Dynamixel torque control using N.m

[Roque]

Hi,

I would like to control the torque of the motor but, instead of using current, using N.m. For this I need to know the current power drawn by the motor (I can get this by measuring torque goal and converting to Ampere and the current voltage, multiplying them) and the velocity (which I can get with the Present Velocity register).

Now the tricky part is giving N.m setpoints, and converting those to current values, as velocity seems somewhat to be a free variable here that I cannot control.

What would you suggest?

[jwatte]

Which Dynamixels?

You may be able to set them in wheel mode, and control current by updating the "max duty cycle" parameter?

It will only be an approximate control, though.

[Roque]

Which Dynamixels?

You may be able to set them in wheel mode, and control current by updating the "max duty cycle" parameter?

It will only be an approximate control, though.

I am using the MX64. I thought that the right way to do it was through torque goal registers, but I don't see a way to convert this to Newton meter.

[Roque]

I've searched a bit more and I think that the electric torque is directly proportional to the current on electric motors.

But how does this relate to Mechanical Torque? A torque setpoint should provide angular Accelaration, or not?

Best,
Pedro

[jwatte]

Angular acceleration is a function of torque AND SPEED and thus is "power." This is controlled by some mechanical engineering formulas, which in turn derive from Newton's equations for force, torque, and speed, as well as from the back EMF generated in the motor as it spins faster.

Torque can be applied against a stationary position, which happens all the time if you stall the motor, so acceleration is not necesarily correlated with torque.

The data sheet for the MX-64 has a current/torque/speed curve. http://support.robotis.com/en/produc...ries/mx-64.htm
If you read (or set) the current, and speed, you can use an approximation of that chart to approximate the torque in N.m.

[Roque]

Angular acceleration is a function of torque AND SPEED and thus is "power." This is controlled by some mechanical engineering formulas, which in turn derive from Newton's equations for force, torque, and speed, as well as from the back EMF generated in the motor as it spins faster.

Torque can be applied against a stationary position, which happens all the time if you stall the motor, so acceleration is not necesarily correlated with torque.

The data sheet for the MX-64 has a current/torque/speed curve. http://support.robotis.com/en/produc...ries/mx-64.htm
If you read (or set) the current, and speed, you can use an approximation of that chart to approximate the torque in N.m.

Thank you for the fast response!

All right, but then, speed cannot be set, just read, is this right? So essentially, in order to control Torque in N.m one must use that conversion chart?

On the load subject, I tested the motor without any load and realized that any values greater thank 100*4.5mA provided no change on the motor speed. Is this because there was no load? With load will this change?

Best regards,

[jwatte]

There is a maximum speed for the motor.
When there is enough power to achieve that speed, the motor will run no faster.
As you increase load, current consumption increases to maintain speed, but also, losses increase and top speed will go down.
You can see exactly that relation in the diagram in the manual.
If you look at the registers for the MX-64, it can be in "wheel mode" where you can set target speed, or it can be in "position mode" where you can set target position.
You can also control maximum PWM duty cycle applied, which controls maximum current supplied, but it doesn't "force" current into the windings.

The more expensive Dynamixel Pro servos have a "target torque" register that I presume let you control desired torque output directly.