3D Printed Axial Flux Alternator and Dynamometer

STOP!! READ THIS FIRST!!! This is a record of a project that's still in development, please feel free to offer support.

My eventual goal is that this type of motor/alternator can become a parametrized open source design. A user should be able to enter some parameters, like torque, speed, current, volts/rpm, common magnet sizes and perhaps space available, and a series of 3D printable .stl's and .dxf cut files should be generated.

What I've done is created a platform that can validate a simulated design, which can then be evolved to a more optimal device by community.

In part, this is one reason that I've set this up with a dynamometer. A dynamometer measures torque and speed to allow hp, or shaft Watt's to be measured. In this case I've built the alternator with a pass through, stationary shaft, which makes setting up a dynamometer system simpler, and so it can be configured to be driven as a motor by an RC ESC (I hope), and torque measured on the output, as well as speed, V and Amps, allowing motor efficiency to be determined.

For my purposes it can be driven by a variable speed motor (surplus from cordless drill, with step down gearing), and shaft torque input measured, as well as V and Amps out, allowing real efficiency to be generated, and the expected turbine loads to be simulated.

In this mode I hope to use an RC ESC capable of regenerative braking, and perhaps an Arduino to control the load my VAWT carries to achieve MPPT (Multi Power Point Tracking).

MPPT is used in solar as well as wind turbine control, but a it's a bit different for wind. With wind power a big issue is that as wind speed doubles 10km/hr to 20km/hr, the energy available from the wind increases by the cube, so by 8 times. If 10W were available at 10km/hr, then 80W are available at 20km/hr. It's great to have more energy, but alternators output only doubles as speed doubles. So if you have the perfect alternator for a 20km/hr wind, its load may be so strong that at 10km/hr it won't even start.

What MPPT does is use a heavy duty solid state switch, to disconnect and then reconnect an alternator very quickly. It allows you to adjust how much load an alternator carries, and the Multi, of the MPPT, means that you can set different loads for different speeds.

This is very useful, as all types of turbines collect their maximum energy when the load is matched to the energy available, or wind speed.

SO

This is not a recipe, though I believe it could be copied from what I've posted, and would be happy to supply further information but I suggest that the best option would be to suggest improvements to me, before the Circuits and Sensors contest ends, so that I can consider, respond, and perhaps improve this instructable.

I will continue to update, revise and add information, so if it's interesting now, you might want to check in again in a bit, but I hope to get quite a bit done before the Sensors contest ends on July 29/19.

Also, I'm not a particularly social beast, but I do like a pat on the back now and then, and that's one of the reasons I'm here :-) Tell me if you enjoying seeing my work, and want to see more,, please :-)

This project came about because I wanted a controllable load for testing my turbine designs, and I wanted it to be easily reproducible, so that others could use it too. To this end, I constrained myself to designing something that could be built with only a FDM printer, no other machine tools needed. There don't seem to be many commercial products that fill the need for a high torque, low speed, non cogging alternator, though there are a few from China. In general there isn't much demand because gear systems are so inexpensive and electricity is so cheap.

What I wanted was something which produced around 12V at 40-120 rpm, and around 600-750W at 120-200rpm. I also wanted it to be compatible with inexpensive 3phase PMA controllers from the RC world (ESC's Electronic Speed Controllers). A final requirement was that it be an out runner (case or shell with magnets rotates, while shaft with stator, is stationary), with a shaft that passes all the way through the case, and a stator that clamps to the shaft.

This instructable is a work in progress, and I'm posting it so that people can get a view of the process, not so much because I think they should copy it. A key thing I'd change is that the wire backing plate I built isn't nearly strong enough to properly channel the magnet fields around the ring, so a great deal of the magnetic flux paid for in those magnets is wasted out the back. When I redo the design, which I will be doing shortly, I'd likely do it the magnetic backing plates as cnc cut steel plates. Steel would be fairly inexpensive, much stronger, and would simplify most of this build. It was interesting to do the FDM/wire/plaster composites like I've illustrated here, and with iron loaded PLA, things would have been different too. I decided though that I wanted something that would really last, so steel plates.

