Baldor 3-Phase Motor Conversion 2010

My most successful motor-conversion yet.  It is still running, 10 years later.

If I were to give it an output power rating, I would estimate it around 40 Amps at 28 Volts (1120 Watt), when rectified to DC. The wiring offers many series and parallel options, so a small change would make it suitable for a 48 Volt system with about the same power rating.  To make it last a very long time, I am using it at much less than its rated power.

The original motor

It’s another motor that I found discarded, but still usable. There is a bit of corrosion pitting on the shaft, but that’s easy enough to clean up. Taking it apart, I find a rather odd winding pattern in the wires, but otherwise everything is recognizable and ready to go. In fact, this motor is obviously of very high quality, just looking at how well it’s been built.

Specifications of this motor were:

  • Baldor M3211T
  • Power: 3 HP
  • Speed: 1725 RPM
  • Frame: 182T (Fan cooled)
  • Footprint: 7.5″ x 5.5″
  • Shaft: 1.125″
  • Rotor Diameter: 4.375″ OD
  • Stator Diameter: 4.4375″ ID
  • Stator Length: 2.9″
  • Stator teeth: 36
  • Current ratings: 8.6A @ 230V
  • Wire resistance per phase leg
    (wires 1 to 10): 2.5 Ohms
  • Wire resistance across the lines in Star
    (wire 1 to wire 2): 5.0 Ohms

The conversion process

Having done a couple of motor conversions now, I followed a process like this to get what I want:

  • Completely disassemble the motor
  • Measure everything that I can
  • Draw the motor to scale in CAD
  • Export the CAD drawing into FEMM
  • Use FEMM to select a magnet configuration
  • Decide to keep the existing rotor or make a new one
  • Turn the rotor to size (new or old)
  • Machine faces for the magnets on the rotor
  • Install the magnets on the rotor
  • Replace the old bearings with new ones
  • Make changes (if any) to the stator wiring
  • Reinsert the rotor back into the stator
  • Pull tests (open-circuit) to determine the new voltage range
  • Depending on the results of the test, change wire connections
  • Repeat tests, then conduct a run-up test using a lathe to drive the shaft

So it looks like a complicated job, but only a few of these steps require much work, and several can be skipped entirely. Rarely do you need run-up tests, and few would bother making CAD models before starting. Drawing it on paper is plenty! Taking careful measurements, though, is very important. Obviously you can’t end up with a rotor too big to fit back inside, but almost as bad is too much air space in the stator. It lowers the output and every millimeter counts.

Design choices (January 2009)

Arc Segment Magnets
1" x 3/4" Magnets

I used my computer to try to find the “best” choice. The software that I use to do this, called “finite element magnetic modelling” or FEMM for short, turns the diagram into beautiful graphic pictures of the magnetic flux paths. By making them as strong as possible, I am improving my odds of having a powerful alternator at the end. Here I can show the difference made by different arrangements of magnets. There is also a cost factor to consider – the curved magnets are much more expensive than brick-shaped ones!

Furthermore, cogging must be considered. ‘Cogging’ is the term for the tendency of motors with permanent magnets in them to stick in one position. If it’s really bad, wind turbine blades will not be able to turn in light winds. If it’s moderate, the blades may work all right, but there will be vibration. If it can be eliminated, then a very smooth wind turbine is the result.

1" square AND 1/2" square Magnets

Here is a very densely packed magnet configuration, because the smaller magnets fill up the empty spaces between the larger ones. Making the most magnets fit into the space available is usually the goal.

I finally chose this rather complicated configuration! This was as much a challenging idea as a practical one. I could fit 3 rows of magnets like these on the rotor by making it just a bit over 3″ long. This still fits effectively inside the stator which is 2.9″ long. By flipping the direction of the rows of magnets, I have cogging forces that cancel each other out.

In theory, at least…

The final design

Machining the rotor (February 2010)

I have access to nice metal-working machinery where I work, as long as I don’t get in the way of “real” work. On weekends I turned down solid steel stock on the lathe, then put the cylinder in the milling machine to put the faces on it. With that done, I pressed the old rotor off of the shaft, and then pressed this one on in its place. This didn’t work out perfectly because “press-fit” forces are huge when the parts are this big. I repeatedly opened up the bore through the rotor until it would go on. This took hours to get right because I’m not a seasoned machinist by any stretch of the imagination.

After I had it on, I had an idea, and drilled two more holes in the end of the rotor. You will see them in later photos.

Fastening the magnets

The last step was to drill and tap all of the holes. While I was doing this, I decided I needed to “lock” the rotor onto the shaft, due to the trouble I’d been having with the so-called press fit. I drilled some of the screw holes extra deep, penetrating a little into the shaft, then sinking dowel pins to the bottom. Those dowels I DID manage to press fit. The screws to go into these holes will serve to hold the magnets and to ensure the dowels stay in.

