combined free speed of 3 motors

[James Kuszmaul]

Originally Posted by Ether

The derivation I provided in the original post applies if the same voltage is applied to all the motors, and the individual motor stall torque and free speed values are all at the same spec voltage. The resulting "combined" Free Speed (together with the simple-to-calculate Stall Current and Stall Torque) allow you to view the combination as equivalent to a single motor with those characteristics. Treating the combination as equivalent to a single motor allows you to use all the single-motor math you are already familiar with.

However, using all that single motor math will cause you to result in suboptimal results (if you attempt to optimize in any way, beyond just modelling the system). Also, has anyone gone through the proof that you can validly treat the two motors as one, given that you input the same voltage?

[Ether]

Originally Posted by James Kuszmaul

However, using all that single motor math will cause you to result in suboptimal results (if you attempt to optimize in any way, beyond just modelling the system).

I'm having trouble extracting an interpretation of the above. Could you explain what you mean? Perhaps give an example?

Also, has anyone gone through the proof that you can validly treat the two motors as one, given that you input the same voltage?

I thought that's what I did in the attachment to the original post. At what point in the derivation do you disagree?

To be clear, the derivation assumes the motors will be used to drive a load (like a drivetrain or heavy arm for example) with a reflected inertia much larger than the rotor inertia, so the dynamic contribution of the rotor inertia can be ignored.

[James Kuszmaul]

Originally Posted by Ether

I'm having trouble extracting an interpretation of the above. Could you explain what you mean? Perhaps give an example?

If you had two different motors, both attached to something (perhaps a shooter wheel), and you wanted to create an optimal controller for the wheel (given some cost function), then I doubt (although I may be wrong) that that optimal controller would have you always apply the same voltage to the motors. Hence, it is suboptimal to make this simplification, although there certainly may be situations where it is convenient, and therefore valuable, to make the simplification (such as if you were limited for space on your Digital Sidecar and wanted to be able to use a y-cable to control two different motors).

Originally Posted by Ether

I thought that's what I did in the attachment to the original post. At what point in the derivation do you disagree?

To be clear, the derivation assumes the motors will be used to drive a load (like a drivetrain or heavy arm for example) with a reflected inertia much larger than the rotor inertia, so the dynamic contribution of the rotor inertia can be ignored.

I see how you would be driving a large load (in my equations, the moment of inertia was primarily a place holder for such a load), but I do not see how a derivation of a formula for the free speed logically leads to the idea that all the single motor formulas must be valid.
I agree that you can calculate a stall torque, stall current, free speed, resistance, and free current for the motor combination, but just because those numbers may be meaningful does not necessitate that the single motor equations are true.
In other words, just because you have formulas for combined stall torque, stall current, free speed, resistance, and free current, why is it that these equations are still valid for values of V, I, omega, and Torque other than those at stall torque and free speed:
V = I*R + omega / Kv
Torque = Kt * I
I may be missing something, but I have not seen any proof that these equations continue to hold, and I like to see proofs .

[Ether]

Many teams that put 3 motors on a drivetrain gearbox want to know the acceleration characteristics when full voltage is applied to all 3 motors. Finding the single-motor "equivalent" motor curve permits the use of existing spreadsheets and models that don't support multiple mixed-motors on the gearbox.

I may be missing something, but I have not seen any proof that these equations continue to hold, and I like to see proofs

Go through the attachment to post#1 step-by-step and tell me at what step you disagree or are unconvinced, and I will explain the justification and make explicit any tacit assumptions.

[Paul Copioli]

Originally Posted by Ether

Many teams that put 3 motors on a drivetrain gearbox want to know the acceleration characteristics when full voltage is applied to all 3 motors. Finding the single-motor "equivalent" motor curve permits the use of existing spreadsheets and models that don't support multiple mixed-motors on the gearbox.



Go through the attachment to post#1 step-by-step and tell me at what step you disagree or are unconvinced, and I will explain the justification and make explicit any tacit assumptions.

Additionally, we have been using this model for several years to determine our robots' acceleration profile and time to distance at different spots of the field. The analysis has routinely been within 5% of actual robot performance.

[James Kuszmaul]

Originally Posted by Ether

[i]
Go through the attachment to post#1 step-by-step and tell me at what step you disagree or are unconvinced, and I will explain the justification and make explicit any tacit assumptions.

Perhaps I was unclear. I agree with your derivation of the free speed. I agree with the idea that there is a free speed as you describe and that it is of value.
What I am wondering about is your statement that

The resulting 'combined' Free Speed (together with the simple-to-calculate Stall Current and Stall Torque) allow you to view the combination as equivalent to a single motor with those characteristics. Treating the combination as equivalent to a single motor allows you to use all the single-motor math you are already familiar with.

(emhasis mine)
I agree that you can calculate numbers for free speed, stall current, and stall torque for the combined motors. It makes intuitive sense why you could use the single motor equations, given that you can find a Kv, Kt, and R, how do you prove (I don't necessarily disagree; I just don't agree) that
T = I * Kt
and
V = I * R + omega / Kv
[Edit: To be clear, how do you prove that these statements apply for the combined motors; I already know that they apply to a single motor]

I presume that I have been unclear in my questions, because I have been asking essentially the same question, phrased differently, the last couple posts.