I've made good progress on this version, which I'll use for testing this VAWT. I'm not quite there in terms of low voltage performance yet. I think my Wattage/Torque is in the right ballpark, I'll update as things progress but at this point what I've got has a good chance of being the controllable load I need. When dead shorted it seems to be able to provide quite a bit of torque resistance, more than enough to test the turbine. I just need to set up a controlled resistance bank, and I've got a friend who's helping me with that.

One thing I'll briefly address is that like many people now, I have had a 3D (FDM-using PLA) printer for a few years, which I've had 20-30kg's of enjoyment from. I often find it frustrating though as parts of any size/strength are either expensive and very slow to print, or cheap, fast and flimsy.

I know how many thousands of these 3D printers are out there, often doing nothing because it takes to long, or costs too much to make useful parts. I've come up with an interesting solution to stronger faster parts from the same printer and PLA.

I'm calling it a "poured structure", where the printed object (made up of 1 or more printed parts, and sometimes bearings and shafts), are made with voids designed to be poured full of a hardening liquid filler. Of course some of the obvious choices for a poured fill would be something like epoxy loaded with short strand chopped glass fibre, which could be used for high strength and light weight assemblies. I'm trying out some lower cost, more eco friendly idea's as well. The other side of this "poured structure" assembly, is that the cavity or void you are going to be filling, can have small diameter high tensile elements, strung pre tensioned on the printed "mould/plug", which, makes the resulting structure a composite in materials, and in structure, part Stressed Skin (the PLA sheath), but with a high compressive stregth core that includes high tensile strength elements as well. I'll do a second instructable featuring this, so will talk about it here, only to cover how it pertains to this build.

The PMA consists of 3 assemblies, each assembly containing or using a variety of parts and materials.

From top (bearing side) to bottom (stator side),

1. Bearing Carrier and Top Bearing Array

2. Stator

3. Lower Magnet Array

1.The Bearing Carrier and Top Magnet Array

For this I used 3D printed parts listed above

1. 150mm8pole upper mag and bearing support CV5.stl,
2. bearing side inner plate
3. bearing side outer plate
4. 1" ID self aligning bearing (like used in standard pillow blocks++add internet link),
5. 25' of 24g galvanized steel wire
6. 15' of 10g galvanized steel wire
7. 2 rolls coarse steel wool

Optionally the heavy steel wire and steel wool could be replaced with steel backing plates, laser / water jet cut, or a 3D printed magnetic backing plate might be possible (but heavy some steel wire is still a good idea as it will resist plastic deformation over time). I've tried casting a backing plate with epoxy loaded with iron oxide powder and had some success. Improving the flux coupling between magnets in the array laterally by using a more effective backing plate should increase Volts out at lower rpms. It is also good to keep in mind that this is the major structural component, and the backplate transfers the forces from the magnets to the jacking posts. The magnetic forces pulling the plates toward each other can be hundreds of lbs, and the forces increase exponentially (Cubed, to the third power) as the plates get close together. This can be very dangerous, and care must be taken with tools and any other objects which may be attracted to the assembled plate or it's back!

I used about 300ft of 24g coated magnet wire in the windings which I'll cover in detail later.

In this axial flux alternator, to minimize cogging, and maximize output, I'm using two magnet arrays, one on each side of the stator coils. This means that no magnetic core is needed to draw the magnetic field through the copper windings, as most motor/alt geometries do. There are some axial flux designs that use ferris cores, and I may try some experiments that way in the future. I'd like to try some 3d printable iron loaded material.

In this case, I've chosen an 8 pole magnet array in about a 150mm circle, using 1"x1"x0.25" rare earth magnets. This size was to ensure that all parts would fit on a 210mm x 210mm print bed. In general I sized this alternator first by understanding that the larger diameter, the better in terms of volts per rpm, so made it as large as would comfortably fit my print bed. FYI, there's more than one reason larger is better: more room for magnets, the farther the magnets are from the centre, the faster they travel, and there's more room for copper too! All these things can add up fast! However a conclusion that I've come to is that in this size range, a conventional flux system might be a better home build. Small rotors don't have much room, and things can get quite tight, especially if you are doing a through shaft like I've done in this design. Also if your magnet (radial length) is small relative to your rotor diameter, like in this one, (roughly 6" diameter to 1" magnet), then the winding gets a bit strange with the inner end winding being only about 1/2 the length of the outer.

