Another Motor Conversion

[SparWeb]

Hi

It's been a while since I updated the forum on my progress.  Most of it has been recovering from a wrecked windmill in the spring.  I had the tower back up in relatively short order, but replacing the windmill has been a long process.

To replace the generator, I wouldn't be able to find an exact copy of the one I'd converted before, but I did need to find something similar.

I had in my garage two candidate motors for conversion, one 5HP, the other 3HP.  The 5HP was more appealing for the potential greater power output as a generator, while the 3HP was seriously considered because it is obviously of higher quality manufacture.  Other things mattered, too, because the 5HP had only 2 poles, while the 3HP has 4 (hence smoother).  Eventually I decided on the 5HP, because I thought I could get more power out of it and there was just the right amount of room for the magnets I had.  Here is a shot of the dataplate.  It was made by USMotor.  A little anagram makes "Sumo", but it's hardly a heavyweight!

The original motor was hard-wired in Parallel-Star, and only 3 leads came to the connection box.  Pretty sure that I was going to need more than parallel-Y on this motor, I cut everything and extended them out to a new connection box with plenty of room.  I had to guess at the wire numbers.  I picked one, and found its corresponding other end, labelling them 1+4.  180 degrees around the stator was another pair of wires, so I tentatively named them 7+10.  Then I worked my way around to the next to find 2+5+8+11 and 3+6+9+12.  (I can't find any pictures of doing this, thought I had taken some).

I turned the rotor down in the usual way, but it was even easier because I did not need to bolt the laminations together to keep them from separating on the lathe.  There was so much aluminum cast on the rotor that I never did cut through it all.  Then I put it in the milling machine, to cut flat faces for the rectangular block magnets.  There are 8 faces in all, but they are not equally spaced 45 degrees apart.  Instead, I spaced the adjacent flats 39 degrees apart, then next flats 51 degrees apart, and repeated.  This made some magnets line up on the stator teeth, while others lined up between them.  The end result was a rotor that had just a bit of cogging.  Not enough, at least, to be a start-up problem.

Time to put the magnets on.  Starting out I had assumed that I could get 1"x1" square magnets for the rotor.  By this time I still hadn't found the right size of magnets with holes in the center to hold them down.  With a rotor 3" long, and only 2" long magnets available, I was faced with the dilemma of finding a way to fill the rotor.  Here's a picture of the rotor with only the 2" magnets on - and blank space to be filled.  

Well, I had 4 extra 2x1 mags, each with two holes, and I needed 8 1x1 magnets, with one hole each... I got out my hacksaw... Oh well it looks like hell but at least I didn't ruin them.

The first time I put the magnets on I did it wrong!  Somehow I put them on NN-SS-NN-SS, which would be right for a 4-pole rotor.  This 2-pole rotor needed them arranged NNNN-SSSS.  Fortunately nothing was glued down!  When I put the rotor in wrong like that, the shaft was impossible to turn.  As if the leads were shorted, there must have been short-circuit currents circulating between coils in the same pole.  It took me a few hours to remember the way I'd put the magnets on, and realize what was wrong.  This of course happens at 1:00 in the morning as I lie awake in bed, wondering.

Okay so it was finally properly assembled.  I discovered quickly that in Delta, the motor acted like it was shorted.  This was an early indication that the wiring was wrong, but after checking the continuity and relative resistances, it seemed okay.  I went ahead with tests in Star and in Jerry (rectifying each phase separately) and I got a batch of open-circuit voltage and charging current measurements.  Later checking the numbers I realized something was wrong, and eventually I realized that I had flipped wires 2 through 11 backwards, so I re-ran the tests in Star.  This mix-up did not affect the results of tests done in Jerry-connection.

I was rather surprised that the open-circuit voltage and current didn't actually change much, though at least in Delta it wasn't as stiff to turn any more.  What did change was the required input load, and it changed into values that made a lot more sense.  I will report the results of the pull tests in series-Y and Jerry.  I decided not to test in Delta because it would be only academic: the cut-in speed for Jerry is already too high, and it won't be any better for Delta.

The simplest test to do was measuring voltage with an open circuit.  All phases came out the same and in Star I had 14 volts per 100 RPM.  This is not what I hoped for!  My previous generator had exactly the same amount, and it was a stock 3HP motor.  It must be the total volume of magnet crammed inside, because everything else about this motor is different.

One mistake that I committed while doing my first runs of tests was setting the digital multimeter ammeter to record the "max" value of current as I did my pull test.  What actually happened was I measured brief spikes of current that didn't represent the average power being generated.  I had to throw away a bunch of work (again) to make measurements with the ammeter set to record the "average" reading.

I tested the motor for power input by wrapping a string around the shaft and pulling with a fish-scale.  By multiplying the shaft radius by the force on the scale, and by the RPM (and the conversion factor) I obtained the input power required to turn the rotor at that speed.  For each test I tried to keep the force and speed as constant as possible, which gave a fairly constant average charging current for each test run.  After pulling the string off the shaft, I wound it all back up, reset the meters, and pulled again.  Of course, one can pull with only so much force, so these tests were limited to only a couple of amps.  I have to extrapolate all of the rest.

For other tests, where I needed to have my hands free, I rigged up a pulley to the ceiling and the weight on the string pulled down with a constant force, pulling on the shaft with a constant force.  I have that setup illustrated on the picture above.  I put some notes on the picture to explain a bit better.

Here's the graph of the results (I can post the detailed numbers later) so that we can compare the Star and Jerry configurations.

Obviously, there's a big difference in cut-in RPM between Star and Jerry.

Overall I'm disapointed because I expected this 5HP motor would provide a lot more EMF, and cut-in sooner than it did.  I am still suspicious of a wiring mishap.  I should have been more careful to try to label the wires before cutting them out of the stator.  Rushing through that step has already cost me a couple of days' work.  

If there is an error and I can correct it I may get the 20+ volts per 100 RPM that I was looking for.  Then that would allow connecting in Jerry or Delta, or experimenting meaningfully with a Star-Delta switch or whatnot.  Otherwise I will be stuck with somewhat diminished performance compared to last year's generator, and I will have to manage the battery power more stringently.

There is still more to do!  I will take it all apart again, glue the magnets, paint the assembly, and then replace the bearings.  The bearings are badly worn and I wonder if this old motor had a rough life before I came along.

I have also finished carving the replacement blades.  Pictures of that to come after they're sanded, drilled, oiled and mounted.

Thanks for all the inspiration, folks!

[phil b]

Steven,

Thanks for the post. Your method of presentation is thorough. I have a 3,5 and a 20hp 3 phase motors stacked back for possible conversions. I can't wait to get started.

[SparWeb]

Going to start with the 3HP?  Are they good candidates?  (Low speed / big bearings / high voltage?)  20 HP is a bit big isn't it?  

I wrote in more detail about my previous motor conversions here

Thanks

[ghurd]

It looks good to me.

The numbers are series Y and series Jerry?

How did you hold the stator & rotor for the photo?

G-

[adaml]

Steven,

Great write up, look forward to seeing the rest.  Great inspiration as always.  Many thanks.

[oztules]

Hi Stephen, nice work as always. Just thought I would comment on this.... and maybe be corrected as well.

"The simplest test to do was measuring voltage with an open circuit.  All phases came out the same and in Star I had 14 volts per 100 RPM.  This is not what I hoped for!  My previous generator had exactly the same amount, and it was a stock 3HP motor.  It must be the total volume of magnet crammed inside, because everything else about this motor is different."

I suspect that the results are as they should be. If you think about it. These motors are really a transformer with a shorted secondary (rotor). The primary (stator) will be designed to just stay out of saturation. This will come down to so many turns per volt for the input voltage required. The thing that will change between the motors in the # of turns and the wire diameter (effective). The big core will be less turns per volt, as the square inches covered by the coils is greater... so less turns/volt, allowing for bigger wire. The smaller one will be converse..... but they are for the same voltage. So for a given flux I would expect similar emf generated. What will change is the current carrying ability of the windings... not the emf as they are both designed for the same line voltage.... the turns are different, the stator cross sections are different, but should both generate the same emf for those combinations for a given flux (magnet pack... which is the same I think from your story).

It may be worthwhile reading Gordons capacitor thread (backshed) and Phils axfx story. They are getting good results from inexpensive doublers on their machines. Phil has about half the turns he should, but gets very impressive results because of the cap doublers. (>2.5kw from sub 12" disks"

If I'm wrong about this explanation, we should soon know.

..........oztules

[Flux]

Perhaps if you think this way you may see that the results are fairly close to expectation.

At best if you can get enough magnet in you will just about saturate the core. Then you will get close to its original voltage at design speed.

Two pole machines are fast and hence are the smallest physical size for a given rating.

If you get nominal volts at 3600 rpm then you can expect about 10th of this at wind speed ( say 360 rpm) you will have 1/10 the original rating unless you overload it) and you have the resistance suited to the original voltage still hindering you at 1/10 voltage so not only have you drastically reduced output but you get it at rotten efficiency. If you can't match the windings it can be much worse.

You wouldn't expect as much of a 5hp 2 pole machine as a 4 pole one of 5hp if you run at reduced speed. In fact a 6 pole one would come out even better. Looking at it another way, the 2 pole machine will be small, 4 pole larger and 6 pole will be a heavy brute for the same rating.