Putting these magnets on was a slow process, haha! I had to mix small batches of epoxy at a time, bond 2 of the little bricks, hold them in place with a clamp, then come back the next day for the next pair. If they weren’t clamped, they would slither and slide under the influence of the neighbouring magnets.

To finish it off, I added epoxy to fillet between each of the magnets’ gaps. Extra holding power, without completely submerging them. I’m afraid of the ability of the rotor to loose heat, and covering them with epoxy (an insulator) would be detrimental to that. I know some people haven’t had any such trouble, but I don’t want to take the risk.

I have test run generators like this before. I can tell you they do get hot!

The last thing I did to the rotor before assembly was to spray-paint it with stove paint. I found a white paint that’s resistant of high temperatures. Perfect for the tough working environment inside the motor where heat builds up rapidly and there are limited means of cooling. (This one at least has a fan to help.)

The stator wiring (April 2010)

I hadn’t decided yet whether I wanted to separate the star point in the Baldor. It has 9 wires connecting the windings already available in the connection box. This made series-star and parallel-star connections possible without any changes. My work with FEMM told me I would want parallel-star.

Adding wires to the stator takes some prior knowledge and careful cutting in the right spots. When I went about modifying my SUmo motor conversion, I made mistakes in the re-wiring, so I’m a bit more leery of just snipping away at this nice Baldor wire. The bundles are very tightly packed and I know that my skills would not get it back to such a high quality state if I were to cut them apart. For now, I decided to leave them alone.

Re-assembly (May 2010)

Sliding the rotor, containing 36 high strength magnets, into a steel stator, containing lots of delicate wire, is a dicey operation! Inadequate preparation can lead to broken magnets, severed wires, or smashed fingers!

I clamped the housing of the motor to my workbench, face up. Then I suspended the rotor below a sturdy ceiling attachment with a pulley. With this I was able to slowly lower the rotor into the stator, without allowing them to crash together. In addition to that, I put sheets of plastic around the edges of the housing so that where the rotor rubs it would not scratch the wire or chip the paint off of the magnets.

I should mention that before starting this, the motor should be ready to STAY together. Pulling the rotor in and out is risky and there is always some minor damage. Doing this too often invites disaster!

I made sure that the bearings fit on properly, and that their sockets were clean before putting them together, too. This motor has a fan disk, so I checked that it hadn’t been warped during all the previous work.

I used a piece of wire (about 3/16″ diameter) as a feeler gauge. The rotor is off center in the photo, so when it is centered the gap will be 3/32″. I was aiming more for 1/16″ gap between rotor to stator, but it seems I didn’t measure the stator accurately!

I put the end shell back on and, of course, gave the motor conversion its first trial spin.

De-cogging check

That first turn of the shaft was very satisfying. My work to prevent cogging paid off, and there are just a series of little “bumps” to feel as it turns. I took some starting torque measurements by wrapping string around the shaft or a pulley and pulling the string with a spring scale. The shaft would resist turning until I applied enough force.

The torque to kick it over varies a lot. I am sure the cause of this is the variation in magnet size that I used on the rotor. When the alignment of the rotor puts more large magnets than small ones in proximity to a stator tooth, the attractive force to be overcome is slightly larger than when mostly smaller magnets are in proximity and the larger ones are “in between” two teeth of the stator.


Initial testing of the generator

Now the fun begins! This alternator with permanent magnets inside will generate electricity at extremely low RPM’s. Simply by turning the shaft by hand I can produce dangerous voltages! I put a crank handle on the shaft, then measured the output voltages, just open-circuit at first.

I did these tests with the 3-phase output rectified and the digital multimeter set to DC. I intend to use the generator for battery charge power, therefore the AC values are not of direct use to me. Using the DC voltage measurement also avoids the vagaries of 3-phase power calculations.

The open-circuit voltage measurements are consistent: in series-star the electromotive force is 37 volts per 100 RPM. This is fantastic! 

150 RPM 57VDC 0.37 V/RPM
160 RPM 58VDC 0.36 V/RPM
110 RPM 42VDC 0.38 V/RPM
65 RPM 25VDC 0.38 V/RPM
280 RPM 100VDC 0.36 V/RPM
300 RPM 110VDC 0.37 V/RPM
160 RPM 57VDC 0.36 V/RPM
170 RPM 63VDC 0.37 V/RPM

Average EMF: 0.373 V/RPM

Different wiring schemes will therefore offer different potential speed ranges:

ConnectionEMF28V cut-inPhase Resistance
Series – Star37.3 V/100RPM75 RPM5.00 Ohm
Series – Jerry18.7 V/100RPM150 RPM2.50 Ohm
Series – Delta21.5 V/100RPM130 RPM1.67 Ohm
Parallel-Star18.7 V/100RPM150 RPM1.25 Ohm
Parallel-Jerry9.3 V/100RPM300 RPM0.63 Ohm
Parallel-Delta10.8 V/100RPM260 RPM0.42 Ohm

I experimented with it in other ways.  These tests were mostly just for fun:

Selecting the right 3-phase wiring configuration (May 2010)

Having done these tests, it’s obvious that there is so much EMF that cut-in would be too low unless I use a parallel winding connection. Furthermore, even Parallel-Star has a very low cut-in speed. I expect to be using 10-foot diameter blades on this generator. If I designed for a cut-in TSR of 8 (for example), the corresponding wind is only 2.5 meters per second. But at this wind speed, there is less than 10 Watts of power available to turn the blades! Since it takes about 15 to 25 Watts just to overcome the cogging, slight though it is, I can already see that there is no need for a cut-in speed as low as 130 RPM.