Back to the instructing! The way I've assembled the magnet plates of this alternator is to first glue the magnet plate (green) to the red flange/backing plate. I then placed the magnet plate on a few thin layers of plywood (about .75" thick), and placed the both on a heavy steel plate, to allow the magnets to clamp the assembly in place. Then I wound steel wire, onto the back of the magnet plates. This didn't go quite as I'd hoped. The strong magnetic field pulled the wire toward the centre of the magnets, and I was not successful in bending each row, of wire to perfectly fit the next spot, without jostling the first wrap. I'd hoped that I could just spool the wire in, and the magnetic flux would lock it down. Next I tried cutting rings of wire, and this was better, but still far from what I'd hoped in terms of getting a nice consistent backing plate from wire. More complex ways of getting this done are possible, and might be worth future experimentation. I also tried using steel wool, compacted in the magnetic field, as a backing plate, or flux return path. This seemed to work, but the actual iron density didn't seem to be very high, so I didn't test its effectiveness, in part because I believed the wire structure to be important to the mechanical loads on the magnet plates. The steel wool also may be worth future investigation, however water jet cut steel plates are likely the next option I'll try.

Next, I took the orange 3D printed part, and wove wire through and around it, along what seemed to me to be the directions of highest load, bolt to bolt, and bolt to centre a few times on each corner. I also wound it around the bolt holes where the all thread rod passes as jacking posts to keep and make adjustable the spacing between the plates.

After being satisfied that the magnet plate and flange were good enough, and the orange backing plate was satisfactorily threaded with reinforcing wire, I joined the two with glue. Care needs to be taken as this glue joint will need to be water tight, or close. I had leaks the first two times, and it's a mess, wastes a lot of plaster, and is more stress than you need. I'd recommend keeping some blue tack or other bubble gum like non permanent adhesive around to patch leaks quick. Once the parts are joined, fill with the reinforcing material of your choice. I used a hard plaster, modified with PVA glue. The plaster is supposed to reach 10,000 psi compressive, but not much in tension (thus the wire). I'd like to try epoxy with chopped glass, and cabosil, or concrete and admixtures.

A handy thing about the plaster, is that once it kicks you have quite a bit of time where it's hard, but fragile and leaks or blobs can easily be scraped or knocked off.

In this design, there are two magnet plates. One has a bearing, a standard 1" pillow block self aligning unit. I pressed mine into the magnet array early on. For the application I've designed it for, a second bearing will be located in the turbine above the alternator, so I only used the one self aligning bearing. This was a bit of a pain in the end. These parts could also be assembled with each magnet plate having a bearing, if the output wires from the stator were lead internally through the mounted shaft. This would allow contra rotating propellors to be mounted to a common, non rotating shaft/tube.

In keeping with my theme of trying to explain what I've done, and why it seemed like a good idea at the time, the stator will require a bit more space.

In a PMA, generally the windings are stationary, while the magnetic assemblies rotate. This is not always the case, but almost always. In an axial flux assembly, with the understanding of the fundamental "right hand rule", it's understood that any conductor encountering a rotating magnetic field, will have current and voltage generated between the ends of the wire, with the amount of useful current being proportional to the direction of the field. If the field moves parallel to the wire (eg, in a circle around the axis of rotation), no useful current will be generated, but significant eddy currents will be generated, resisting the movement of the magnets. If the wire runs perpendicular, then the highest voltage and current output will be reached.

Another generalization is that the space within the stator, through which the magnetic flux passes while in rotation, for maximum wattage output, should be filled with as much copper, all radially laid, as possible. This is an issue for small diameter axial flux systems, as in this case, the area available for copper near the shaft is a fraction of the area at the outer edge. It's possible to get 100% copper at the inner most area encountered by the magnetic field, but within this geometry that only gets you maybe 50% at the outside edge. This is one of the strongest reasons for staying away from axial flux designs that are too small.