You could argue that there is little space in a 2 pole machine for magnet compared with a 6 pole one and this is another way of seeing that the 6 pole machine will make a better conversion than the 2 pole. You can't hope for a tiny machine to compete with a big brute as the big one will have much more material in it for your same expected output.

Even rewinding to make a 2 pole 6 or 12 pole still leaves you with a small core and little space for magnets although you may get magnets a better fit than the 1/3 of a circle arc that you need for 2 pole. the 2 pole core also has long teeth ( deeps slots, call it what you will) and this adds leakage reactance as it increases flux leakage and this will limit your output if you try to push it beyond the limits set by resistance alone.

There is no escaping the fact that a low speed pma needs to be large diameter with lots of material in it. If you want to skimp on material for a given output you need a gearbox and that makes it a non starter.

Capacitors in parallel may increase excitation a bit but they will be monsters at low volts and low frequency. Not very practical unless you run high voltage for heating. I don't think series capacitors are going to help but you have done no tests beyond the resistive limited part of the curve, you show no reactance limiting because the resistance swamps it and you will probably burn it out before entering the reactance region. Capacitors are fine for high pole number very reactive F & P and similar motors but won't help you here. The voltage multiplier may help you but that is a matching issue and not related to the actual winding characteristics.

it is effectively a boost circuit and with your winding you may need a buck to match better.

Flux

[SparWeb]

Yes, series Y and Series J.

When I inserted the rotor, I clamped the motor to 4x4 posts and then clamped the 4x4 posts across the table.  I left it that way for the pull tests, so it never moved while pulling on the string.  There are about 7 pipe clamps and c-clamps involved!

Of course you can only pull so hard on the string.  And on the same note a human can only put so much power into it anyway, either by turning fast with limited force or gaining mechanical advantage but turning slower.  I found that I could only put about 200 watts into it, the force/speed tradeoffs always came out even.

I could have use a beefier cord, but that defeats the test.  Every time the string wraps over other wrappings of string, the radius of the shaft is changed.  When doing a test like this I want a constant force pulling on a constant radius.  So a rather small string was necessary.  1/8 braided nylon was just strong enough for me to put my weight against it without breaking.  I could wind up enough of it for a steady pull test, but didn't need to wrap on so much that it all bunched up on the shaft.

[SparWeb]

The windings are noticeable different in this one.  The wire must be 2 or 3 gauges heavier than the 3HP motor.  I'll have a chance to measure for sure when it's apart again to replace the bearings.

I've followed Gordon's thread as best I can, and in fact experimented with those capacitors myself.  I also saw the potential, and even tried to keep up mathematically.  I was not able to measure any effects even using capacitor banks of equal or larger mF than he is, when installed across phases of my windmill.  When Gordon finally got around to publishing a schematic of what he was doing, I realized I'd completely misunderstood.

I would have to go back to the little brick rectifiers to split everything apart.  Currently I rectify through a 3-phase bridge, and Gordon's scheme doesn't work with that.  I also think the frequency difference between the 14-pole F&P and a 2 or 4-pole conversion may taking my machines out of the running.

I'd love to try more experiments but I'm afraid it would take time that I don't have.  I need to get this machine in the air by mid-October!

[SparWeb]

Flux,

"...you need a gearbox and that makes it a non starter." was that a pun?

I've definitely noticed the different heft of these things.  As the number of poles goes up so does the weight.  As a 2-pole motor this one is even lighter than the 4-pole GE was, and had a lot less space for magnets on the rotor to boot.  Looking back at the GE I wonder if it was a design of frame that they could wind for different speeds as they saw fit.  That motor was old enough it might have pre-dated the modern NEMA dimension standards!

This is my third conversion, on a motor half way between the first and second's original power ratings, so here's a comparison:

Type       original   speed   mag volume         emf          R         Test results

GE             3HP    4-pole    12 cu.in.     16v/100RPM    5.8 ohm   24vDC   500W

US Motor    5HP    2-pole    12 cu.in.     14v/100RPM    3.6 ohm   24vDC   ?

Toshiba   7.5HP     4-pole    24 cu.in.     48v/100RPM    6.0 ohm   24vDC  1200W

Looked at this way, maybe what I can expect in the future is a 4-pole 5HP motor would have 28v/100 RPM instead.  That would make some sense, with having more turns of wire.

There's no telling how much reactance will limit the current output of this one at higher speeds (500+ RPM) but thank you for giving it some thought.  I'd like to do a test on a lathe but that takes preparation time that I don't have.

If I was interested in re-winding the motor maybe I could make a beter generator out of it.  I'm not terribly impressed with the housing (I'm worrying about how full of snow it will get).  Well, I guess I will have more luck converting that Baldor!  A winter-time project for me.  There is more room for magnets, too.