General Electric 3-Phase Motor Conversion
The original motor
This project started with an old air compressor was being thrown out at work. The compressor had cracked cylinders, and the old motor was going into the bin along with it. Steve to the rescue!
It was a perfectly serviceable motor. It was also a very old motor, probably built in the ’70s. Here are several pictures of the motor taken apart. The rotor and stator are “short and stubby”, and to some that would indicate a high number of poles on the motor. Actually this motor has a typical 4 poles, so perhaps the high proportion of diameter to length allows one motor design that can be wound with many pole numbers.
Converting the rotor
I had access to a machine shop with a lathe, CNC milling machine, and hydraulic press. My design reflects this by being rather complex. The lion’s share of the work converting my 3 HP motor was taken making the rotor. I had to make 3 parts, each sliding into each other with a (supposedly) light press fit. Actually, one press fit was to tight and the other too light, but some quick improvisation solved both problems.
I discovered that a piece of 4” pipe that I already had was exactly the right size to machine as a ring to mount my magnets, provided that I inserted a plug inside the piece of pipe to mate it with the shaft. You may find that starting with a solid block will be easy enough. By calling around local metal suppliers, you may be able to find some large hexagonal stock, which would allow you to mount several flat magnets on the flat sides without any machining. I couldn’t find any larger than 3”, so that was out of the question for me.
I built in a skew on the rotor by offsetting the magnet mounting screws. The corners of the magnets stick out more. This reduces the clearance between the rotor and the stator. That is, in order to preserve the clearance, I couldn’t make the magnets as close to the stator as I could without skewing them. The purpose of skewing is fully explained in the test results that follow.
I used screws to fasten the magnets to the rotor. I found some stainless steel screws that just “barely” responded to the powerful neos. Brass is another good choice, but the choice of head sizes & shapes may be limited at your hardware store.
The way I built it, I could still add 4 more magnets to the 4 bare flats. When I looked at the size of the coil groups on the stator, however, I concluded that two magnets per pole are sufficient, and the third would not likely pull its weight. I left room for extra magnets, though, and the extra holes allow for stripping one or two without tragic consequences.
The pipe wall is a bit thin for the purpose. The experts say to back the magnets with a thickness of steel or iron equal or greater than the thickness of the magnets themselves. I agree, and my part is thin only because I had the 4” pipe available.
Putting the motor back together could have been a risky operation without guiding it under control. I inserted the housing bolts backwards through a pair of boards to hold the stator down. The rotor was lowered on a block-and-tackle pulley above. This worked well after a few practice approaches. What also helped, was cutting up a plastic carton into strips to protect the magnets scraping the stator.
Converting the stator
I opened the stator up and I located the “star point”. This is the point in the motor where three wires are joined together. Each wire leads to each phase, and when connected the form a common central point and the three phases radiate outward until you reach the line wires in the connection box. I cut the star point, and soldered 3 new wires that I fed out to the connection box, too.
Now I can connect the motor in “Delta.” I attach the former outside to the former inside of the next phase, creating a triangle. Since this motor also allowed parallel/series connections of the coils, I now have a total of twelve wires!
I also improvised the connection box and terminal block out of what I scrounged up from a motor re-wind shop and the local hardware store. The 6-stud terminal block fit just nicely inside a cable box, which screwed just nicely on the threaded socket on the side of the motor, and accepts pipe thread attachments to shield the exiting cabe.
I am interested in the parallel connections because they allow double the current output of the series configurations. Before investigating in detail, I expect that a switch between 1-Y and 2-Y will be the easiest to create (or maybe 1-Delta to 2-Delta).
Because the stator of the motor has “teeth” between windings of wire, the magnets are attracted to the teeth. The magnets tend to pull to a point that is closest to a tooth, and it takes some extra torque to get it past that point. I experimented a little with the phenomenon when assembling it, to get a “feel” for the strength of the effect. I started with magnets on all 12 faces of the rotor aligned with the axis. No skewing at all. Inserting the rotor and assembling the faceplate went fine, and when I attempted to turn the shaft with my hand, it was obvious what was going on. I could rotate it one “step” at a time, 36 steps per revolution.
I then wrapped a string around the 1-1/4″ shaft and pulled with a spring-scale. It took 25 pounds to start the rotor turning against the first step. Once it got going, it took 11 pounds to keep it going at a constant speed. Later, I disassembled the rotor again and skewed all 12 magnets. I tapped holes in each rotor face with a slant and fixed the skew angle at 10 degrees. The 10 degrees refers to a view looking down the shaft of the rotor. You can see one end of the magnet and in your mind’s eye draw a line down to the axis. You can also draw a line down from the back end of the magnet and it is turned 10 degrees away. The 10 degrees works on a motor whose stator has 36 teeth (360 deg / 36 teeth = 10 degrees). If your stator has 24 teeth, then your magnets need a (360/24) 15 degree skew.
I re-inserted the rotor with all skewed magnets and performed the same pull test. The “steps” were much smoother, and to start the rotor took 14 pounds, and 8 to keep it going. The starting torque has nearly been cut in half, and the running torque is also down 1/3. As a last test, I removed 4 magnets, leaving the remaining 8 skewed. As expected, the pull test gave starting and running torque numbers 2/3 of the results with 12 magnets. The extra torque required to start was never equal to the running torque, but I don’t think there is a point where that occurs, because even the bearings behave that way. It always takes a bit more oomph to get going, than to keep going.
Testing the generator
My first test of the generator was to crank it with my hand and measure the voltage on a phase. I installed the bike tachometer to the face of the housing. I was able to crank it up to 230 RPM and 18 volts. Not bad.
Next was time to drive the generator with a motor. My drill press has been handy for testing the axial flux generator, but the vertical axis posed a problem with this converted motor, whose axis is normally horizontal. I settled for a half-twisted belt between them and put a pulley on each.
With 300 RPM at the generator shaft, over 24VAC could be generated on a phase. Connected in series star, about 42V would be available. When connected to a 12V battery, 4 Amps DC flowed at 300 RPM. This equals only 48 watts, but the drill press couldn’t turn it any faster. With all the belt losses, the drill press could not turn the generator faster than 300 RPM. Speeding up the drill press belt ratios only reduced its torque, which was then lost through belt slippage, and still the generator turned at 300 RPM!
With this information, I concluded that the open-circuit voltage available in 1-Y connection is 14V / 100 RPM. This is just enough that I can use it either for 12V charging with 2-Y connections, or for 24V charging with 1-Y.