As I've said previously, this instructable isn't about how I would do it again, more it's to point in some directions that seem promising, and show off some of the potholes that can be reached on this path.

In designing the stator I wanted to make it as flexible as possible in terms of volts output per rpm, and I wanted it to be 3phase. For maximum efficiency, via minimizing eddy currents generated, any "leg" (each side of a coil should be thought of as a "leg") should only encounter one magnet at a time. If magnets are close together, or touching as is the case in many high output rc motors, the during the time the "leg" is passing through the magnetic flux reversal, significant eddy currents will be developed. In motor applications this doesn't matter as much, as the coil is energized by the controller when it's in the right locations.

I sized the magnet array with these concepts in mind. The eight magnets in the array are each 1" across, and the space between them is 1/2". This means that a magnetic segment is 1.5" long, and it has space for 3 x 1/2" "legs". Each "leg" is a phase, so at any point, one leg is seeing neutral flux, while the other two are seeing accending flux and declining flux. Perfect 3 phase output, though by giving the neutral point this much space (to minimize eddy currents), and using square (or pie shaped)magnets, the flux almost peaks early on, stays high, then falls off to zero quickly. This type of output is I think called trapezoidal, and can be difficult for some controllers I understand. 1" round magnets in the same apparatus would give more of a true sine wave.

Generally these home built alternators have been built using "coils", donut shaped bundles of wire, where each side of the donut is a "leg" and numbers of coils can be attached together, in series or parallel. The donuts are arranged in a circle, with their centres aligned with the centre of the magnet path. This works, but there are some issues. One issue is that since the conductors are not radial, a great deal of the conductor is not passing at 90 degrees to the magnetic field, so eddy currents are generated, which appear as heat in the coil, and resistance to rotation in the magnet array's. Another issue is that the because the conductors aren't radial, they don't pack together as nicely. Output is directly proportional to the amount of wire you can fit in this space, so output is reduced by non radial "legs". While it would be possible and is sometimes done in commercial designs, to wind a coil with radial "legs, joined top and bottom, requires 2x as much end winding as a serpentine winding where the top of one leg is joined to the top of the next appropriate leg, and then the bottom of that leg is joined to the next appropriate leg, and on and on.

The other big factor in Axial flux alternators of this type (rotating magnets above and below stator), is gap between the plates. This is a cube law relationship, as you reduce the distance between plates by 1/2, magnetic flux density increases by 8x. The thinner you can make your stator, the better!

With this in mind I made a 4 lobed winding jig, set up a system for measuring out about 50ft of wire strands, and wrapped the jig 6 times, creating wire bundles about 6mm diameter. These I fit onto the blue spacing ring, tying them down through the holes so the wire ends came out the back. This was not easy. It was helped a bit by having carefully taped up the bundles so they weren't loose, and by taking my time and using a smooth wooden forming tool to push wires into place. Once they were all tied in place, the blue spacing ring was placed in the largest of the light green forming tubs, and with the help of the dark green donut forming tool, on the other side of the light green tub, carefully pressed flat with a bench vice. This forming tub has a groove for the tie wire twists to sit in. This takes time and patience as you carefully rotate about 1/5 turn, press, rotate, and keep going. This forms the disk flat and thin, while allowing the end windings to stack up. You may notice that my 4 lobed winding has straight "legs" but the inner and outer connections are not round. This was supposed to make it easier for them to stack. It didn't work out that well. If I were doing it again I would make the inside and outside end windings follow circular paths.

After getting it flat and thin, and the edges packed down, I wound a flat ribbon around the edge to compact it, and another up, down and around each leg and then to the one next to it as well. After this is done you can remove the tie wires and switch to the smaller pressing tub, and go back to the vice and press it as thin and flat as possible. Once it's flat, then remove from the press tub. Instead of the complex process of carefully waxing and coating moulds like this with release compounds, generally I just use a couple of layers of stretch wrap (from the kitchen). Lay a couple of layers in the bottom of the mould and lay the fibreglass onto the stretch wrap. Next add the stator mounting tube, which fits to the top of the light green forming tub, but has the layer of stretch wrap and fibreglass between. Then add the stator winding back into place to push down both the stretch wrap and the fibreglass and lock the stator mounting tube into place. Then return to the vice and press flat again. Once it's fitting well into the tub, with the stretch wrap and fibreglass sandwiched in, then fibreglass cloth is added (with a hole in the centre for the stator mounting tube).

