Method to compute relative motor torque

[mmagliaro]

At long last, here is a torque vs load graph.
Remember now that I am using a Maxon motor for which I have actual torque/A spec (13 mNm per amp), so these torque values are not "relative" numbers.  These are the actual measured torque values of these motors.  Look at the graph, and I will try to clarify here:

From left to right, on each plotted line, you are looking at how the torque (x axis) and RPM (y axis) value changes as the load on the motor INCREASES.  The 4 points on each line were at load resistors on the generator of 1220, 470, 180, and 100 ohms, respectively.   Each COLOR represents one fixed voltage.  So the 4 purple lines are the 4 different motors all running at 4 volts as I increase the load on them.  The red lines are all 4 motors at 2.5 volts.  The green lines are the 4 motors at 1.5 volts.  Some motors could not run at 1.5 volts, so there will not be 4 lines in that case.  Also notice that some motors could not keep running as I increase the load over the 4 settings, so some lines may only have 2 or 3 points.
For example, if a motor could run at the 1220, 470, and 180 ohm load, but stalled at 100 ohm, then there would only be 3 points on that line.

What does the graph show?
Well, it's pretty obvious that both armatures with neodymium magnets outperform the stock Rivarossi by a lot.  Not only are they able to run slower, but as the load increases, the Rivarossi's RPM nose-dives compared to the others, and its torque values are nowhere near as high.

The other interesting thing is looking at my hand-rewound armature vs the stock "type 1" armature.  As you can see, the hand rewound runs a lot slower at each voltage.  And if you look at the X-axis values of the rewound vs type 1, it does put out similar torque at LOWER RPM, which is what I had hoped.  But once we get up to about 4 volts, it no longer does this.  Or example, the Type 1, 4th point, 100 ohm load (purple line with diamond markers), at 4 volts, has an RPM of about 2400, and torque= about .29   The rewound armature at 4 volts turns at 1800 RPM, torque=.21

One thing not shown on the graph is how much torque the rewound motor makes with the 100 ohm load when it is turning as fast as the type 1 armature?   But that should be obvious.  If it can turn the same load at the same speed, it makes the same torque at 2400, and in fact, my measurements showed that it does.   

So in keeping with metalworkertom's suspicions, the trade-off of more turns of higher resistance wire for lower current, does appear to end up giving you the same torque after all.  I guess the only way to make the armature STRONGER would be to put more turns of the SAME gauge wire on there, not thinner wire.  And that isn't physically possible on this motor.

The rewound armature did appear to be able to physically turn at lower speeds.  It is able to get down to 355 RPM vs the Type 1 which can only make it to 420.  But I'm not even sure that is conclusive when we are looking at only one sample of each motor.  I would say that subject to the variability of these motors and my ability to hand-wind an armature, that isn't any difference at all.

[nickelplate759]

Science!

[peteski]

Science!

[peteski]

That was an interesting experiment Max.  The results are what I expected.  The torque is produced from the interaction between the magnetic field of the permanent magnet and magnetic field generated in the windings.  So stronger permanent magnet will produce more torque than a weak magnet (in the same motor).

The magnetic field produced in the windings is related to the power consumed by the coils.  Looking at the coils strictly as resistive load, since you use the same reference voltage for original and home-brewed windings, at the same voltage the higher-resistance (home-brewed) winding will pass less current. Since power is a product of current and voltage, at identical voltage, the home-brewed armature has a weaker magnetic field than the original armature.

By rewinding the armature you basically created a motor which can operate at higher voltage.  So to produce the same RPMs and torque as the motor with the original armature, it will require higher voltage.  It all adds up.

[mmagliaro]

Thanks, Petesk.

Now, I have some other observations.

When these motors are under a heavier load, they can actually run down to a SLOWER minimum speed than they can at light load.  That really surprised me.  My supposition about what's going on is this: Under heavy load, the motor can't move.  And as you turn up the voltage, it draws more and more current, until finally it CAN move.  But the load is heavy enough that it can't move very fast.
Recall that I did most of my tests down to a 100 ohm load on the generator.
When I tried a 5 ohm load, I found that all the motors, Rivarossi included, could manage 988 RPM, which equates to 10 scale mph in a Rivarossi 4-6-2.  All of them EXCEPT the Rivarossi could handle a 5 ohm load and run at 494 RPM, which equates to 5 mph in the 4-6-2.  But this is far more load than these motors experience in a 4-6-2.  I know this because when I run that same engine on a test track, these motors all draw 30-60 mA, even at 8 volts (the RR draws about 125).  But when I put them on the generator with a 5 ohm load at 8 volts, they draw 130 - 150 (the RR draws over 200 and starts getting HOT).

Thing 2:
I tested one other variation of motor that I haven't mentioned all this time.  Take an existing armature, that has no epoxy blobs in the windings, and happens to have some room to spare on the armature, and simply snip the wires and solder-splice on with magnet wire and add more windings to what's there.  The second type of armature I got from eBay was exectly like this.   I added 100 turns to each coil.    That bugger can run the 5 ohm load on the generator at 350 RPM.  None of the other variations could do that.
I am in the process of doing adding windings to another armature this way, to see if the results are repeatable.

Thing 3:
Since the Maxon has a published torque/current constant, it is now possible to actually measure the torque of any motor.  Think of the possibilities!  Atlas "regular speed" vs "scale speed" at various loads and RPMS to see if the scale speed is really any stronger, weaker, the same, just "slower", etc.

[peteski]

Your testing setup is very impressive Max, and I'm sure it will come in very handy for evaluating all sorts of motors (Atlas, Kato, etc.).
What you should also consider including in your graph is the power consumed by the motors (voltage multiplied by current).  I expect that you will find in order to obtain certain rpms under certain torque, similar amount of power will be consumed by every motor you tested.

The only thing I'm not sure about is the role of the BEMF generated by the running motor.  The BEMF is at its maximum (in relation to the voltage running the motor) when the motor is not under load (free running).  Since BEMF is of opposite polarity, it reduces the current from the power source running the motor. That is why a free running motor consumes very little current. Much less than if you were to calculate the current simply by dividing the voltage supplied to the motor by the winding's resistance.

 But when the motor is under load, the rpm's are reduced, and so is the BEMF. That causes the current running through the motor to increase. That probably generates stronger magnetic field, and more torque.  But unless the motor is stalled (no BEMF generated) then the motor likely still consumes less current than what you would expect when going strictly by the winding's resistance.

[mmagliaro]

Your testing setup is very impressive Max, and I'm sure it will come in very handy for evaluating all sorts of motors (Atlas, Kato, etc.).
What you should also consider including in your graph is the power consumed by the motors (voltage multiplied by current).  I expect that you will find in order to obtain certain rpms under certain torque, similar amount of power will be consumed by every motor you tested.

The only thing I'm not sure about is the role of the BEMF generated by the running motor.  The BEMF is at its maximum (in relation to the voltage running the motor) when the motor is not under load (free running).  Since BEMF is of opposite polarity, it reduces the current from the power source running the motor. That is why a free running motor consumes very little current. Much less than if you were to calculate the current simply by dividing the voltage supplied to the motor by the winding's resistance.

 But when the motor is under load, the rpm's are reduced, and so is the BEMF. That causes the current running through the motor to increase. That probably generates stronger magnetic field, and more torque.  But unless the motor is stalled (no BEMF generated) then the motor likely still consumes less current than what you would expect when going strictly by the winding's resistance.

I did collect the input voltage and current to all the motors under all these tests.  I just didn't post it.
In general, all my rebuilt variations can handle the 100 ohm load from 0-12v without getting too warm.    The Rivarossi, however, gets monstrously hot under that load by about 8 volts.
This also makes sense.  The armature and the magnet in the rebuilt motors provide a much stronger, and therefore
more efficient, field.  What my torque graph did not show was that even at the points where the Rivarossi motor
can compete with the others for torque, it is using a lot more current and making a lot more heat to do it.

One big thing MISSING from all this testing is the motor behavior with simple sine wave pulses, which greatly enhance the performance.  I can run the RR 4-6-2 under 2 mph with the motor where I added the 100 turns to the armature.  That's 197 RPM. and it can't come anywhere near that on smooth DC.

[peteski]

One big thing MISSING from all this testing is the motor behavior with simple sine wave pulses, which greatly enhance the performance.  I can run the RR 4-6-2 under 2 mph with the motor where I added the 100 turns to the armature.  That's 197 RPM. and it can't come anywhere near that on smooth DC.

Yes, I would expect to see that. We know that any sort of pulse improves low speed performance of these motors.   But I also suspect that if you used pulses for higher speeds, the motor torque would be reduced (since pulses provide less average power to the  motor).