[Ether]

Originally Posted by James Kuszmaul

I agree that you can calculate numbers for free speed, stall current, and stall torque for the combined motors. It makes intuitive sense why you could use the single motor equations... how do you prove...

Perhaps it will help to present the math in a slightly different way. Attached is a derivation for the torque vs speed curve for the 3-motor combination directly from the torque vs speed curves for the individual motors.

It's straightforward algebra. It follows naturally from the assumptions that 1) all three motor speeds are the same (since they are mechanically linked) and 2) the individual motor torques are additive.

As you can see, the combination behaves exactly like a single motor with torque = Tstall*(1-Speed/Sfree), where:

Tstall = T1s + T2s + T3s .... and .... Sfree = (T1s + T2s + T3s)/(T1s/S1f + T2s/S2f + T3s/S3f)

[James Kuszmaul]

Originally Posted by Ether

Perhaps it will help to present the math in a slightly different way. Attached is a derivation for the torque vs speed curve for the 3-motor combination directly from the torque vs speed curves for the individual motors.

It's straightforward algebra. It follows naturally from the assumptions that 1) all three motor speeds are the same (since they are mechanically linked) and 2) the individual motor torques are additive.

As you can see, the combination behaves exactly like a single motor with torque = Tstall*(1-Speed/Sfree), where:

Tstall = T1s + T2s + T3s .... and .... Sfree = (T1s + T2s + T3s)/(T1s/S1f + T2s/S2f + T3s/S3f)

This is definitely the right step getting to where I want to be, however, although I do see the linear relationship between net speed and torque, I do not see how that necessarily allows us to use the single motor equations (V=IR + v/Kv and I = KtI) for the combined motors. I have worked out some of the equations and attached the derivations. Hopefully you will be able to see where I am going and can help get it that last few steps.

[James Kuszmaul]

Originally Posted by Ether

My interest was in finding the equivalent torque vs speed curve for the motor combination.

If you want the equivalent R and Kt, they would be given by:

R = 1/(1/R1 + 1/R2 + 1/R3) ... and ... Kt = (T1s + T2s + T3s)/(I1s + I2s + I3s)

Yes; those are what those constants should be (and what the resistance is). But as for Kt, it makes sense for it to be that (I say as much in the attachment to my previous post). Prove it. I may be sounding a bit picky, but I really dislike taking anything or granted that I can't see the derivation of. And I can't see how you derived that Kt other than to say it is combined motors' stall torque divided by their combined stall current. But why is it valid to say that? Why do Torque and Current still relate linearly? Intuitively, it makes sense. However, I have not been able to figure it out with actual equations.

[Ether]

Worthy of note: If you combine motors with substantially mismatched free speeds, the free current for the combination can be large, and thus the value of Kt for the combination at high speeds will be less than the value calculated at stall.

I think this may be what James was trying to say.

edit: If you look at the last graph, you'll see that the 550 probably won't last long in this setup.

[James Kuszmaul]

Originally Posted by Ether

Worthy of note: If you combine motors with substantially mismatched free speeds, the free current for the combination can be large, and thus the value of Kt for the combination at high speeds will be less than the value calculated at stall.

I think this may be what James was trying to say.

That is what I was trying to figure out; I just wasn't able to figure out a way to tell for sure (ie, make the graph). Thanks.

[akeisic]

Creating a tool to help with this problem has been one of my back burner projects for a while now. I was never satisfied with just matching free speeds for different motors as there are many different ways to match motors, including matching free speed, stall torque, max power, max efficiency and matching slope. Depending on the application, you may want to choose a different matching condition.

As for the tool itself, I've worked through many of the thought processes mentioned in this thread and have a ~90% complete tool that combines up to 3 motors at any quantity using different operating voltages and fuses and matched by the conditions mentioned above. I then give the user the choice of plotting the torque or speed on the x-axis as well as giving the choice of plotting the super motor.

I've never gotten the workbook to a fully releasable version, which is why I haven't posted it yet, but it seems prudent to post it now. Plus, this conversation has me motivated to finish it through. I just uploaded the latest version to my motor performance white paper:
http://www.chiefdelphi.com/media/papers/2292

[Paul Copioli]

Ether,

Joe Johnson and I worked through this for two motors some time ago. Joe had a post on it back in either 2002 or 2003. His terminology and mine were different, but recently I made sure our two analysis methods came to the same conclusion. They did and yours matches our conclusions, but adds one more motor.

We did this analysis to try to kill the myth that you had to match free speeds. Joe went into detail about loading conditions that will cause one motor to actually be pushing the other, therefore the slower motor would contribute negative work to the system. Again, your analysis corroborates our conclusions back then.

I attached my hand calculations to prove Joe and my analysis methods were the same. I now use Joe's terminology because it is easier to put in Excel.

Paul

[Ether]

Originally Posted by Paul Copioli

Ether,

Joe Johnson and I worked through this for two motors some time ago.

Thanks Paul. I thought I remembered seeing something like that, but couldn't locate it.

yours matches our conclusions, but adds one more motor.

Notice the nice symmetry in the solution. It's easy to see how to extend this to N motors, instead of 3.

[Adam Freeman]

Originally Posted by Paul Copioli

Joe Johnson and I worked through this for two motors some time ago. Joe had a post on it back in either 2002 or 2003.

I bookmarked this post the last time this came up, just so I could reference it quickly anytime I needed to:

http://www.chiefdelphi.com/forums/sh...39&postcount=3