This leaves me with the remaining choices of either Parallel-Jerry or Parallel-Delta. I would prefer to use Jerry, rectifying each phase separately. Delta isn’t a great choice for a wind turbine, IMHO, because currents can circulate in the windings instead of running through the power lines, wasting energy as heat that should otherwise be useful.

The only way to accomplish a Jerry connection is to separate the Star point. Just what I wanted to avoid before!

By the way, the term “Jerry” comes for a member of Fieldlines’ forum who championed this means of generator connection a few years ago, especially for motor conversions. Looks like I’m following in somebody’s footsteps…

Rewiring the stator (June 2010)


Cracked the case open again, extracted the rotor, and examined the wiring. After a few minutes of carefully looking for the pattern of wire connections, I could see which joints led to the wires coming out to the connection box. This left three that didn’t come out. Each wire joint is covered with a fiberglass sleeve. The sleeves are red (hard to distinguish between the burgundy-orange wire varnish). Between a couple of sleeves I found a wrap of tape, which is what I suspected contained the star point all along.


With the joint identified, I carefully cut away the binding strings and peeled the wires apart where I needed to move them. At the factory, completed stators are dipped in varnish to protect them, and this varnish sticks everything together. Not only does it prevent separations, but by filling all the little spaces between turns of wire, the motor “hums” less. Literally. Anyway it makes separating the sleeves and wires a job worth doing slowly, lest I pull their own protection off.

While removing the tape, I was surprised to find a bit of black soot. I tested the wires again for continuity but didn’t find any problems. The soot seems to originate at either the tape or the sleeve, or perhaps something was contaminating this joint.

I cut the wires apart (you see proof in the photo that each phase was wound “two-in-hand”.) Actually, I’m not sure about this. The stator is wound in the “Wave” style, so my assumptions about this may be incorrect. To each phase I connected my own wires. These three wires I tightly twisted to the motor wires and thoroughly soldered them to keep them together. My new wires run out to the connections box, making for a full complement of 12 wires.

Tied the wires back up tightly (sorry no boy scout knots) and I might even look for a way to put a varnish on this, as protection against vibration and the elements. Each joint is in a new sleeve that I put on too.

Checked continuity of the joints and that all of the connections had the same resistance. It appears that between any two of the split-phase wires (eg, between wire #1 to #4, #7 to 10, etc.) they are all between 1.0 to 1.1 ohm.

And finally, I hooked it all up in Parallel-Jerry, and checked that all phases are consistent again. And they are. Success!

Checking the wiring (June 2010)

Here’s yet another check of the wiring, on my workbench, to verify that I’ve done my work properly. This led to a little surprise. I had done my math wrong a few weeks ago, leading me to think that a Jerry connection (individually rectified phases) would offer a cut-in speed near my target of about 150 RPM, but that was actually incorrect. I’ve fixed the table of measurements above, too, to reflect this.

The cut-in speed in Delta and in Jerry is virtually the same. Either it’s 112 RPM or 223 RPM, too slow or too fast for a typical 8-10 foot diameter wind-turbine rotor. I must look again at Parallel-Star. Most of the time it must deliver 26 to 28 volts to the fully-charged battery, so the 128 RPM cut-in drifts up to almost 150 RPM. This will do, but if I hadn’t been successful at minimizing the cogging, the start-up behaviour would be poor.

Performance testing of the generator (June 2010)

Using the lathe is very convenient for me. It’s a 5HP power source that can drive my generator at any speed I care to use.

I hauled a whole bunch of batteries with me, and I wired up that board in advance so I wouldn’t piss around while at work. The planning paid off because I got in, did all the tests, and packed out in 5 hours. I brought my old Wattmeter with me, and it was indispensable. It’s much simpler just reading off that steady needle than trying to decide which of the 3 decimal places to keep on a DMM. Speaking of DMM’s, they actually agreed with each other most of the time, a nice surprise. Just to illustrate and have fun, I ran the current (about 40 amps) through a thin piece of wire (last picture) and made it glow nicely!

The reason for all this stuff is so that I can measure both the input torque and RPM. With those figures I have the input power. With the wattmeter on the output side, the analysis gets pretty simple: efficiency is power_out divided by power_in.

All of the tests used 24V batteries. Maybe I could have tried 48V but I can’t contemplate a 48V system at home right now, so it would be totally academic. The goal of the tests was to size up the prop, and identify the preferred connection scheme for the generator. 

The test results are graphed here: