
09062013 #1
Hi all,
As my workshop and office are all in storage, not that i could get to them in my wheelchair anyway, i need someone to carry out a little experiment for me. i'll explain why in a later post if the results are useful....
kit needed:
a stepper motor, any sort but lets say a nema23 bipolar for now
some way to turn the shaft at a regular but lowish speed up to 100rpm
a dc power supply with adjustable current limit up to the stepper current rating
an oscilloscope
Basically i want to spin the motor at various speeds while applying a controlled DC current to one phase winding and seeing what gets generated at the other phase.
My expectation is that the motor will get harder to turn as the current is increased and the other winding will generate a sinusoidal voltage, the amplitude at a given speed will have some relationship to the effort needed....

09062013 #2

09062013 #3
Thanks Jonathan

10062013 #4
The last comment in your first post implies the plan might be to try and use the stepper motor as a means to measure torque?
Stepper motors that I can easily test (i.e. they're not attached to a machine) are:
3.0Nm Nema 24, 4.2A/phase (parallel), one from Zapp
1.0Nm Nema 23, 1.4A/phase (parallel), Astrosyn MY103H702.. Quite high (10mH) phase inductance which might actually be useful here.
12.2Nm Nema 34, 6.2A/phase, SY85STH1566204B
Plan:
 Hold stepper motor shaft in lathe chuck, clamp motor to saddle.
 Lathe induction motor controlled via VFD/gearbox, so can easily set various speeds.
 Got an upto 50V, 20A lab PSU which is plenty...
 Can get (more) exact motor speed from oscilloscope.
Which motor (or motors) would you prefer me to test?
Do you just want the open circuit voltage waveform of the nonpowered coil, or the voltage under load (i.e. with a resistor attached)?

10062013 #5
Try the Astrosyn MY103 to start as I have a few of those. Measure the voltage across a light load, say 1k.
The main goal is to see if putting a DC current through the winding is an effective way of using a stepper as a variable brake. And whether induced voltage in other winding tells anything about the load its putting on the driving force.
I need to generate a 'load' of between 5 and 100W at 60rpm approx (6rad/sec), equal to a torque of between 1 and 17Nm (obviously a 1Nm stepper wont do this so final design may involve gearing up with timing belts)

10062013 #6
Using the 1Nm motor, 700mA, 42rpm (calculated from graph, seems reasonable as lathe was set to 10Hz using 205rpm ratio so 10/50*205slip etc)
Many more graphs attached:
MY103Test.zip
That's with 700mA through one phase and a 1K resistor in parallel with the other phase. I also added a 1nF capacitor in parallel with the resistor to filter out the many spikes on the waveform, which the oscilloscope reckoned were at about 20kHz. I think this motor may be damaged though, as to get 700mA through it in bipolar parallel required approximately twice the voltage on the phase I used, compared to the other phase, which implies one coil is open circuit. I should try again with a more reliable motor! This does mean that the test at 1.4A was exceeding the rating somewhat, so I didn't do it for long... may cause a problem due to saturation.
I'm not sure that we can glean much from these results, as I'm not measuring the torque, so we there's not enough information to plot power vs phase current, or P vs V. I could measure the torque using a longish bar and a force meter (= hanging scales)...awkward to calibrate though, but still fine for comparing readings. Alternatively could assume that (pretending for the moment I'm not using a dodgy motor) at rated current the mean torque will be rated divided by squrt(2) and assume torque (and therefore power) is proportional to current but that method clearly has problems.
Thinking about it I should have measured the input voltage to the energized phase then at least we'd have input power...just done that and at 42rpm and 84rpm input power is 14.8W with 1.4A, similarly 3.7W for both speeds at 0.7A. Still 3.7W at a substantially higher speed, so looks like input power isn't a function of speed. At higher speed the waveform peak splits into two, and if you look at the FFT the harmonics are substantially greater. If the speed is greater still (tried 1kHz, so 1000/50*60=1200rpm) the output is a very much purer looking sinosoid, power is still 3.7W ... might be going a bit off the topic here so I'll stop.Last edited by Jonathan; 10062013 at 11:38 PM.

11062013 #7
Jonathan
Thanks for doing that. not sure i got the answer i was hoping for :(
Basically I was hoping that applying dc to one coil of a stepper it would act as a brake. the lathe clearly has much more grunt so no braking effect was seen (frequency is near identical) and the output voltage from the second winding is essentially the same so isnt a measure of braking force.
i wonder if turning it by hand would empirically show the braking effect and maybe measure the effect using pulley and falling weight approach

11062013 #8
The stepper motor is definitely applying a torque opposing the lathe and if I just hold the motor it feels like that torque is a function of the phase current  presumably they're proportional. The torque ripple is high though since we're operating it as a single phase motor.
I think measuring the torque using the method I suggested earlier would be easier?

11062013 #9
Just spotted something interesting...
I measured the backemf voltage from the 3.1Nm stepper motor at 1200rpm and it's 66 volts, so using those numbers k=0.525 Vs/rad. The unit Vs/rad is dimensionally equivalent to Nm/A ... i.e the 'torque constant' of the motor, so we can use this to find the torque for a given phase current. Rated phase current is 4.2A, so for both phases T=2^0.5*4.2*0.525=3.12Nm. That matches the holding torque specification rather nicely, so maybe this could be a simple way to find the rated torque of stepper motors. Or more usefully, use the formula for phase current (something like i(t)=V/R(V/RI)*e^(t/(L/R))) multiply by k to get torque as a function of time then integrate to get the mean torque and then the torque vs speed curve for the motor. I think this should be more accurate than the usual approximation.Last edited by Jonathan; 11062013 at 11:25 AM.

11062013 #10
Bookmarks