I disassembled enough of the John Deere motor to see that internally it looks very similar to the 775 Nichibo for which I have electrical data. The 775 portion of the motor's part number indicates a standard physical size, but from what I've noticed during my online research the electrical specs can vary somewhat from one 775 motor to another. Since my unloaded rpm and current measurements of the two motors were nearly identical, I'm assuming the motors are the same except for some cosmetic differences. The John Deere version has a plastic boot wrapped around its lower end to block off air vents that would otherwise allow grass clippings inside the motor while the Nichibo has a stator flux shield wrapped around the outside of its housing.
After taking some time to think about the disappointing and puzzling tests that I ran a few days ago, I believe I now have an understanding of what happened.
The speed of a dc motor is directly proportional to the voltage applied to its terminals, and its constant of proportionality is Kv in rpm/volt. When a fixed voltage is applied to an unloaded motor, it will run at its unloaded speed and draw its unloaded current. When mechanical load is applied to the motor, its speed will decrease and its current will increase. Both of these changes are linearly related to the size of the load, and the slopes of their straight-line relationships are available from the motor's data. They can also be measured, but a calibrated mechanical load isn't a typical fixture in a home shop.
For the Nichibo motor, the speed decreases 71 krpm per ft-lb., and the current increases 268 amps per ft-lb. The maximum torque that the motor can supply is called its stall torque, and for the Nichibo this is .32 ft-lbs. The current at this torque is called the motor's stall current, and for the Nichibo this is about 88 amps. In general, as the Nichibo curves show, operating a motor near its maximum torque can be very inefficient, and the resulting internal dissipation will shorten the life of the motor. In briefly powered starter applications this may not be a major concern, but the motor's wiring and controls must be able to handle the additional wasted current.
As the load on the motor increases, its speed will drop, and its current will increase as predicted by the two linear constants. These constants are fixed by the physical design of the motor and are independent of the applied voltage. If the motor is spinning faster than desired under a given load, though, the applied voltage can be decreased to drop the speed of the motor. Likewise, if the motor is spinning slower than desired under a given load, the applied voltage can be increased but only up to the maximum working value specified for the motor.
An efficient way to manually vary a battery voltage applied to a motor is with the use of a PWM controller. This device will efficiently switch the voltage applied to the motor on and off at a high rate and at a duty cycle that can be manually controlled. The effective voltage applied to the motor will be the average value of the PWM waveform. The frequency of the switching should be much greater than the motor's ability to respond to speed changes. My tests on this particular motor showed that 240 Hz was much too low, 2.2 khz was OK, but 22 khz was much better.
The crude load test that I ran while gripping the prop shaft with my hand got the motor to draw 13 (average) amps while spinning the crankshaft about 1500 rpm. The measured average voltage applied to the motor was 2.8 volts. Therefore the duty cycle of the PWM worked out to be 2.8v/12v =23%, and all seemed well. I didn't realize at the time there was trouble brewing under the covers. The peak current supplied to the motor by the controller was 13a/.23 = 56 amps, and this was higher than the controller's continuous 40 amp rating which I mistakenly assumed was spec'd in terms of average current and not peak current.
The third test was performed with some of the spark plugs installed in order to approach the load that the starter will need to handle. Everything went wrong during this test. The motor wouldn't spin, the pot had little control over the controller's low output voltage, and the average measured current was off-scale at 30 amps.
After mulling over the tests and discovering my error about the misinterpreted current rating, I realized the controller was probably at least part of the problem. I replaced the 12 volt battery I was using with a pair of series-connected batteries which I hoped would allow me to source a little more power to the motor through the controller. This change got the motor just barely spinning with three spark plugs installed, but there were still significant hesitations at each of the three compression humps. The starter sounded very much like an old big block Chevy trying to start after its starter and solenoid had been over-heated by the engine's headers.
This test seemed to confirm that the controller was probably most of my problem. After installing all the spark plugs, I threw caution to wind and removed the controller completely so I could briefly energize the motor directly from the battery. The motor abruptly woke up and effortlessly spun the crankshaft. Over a number of test cycles I was able to measure the crankshaft spinning at 620 rpm while drawing 77 amps from the battery with 10 volts across the motor's terminals.
Although a lawn mower engine starter would likely have no issue with running at such an operating point, I'm uneasy about stressing the Quarter Scale's starting system with some 2-3 times more power than required because the motor is spinning faster than needed. If it weren't for my concern about the robustness of the starting system I would probably just add a starter relay to what I already have and be done with it. I have to admit that that its current high-pitch whine sounds pretty good.
Spinning the engine with the starter eventually became infectious, and I spent a lot more time playing with it than I probably should have. I was careful to keep the bearings and cylinder walls doused with oil since I don't yet have the oil pumps connected. As an aside, I placed a finger over the carburetor intake during one of the cranking tests. The suction was very strong and a good indication that the induction system is probably working as it should.
Even though the starter system has already survived several minutes of cranking time, I'd still like the ability to control its speed, and a PWM controller is still the most efficient way of doing this. This time I went shopping for one in the RC car/truck world where 70 amp brush motors need to be controlled. In this specialized marketplace the controllers are called ESC's (electronic speed controllers), and they're much smaller and more capable than the general purpose controller I've been using. Once more, eBay came to the rescue:
http://www.ebay.com/itm/302145096583
I don't know what extraterrestrial technology the manufacturer of this PWM is using to be able claim 320 amps continuous operation within such a small package. Even though this time I realize they're talking about peak current, my expectations aren't high.
Seemingly, the main issue with adapting an ESC to a non-RC application is that these controllers have an input control port of their own that expects a small signal PWM control waveform from a servo. Fortunately, this input can be generated and manualły controlled by yet another piece of ($6) specialized RC hardware called a servo tester. Alternatively, some simple circuitry based upon a 555 timer can be breadboarded to do the same thing. My concern is that the ESC/servo tester combination that I've ordered may include some extra built-in intelligence to support bi-directional and dynamic breaking operation that could get in the way of my plans to use it as a simple single direction speed controller. - Terry
