[bwilson4web]

Another hard problem comes from testing by rotating the magnet:

What I found is the magnetic field strength, measured at the nickel plating, is only 0.477 T. Without some sort of paramagnetic material to concentrate the field, it is way too weak to generate appreciable voltages needed to moderate the current flow.

So I've looked at prototyping a dual-disk configuration:

But the problem is the active, voltage generation is really just the outer rim of the disks. Two disks are less bad but the air gap means the field intensity is likely to be less than the 0.477 T measured in the nickel plating of the rotating magnet.

I'm interested in a homopolar generator because of the potential of a high velocity conductor. So one obvious solution is a rotating cylinder with parametric material from the magnet to concentrate the magnetic field through the rotating conductor. It turns out there is a 1994 patent for such a device that rotates two magnets in a conductive cylinder. But the challenge is the relative mechanical strength of the magnet in a high G field that risks mechanical fracture. This would not only unbalance the rotating mass but even if the bearings held, the brittle, magnetic material could collapse into little more than randomized, dust without an external field.

So I have a mental image of the fixed magnet with an outward tapered, ferrous ring on each pole designed to concentrate the magnetic field radially outward. After a gap, there is a ferrous pipe with a fixed gap between the two. Within this gap, is a cylinder with two, conducting rings between the magnetic poles. Bushes pickup the voltage and again, put them in series. This is not a trivial design but it maximizes the voltage generating material in the strongest possible magnetic field at the highest possible velocity.

Bob Wilson

[jdenenberg]

Bob,

This would make a great senior project at Georgia Tech (with you as the customer/mentor).

JeffD

[bwilson4web]

jdenenberg said:

. . .
This would make a great senior project at Georgia Tech (with you as the customer/mentor).

Or better yet, U. Texas where some of the really huge, homopolar work has been done. I have to admit every time I run into a problem, it just 'fires me up again.' <GRINS>

I have another test I want to run, a smaller model, that will let me know whether this scheme will ever work. I also need to model it but right now my 'sense' is the ratio of active volume, the space that actually generates electrons, is relatively low. I suspect that when I model the generator versus the traditional stator I'm going to find a low 'power density' and this may be the fatal flaw.

I want a small, dense, up to 20kW generator driven by a turbine.But to achieve this, I need to maximize the moving conductor speed (aka., Lorenz voltage equations.) This means a cylinder, not a disk, needs to be the current generating 'volume'. Now layers can increase the effective voltage and wired in series, we can reduce the current needed for a given power rating. But it also means tricking a very strong, magnetic field through the rotating cylinder. . . . This is not trivial but I think I know enough to make a model and test. <grins>

It is great to be in 'new territory.' <GRINS>

Bob Wilson

[bwilson4web]

I continued working on a high-speed, homopolar generator that is based upon a rapidly rotation cylinder. This is needed to maximize the velocity of the conductor, an optimum design. I got this far and realized I needed some high-speed, bearings. Without distributed support, any shaft based, rapidly rotating mass is unstable, bend the shaft, and it will fly apart:

This design has the following features:

1. Magnet manifold - a layer of steel washers tapered so their cross-section is the same as the surface area of the poles but carried to the rim. At the rim, the surface area is equal to the pole surface area. Given steel has 100 times the permeability of air, we are expecting the bulk of the magnetic field to be concentrated at the rim. Note that at the outer rim, a recess helps concentrate the magnetic field into the magnet manifold to maximize the field in the working gap. We are trying to minimize the field leaking through the center hole. The best magnet would have no center hole or the smallest one needed to mount the magnet and not leak any or minimum field.
2. Gap manifold - a steel cylinder circles the magnet with a recessed, air-gap to encourage the field to take the steel path, not the air gap. We want the magnetic field to concentrate between the brushes.
3. Brush ports - offset, they hold multiple, dremel tool brushes, 0.125" square, in a spring tensioned housing. For testing, we are less interest in supporting high current flows as these are probes to measure the electric field.
4. Aluminum cylinder - supported by high-speed bearings, the cylinder rotates as fast as possible. Before trying to reach speeds of 10**4 rpm, we need to know if the magnetic field through the enhanced air-gap is strong enough to generate a usable voltage. If the field is too weak, there is no need to proceed with a fixed, non-rotating magnet.

If we can not get enough voltage, the only remaining option is to encase the magnet and field shaping washers in an aluminum cylinder and find out how fast it can rotate before mechanical failure. Re-enforced with S-glass, epoxy, the encapsulating cylinder becomes two, electrically isolated, voltage generators. These can be wired in series to double the voltage into more usable values. But searching for the high-speed bearings for testing the single cylinder led me to a short-cut.