Now it's ready to pour the bonding material, epoxy, or polyester resin are commonly used. Before this is done careful preparation is important as once you start this process you can't really stop. I used a 3D printed base plate I'd previously made, with a 1" hole in the centre and a flat plate around it. I used a 16" piece of 1" aluminum tube, that the stator mount tube would fit over and be held perpendicular to the flat plate. The green forming tub, stator winding, and stator mounting tube were slid down to sit on the flat plate. Before mixing epoxy, I first readied 4 pieces of shrink wrap, and carefully placed a 5th piece on the dark green forming donut, so it would have the minimum wrinkles on the face against the stator winding. After mixing the epoxy and pouring it on the fibreglass cloth, then I carefully laid down the stretch wrap around the 1" tube, and placed the green forming ring on top of it. I'd also prepared a couple of old brake rotors, which gave some weight, and sat nicely on the green forming donut. After this I put an inverted pot on top of the brake rotors, and on top of the pot I stacked about 100lbs of stuff. I left this for 12 hours, and it came out about 4-6mm thick.

There are a number of measurable inputs and outputs from the alternator, and measuring them all, at the same time is not easy. I'm very fortunate to have some tools from Vernier that make this much easier. Vernier makes educational level products, not certified for industrial use, but very helpful for experimenters like myself. I use a Vernier data logger, with a variety of plug and play sensors. On this project I use hall based current and voltage probes, to measure alternator output, an optical sensor to give alternator speed, and a load cell to measure torque input. All these instruments are sampled about 1000 times per second and recorded to my laptop, using the Vernier logger as a AD passthrough device. On my laptop the associated software can run real time calculations based on the inputs, combining torque and speed data to give real time input shaft power in Watts, and real time output data in electrical Watts. I'm not done with this testing, and input from someone who has a better understanding would be helpful.

An issue that I have is that this alternator is really a side project, and so I don't want to spend too much more time on it. As it is, I think I can use it for a controllable load for my VAWT research, but eventually I would like to work with people to refine it, so that it's an efficient match for my turbine.

When I started into VAWT research about 15 years ago, I came to realize that testing VAWT's and other prime movers is more complex than most people realize.

A primary issue is that the energy represented in a moving fluid, is exponential to it's rate of movement. This means that as you double the speed of a flow, the energy contained in the flow increases 8x (it is cubed). This is a problem, as alternators are more linear and in general, if you double the rpm of an alternator, you get about 2x the watts.

This fundamental mismatch between the turbine (energy gathering device), and the alternator (shaft power to useful electrical power) makes it difficult to choose an alternator for a wind turbine. If you choose an alternator match for your wind turbine that will generate the most power available from 20km/hr winds, it will not likely even start to turn until 20-25km/hr as the load on the turbine from the alternator will be too high. With that alternator match, once the wind is above 20km, not only will the turbine only capture a fraction of the energy available in the higher velocity wind, the turbine may overspeed, and be damaged as the load provided by the alternator isn't high enough.

In the last decade a solution has become more economical because of the drop in price of control electronics. In stead of trying to match a range of speeds, the designer calculates the maximum speed that the device is meant to operate at, and chooses an alternator based on the amount of energy and ideal speed for the turbine at that speed, or a bit above. This alternator if connected to it's load, would normally provide too much torque at the low speed range, and the overloaded turbine will not capture all the energy it could have if it were properly loaded. To create the proper load, a controller is added which momentarily disconnects the alternator from the electrical load, allowing the turbine to speed up to the proper speed, and the alternator and load are reconnected. This is called MPPT (Multi Power Point Tracking). The controller is programmed such that as the turbine speed changes (or alternator voltage rises), the alternator is connected or disconnected, a thousand times per second or so, to match the load programmed for that speed or voltage.