and that it'll be rough enough to make me seasick even to do that!

I also have some previous test results from before I put the magnets on the second rotor (October 2006). I reprint them for what they're worth. Somewhat strangely given the more recent problems, the 1/4 HP, 1750 RPM motor started and ran the generator okay. I think I had a 1.5" pulley on it (alas now broken), and the same 9" pulley on the generator, for a theoretical 290 RPM.

These results on each phase:

270 RPM, ~5 VAC, with no load
260 RPM with load...

~28 Amps with ~2.2 V out - (62 watts. demonstrating that the light aligator clip test lead I "shorted" the output with had ~.08 ohms resistance.)

~ 4 VAC with 14 Amps (56 watts - I got these results using, IIRC, 2 test leads as low ohms resistors. They got hot fast!)

~ 4.5 VAC at 10 Amps. (3 test leads?)

Let's see...
12 V [desired voltage] / (5 V * 2) [with both rotors] * 270 RPM = 324 RPM.
That's with no load. With a 10 amp load...
12 V [desired voltage] / (4.5 V * 2) [with both rotors] * 260 RPM = 346 RPM.

These RPM's are a little ways off of the ~400 RPM that it looks like it will actually take to get 12V. But the difference is easily explained by the big air gaps with the two rotors backed off so far.

6. Original Drawings

Before I did any actual work, I drew up some rotors and stators at actual size to get some sort of feel for various possibly applicable sizes. I knew I was going to make the stator an inch thick with one inch nail gun nail strips for iron.

I drew an 8 inch rotors unit. It sounded good, but it would only hold 8 of the magnets I'd ordered instead of 12 (has to be multiples of 4). And it looked pretty cramped for winding the (six) coils. Basically, it seemed like an altogether smaller unit that might have worked better with smaller and wedge shaped magnets.

I drew a 12" rotor such as is common in homebuilt windplants. I even bought the rotors from Forcefield, but it seemed to be bigger than was called for, given that I had an inch stator width to wind the coils in instead of just 1/2 inch. And I couldn't seem to find the matching car hub at the auto wreckers, to boot. (Without checking again to verify, I could probably have done 16 magnet rotors and 12 coil stators!)

I also drew up an 11" rotor. It looked good with ample elbow room for everything, but one would need to find or custom make the rotors - ugh.

I drew up the 10 incher, and later I measured the car disk brake rotors and found they were in fact 10". I had already bought a hub that happened to match them, intending to use it elsewhere on the wave power unit, but I put them together and there it was, the main working assembly for a generator, a good size and virtually ready to use!

Original concept drawings for 10" rotors unit

Finally, I drew up the idea for the multiple rotor/stator pairs with the 10" rotor size, showing a two stator, three rotor unit. This was what I'd planned to build originally, but I can get the voltage I want, with plenty of current drive capacity, at 400 RPM with just one stator and I have no real need to add another stator to run it at 200 RPM, or to get 120 amps per phase instead of 50 from my demo unit.

The basic "MPMG": Double the power with just one extra stator and rotor

7. More Experiments To Try

Here's some additional variations to the theme that would be nice to try out. I have other things to do, so I'm going to print this article without having tried them.

If someone wants to buy a generator off me to cover the cost of the parts and at least some of my time, I could try out some of these variations and write them up here in a future edition, with acknowledgement to the funder.

1. Different thickness stators.

I'm not convinced the one inch stators are optimum - it was just a rough guess on my part to try that thickness. I'd like to try 1.5 inch and then 2 inch thick stators. Naturally, one can wind twice as much copper coil in two inches as in one, either more winds (higher voltage) or fatter wire (higher current capacity). Balanced against this is some reduction in the magnetic field strength (in spite of the nails), which means the performance won't be double - but it may be well worth while for a few extra shekels of copper, nails and resin. If it's promising enough, what about 3 inches? The one inch stator is already double the thickness of the Pigott 1/2 inch windplant stators, which have no iron. But if a 1.5" or 2" stator should somehow prove worse than the 1" or no better, I'd try 3/4". Somewhere there must be an optimum for this design with this size magnets!