The RC community has electric-powered, ducted fans with rpms up to 80,000 rpm driven by brushless motors. Some of the larger ones are rated up to 2 kW. I decided to test a B32, 1.4 kW motor in a rapid build:

Granted, the T04E turbine is over-sized for this application but we should be able to get useful data and build a functional prototype:

1. Use a high-speed, flexible coupling from the turbine drive shaft to B32 motor working as a three-phase generator.
2. Three, 10:1, torroidal transformers to bring the current down to ~6 A and voltages to higher ranges, 70-200V rms, for a boost regulator. Of course we have to deal with the peak, unloaded voltage so the storage cap has to be rated at 1.414*Vrms ~= 300V.
3. With a wide range of generator voltages, a power regulator, rated at 2 kW is needed to couple to the traction battery. The regulator is also the primary control system:

• Input load management - the motor counter torque keeps the turbine from running away. The maximum power available comes from the turbine running as fast as possible but not exceeding the motor rpm limits.
• Output load management - the motor is rated at 65A continious but after 10:1 step-up, we're dealing with 6.5A which is well within the power ranges of common IBGT and power MOSFETs. This also becomes the current limit of the torordial coils which can provide all coils including the regulator.

Now there is a risk that the turbine may provide more power to 'run-away'. If the motor spins fast enough, the voltage could exceed the traction battery voltage and flow freely until something breaks or burns out. We've made the assumption the T04E is oversized and without water injection, this won't happen. But it is a real risk that mandates consideration of how to deal with over-speed conditions.

Bob Wilson

[bwilson4web]

Some new information on the B32, brushless motor workaround. Ohming out the motor, it is in a delta configuration. But given the 2,150 rpm/V rating, it really has to spin fast to generate usable voltages. Using a cordless, drill, I measured, .178 VAC, 25 Hz, and 0.770 A. I tried our Dremel tool but the chuck is too small even so, realistic testing remains a challenge.

I picked up a Harbor Freight, four-speed, rotary tool:

• 6,500 rpm
• 10,700 rpm
• 14,000 rpm
• 22,000 rpm

I've also ordered a shaft coupler rated at 19,000 rpm and 0.5 Nm but it hasn't arrived. When it does, I'll mount the rotary tool, shaft coupler, and B32 motor on a fixed stand. Then I can test the electronics.

Although voltage doublers are well know, they usually are two-phase AC. Three phase voltage doublers are a little harder to find. But thanks to William North's "High Power Microwave-tube Transmitters", Los Alamos National Lab, report LA-12786-MS or DE94005340, pp. 323-324, I found a circuit that made sense. FYI, a single-stage, three phase voltage multiplier was patented in the 1990s. But the circuit behavior is not obvious . . . North tries but his explanation is not quite clear. But other sources indicate each stage of a three-phase, voltage multiplier increases it by the square-root of 3 or ~1.73. This is because of the 120 degree phase angle and an excel spreadsheet quickly verified this behavior. So each stage is:

• V * 1.73 - single stage
• V * 2.99 - two stage
• V * 5.18 - three stage
• V * 8.91 - four stage
• V * 15.32 - five stage
• V * 26.35 - six stage

I started with trying to create a SPICE model but anyone who tries this on their own is likely to get frustrated very quickly. Then I found this Java Applet:

http://www.falstad.com/circuit/

So I made two models, a two-stage and three-stage with a 3-phase source, 0.050 ohm resistance, 20 kHz, and 10 V. This is the two stage unit that brings it up to 30 V:

Now what I've done is use a full-wave, configuration, similar to the patent, 5444357. The diodes were configured using Schottky voltage, 500 mV. Since it a simulation, I wanted to find out what sort of output impedance and lowered the resistance to 1.6 ohms to draw ~600 W. . . Regardless, I already have dual, 35A Schottky and 440 uF, 35 V caps enough to put together this circuit for bench testing.

For field testing, I'm planning to use at least three-stages to bring the voltage up to 50V. So this is that model:

In this model, I used only a 10 ohm load but easily we are getting 300 W. I may go to five stages bringing the voltage up to 89 V. At this level, a boost regulator can easily handle bringing the output up to 240-305 V with current protection to protect the Schottky current limits.

Now one of the curious aspects of a voltage multiplier is the current through each stage behaves strangely. If I used a transformer to step up the voltage, current in the primary would be in an inverse relationship to the secondary:

1A @120V ~= 0.5A @240V :: each cycle

But a voltage multiplier works more like a bucket brigade as each stage passes the charge minus diode and capacitor losses, up to the next stage AFTER the initial capacitor loads are passed up the chain. It took me a while to wrap my head around this concept (and I still have some doubts) but the model suggests this is the case. Best of all, each element within a stage stage works on the same voltage level so identical multiplier circuit elements are just replicated. Between each stage, the voltage increases and the current is reduced by the square root of 3. Low ESR caps are a better choice over cheaper ones.

So let's look at circuit design options for the range of voltages and available power:

• 1200 W / 240 V ~= 5 A.
• 1200 W / 305 V ~= 4 A.
• 20000 W / 240 ~= 84 A. 70A is largest DC current observed (assumes a 3-phase, 20 kW, AC generator)
• 20000 W / 305 V ~= 66 A.

Well this is how far I've gotten.

Bob Wilson