[mmagliaro]

Yes, I would expect to see that. We know that any sort of pulse improves low speed performance of these motors.   But I also suspect that if you used pulses for higher speeds, the motor torque would be reduced (since pulses provide less average power to the  motor).

Ugh... you're going to make me test this, aren't you?   
I'm not so sure about this "less power" thing.  The average voltage of the sine wave (for sine pulses) would determine the speed.  And the speed should be the same as if I put smooth DC into the motor at that average voltage level, no?
So for a given speed, the average power into the motor would be the same.  The only difference is that the pulse power isn't steady, which is what helps the armature keep from getting stuck.

[peteski]

Ugh... you're going to make me test this, aren't you?   
I'm not so sure about this "less power" thing.  The average voltage of the sine wave (for sine pulses) would determine the speed.  And the speed should be the same as if I put smooth DC into the motor at that average voltage level, no?
So for a given speed, the average power into the motor would be the same.  The only difference is that the pulse power isn't steady, which is what helps the armature keep from getting stuck.

I guess that depends on the type of pulses.  First of all, you can't feed a sine wave voltage to a motor (as sine wave is AC voltage). It needs to be rectified. Half-wave will have about 50% duty cycle, and full-wave should be closer do the equivalent DC voltage.  But half-wave is the type of "pulse" voltage that is usually used to improve its slow speed performance.

[mmagliaro]

I guess that depends on the type of pulses.  First of all, you can't feed a sine wave voltage to a motor (as sine wave is AC voltage). It needs to be rectified. Half-wave will have about 50% duty cycle, and full-wave should be closer do the equivalent DC voltage.  But half-wave is the type of "pulse" voltage that is usually used to improve its slow speed performance.

Well, my throttle pulses are a little "fancier" than that.  They are half-wave pulses riding on top of the smooth DC.
So at, say, 6v smooth DC, there are little 2 volt "blips" riding on top of that.  Unlike just using a plain half-wave,
you don't have the output hitting zero between the pulses.  So it's not the same as if you had 8v smooth DC.  But it's
not as low as half-wave rectified sine, either.  I think this is the best way to power a motor if you're going to just
use old-school sine pulses (instead of a PWM throttle).  The pulses keep "kicking" the motor along so it can run at lower speeds, and it never gets hot.

[peteski]

In that case, I don't think there is any need to re-run the test using *YOUR* pulse-throttle.  The mixed type of voltage it supplies to the motor is just too unconventional to be useful to most modelers.

[mmagliaro]

In that case, I don't think there is any need to re-run the test using *YOUR* pulse-throttle.  The mixed type of voltage it supplies to the motor is just too unconventional to be useful to most modelers.

While I agree that this type of pulse is not like the PWM from a DCC decoder (or from a PWM DC throttle), it's not THAT unconventional.  The MRC Tech 4 and Tech 7 both use this approach: a base DC component with pulses added on top.  And I know some of the earlier MRC packs used a variant of this same idea, but they would put out pulses + smooth DC and reduced the amplitude of the pulses as the throttle was turned up.  I think the Tech II 2500 did it that way.
I'd bet that PWM would beat all others.  If I ever get around to throwing a PWM throttle together, I'll try it.

[peteski]

If I ever get around to throwing a PWM throttle together, I'll try it.

That might be interesting since the PMW pulses are at full voltage (even though the average voltage "seen" by the motor is variable.  I wonder how that affects (improves) the low speed torque.

[mmagliaro]

That might be interesting since the PMW pulses are at full voltage (even though the average voltage "seen" by the motor is variable.  I wonder how that affects (improves) the low speed torque.

I don't think it does.   Here's one article:
http://fab.cba.mit.edu/classes/865.18/power/PWM_motor_control_rev5.pdf

Torque and power are not improved using PWM.  In fact, there are always losses from all the switching in the power supply to the motor.  What PWM gets you is compactness, good speed control, and the ability to make the motor run slow because it is "pulse power".  By the way, I did use a very tightly regulated supply when testing all these motors - it doesn't sag more than a couple hundredths of a volt at worst from no load up to 1/2 amp load - so that was not a factor in my data.

My interest in torque began by wondering if we could make a motor move the engine at lower speeds by improving the torque.  I think it helps, but kicking it along with pulses just to keep it moving matters more.  At least, that's my sense of it after doing all this work.