This is an obvious experiment that will determine the best performance thickness with regards to the overall cost of the generator versus the amount of additional power density achieved. Cooling air channels should be considered, as I did with the first one inch thick stator, only more so for a thicker cast.

A late thought: one could put in two one inch stators between two rotors and wire them together, instead of one two inch stator. That would be almost as effective and would try out the idea without making a big fat "white elephant" stator if by chance it didn't perform very well. If performance wasn't anything towards double the single one inch stator, one could then simply add a center magnet rotor between them for the three rotor, two stator design. The performance comparisons would be interesting.

2. Magnet standoffs.

I had intended, when I bought magnets with holes for screws, to experimentally elevate the magnets (which are 1/2" x 1" x 2") on steel blocks, eg, 1/4" x 1" x 2" or 1/2" x 1" x 2", to raise them off the steel end rotors a bit. I'd be interested to see (by measuring and comparing the performance) if this improves the magnetic field depth and-or decreases the rotor iron losses, however marginally. If it does, it's probably worth it, as the cost would be minimal.

However, the magnets were so strong I balked and glued them on the rotors as well as screwing them, rendering further experimentation impractical without buying more magnets and rotors.

3. Reversing Rotors.

In company with the 1/4" standoff blocks above, it should be possible to reverse one or both end rotors (ie to mount the magnets on the recessed face), which would reduce the overall length of the unit by one to three inches. (Without the standoff blocks (1/4" or taller), the raised center hub of the rotor sticks up farther (~3/4") than the magnets (1/2"), at least on my rotors, and thus would be in the way of the stator.)

Owing to the recessing of the hub (bottom rotor) and the nuts (top rotor), this could make the overall length even shorter than flat rotors. The idea of flat rotors shouldn't be overlooked for quantity production, but disk brake rotors seem pretty cheap compared to any custom made piece below at least 50's quantity. (unless the custom pieces came with all the holes including the magnet holes ready made, which would be a good saving in labour.)

A last minute thought: A flat rotor with a large enough center hole could mount on the bottom of the hub plate instead of the top, saving as much space as a reversed brake rotor and also leaving a center space for cooling air to reach the magnets, though one must drag the magnet rotor past the hub assembly to mount it. (Wrap the hub in padding?)

4. Evening the Distribution of Nails to Eliminate Cogging.

My unit coggs madly (at 36 points - where 9 coils and 12 magnets cross). If the nails were evenly distributed, one could much reduce that, in theory even eliminate it. I didn't try. To get max power, I just packed the centers of the coils with nails, and wound some more into the coil windings. There wasn't much room for more elsewhere. (I had hoped it wouldn't be as bad as it is.) If I make another one...

5. Skewed Rotors for Cogging Reduction.

Skewing the rotors a bit should tend to even out some of the high cogging. With the 5 stud axle hub, turning the top rotor by one position is a 72 degree turn. Since two magnets of twelve is 60 degrees, this would offset the magnets by 12 degrees. (A four stud hub will just give you the option to have opposite or like poles facing each other.) There are 5 (surprise, surprise) possibilities, though two are just variations: in-line, +/-12 degrees, and +/-12 degrees but having like poles (almost) facing each other. You can't get like poles exactly lined up. (Note: If you get the magnet polarities backwards on the second rotor, you can't get the rotors to line up with opposite poles across from each other. Not noticing this, I got it right by 50-50 dumb luck, a technique I don't recommend!)

I actually tried skewing rotors briefly and compared the cogging, but the test was pretty much meaningless: I didn't check the gaps, nor note which way the handle of the torque wrench was pointing (the 5 bolts are off center, so it matters). The only thing I found out was that there were still just 36 cogging points. My current thought is that backing off the rotors from the stator probably has about the same effect as skewing them. I did in fact back off the rotors and the air gaps are about 0.15 inch. (So much for the exact, careful tolerances!)

6. Flywheel(s).

This I will probably try. Although my unit should extract power from waves both rising and falling, waves provide power very unevenly, and a good flywheel would help even it out. The magnet rotors are pretty heavy. Since my bolts are a couple of inches longer than necessary, I'll probably just try spinning a couple of extra disk brake rotors as additional flywheels. Flywheels will be less effective smoothers at these low RPM's than at a higher speed, but they can't hurt.

-the end for now-