Dear members,

For tension contol of a the wire spooled on a spooling drum we want to measure the torque applied to the drum by a servo motor.

Does anyone have experience or suggestion how to predict the torque of a servo motor by measuring the input current to the servo motor.

Thanks,
Daniel de Jager

 

What are you going to do with the torque value?
If you just want to get an approximation, you shoule be able to construct a two point characterization of the servomotor, at varying levels of the control voltage. For a given control voltage magnitude, obtain the locked rotor torque, and then the no load motor speed. This is sufficient to draw one line. Pick another control signal magnitude, repeat measurements, and draw another line. ... and so on. Note that you will have to know an instantaneous motor speed, and also the control signal magnitude, in order to determine a corresponding motor torque.

This technique assumes that a control voltage governs the behavior of the servomotor. If you are using a control signal current, you may have to modify the situation, but it should still work. Is this relevant to your case?

[email protected]

 

If you know the current and the coil resistance you can try equating Torque and Power to solve for the Torque. There is really more to it but it might give you your idea.

 

Basically, you just need to multiply the current value by the torque constant of the motor, which tells you the amount of torque produced per unit of current. This is a standard data-sheet item for a servomotor. If it's an AC motor, you need to be careful about DC vs. AC(rms) current values.

Curt Wilson
Delta Tau Data Systems

 

Answer,
There is alot more to it than that! A servo motor tries to stand still as hard as it tries to run a load so the input current will not give you the torque output of the motor. Generally voltage gives you torque and current gives you speed in a DC motor. Servo motors have pulse width modulated square wave DC applied at a couple thousand hertz.
The current feeding a servo motor is about the same running and standing still.

 

Mr. Hinton:

Sorry, but you are just plain 180-degrees wrong on this one. The basic motor equation is T=Kt*I, where T is torque, Kt is the motor torque constant, and I is the current. Look at a servomotor catalog. For every motor, you will see a torque constant listed, with units of newton-meters per ampere, or pound-inches per ampere. You take the current in amperes, multiply this by the torque constant, and you get torque in Nm or lb-in.

Now look in the catalog for the "voltage constant" or "back EMF constant" Ke, with units such as volts/krpm. This is for the basic generator equation E=Ke*w, where E is the generated back EMF voltage, Ke is the voltage constant, and w is the motor speed.

So when you say, "Generally voltage gives you torque and current gives you speed in a DC motor", I'm afraid you've got it exactly backwards.

It's pretty much that simple. There are only a few caveats. First, if you put in so much current that you drive the motor's iron into magnetic saturation, you lose the strict proportionality. More importantly, it assumes that the rotor and stator magnetic fields are 90 degrees apart. The mechanical commutator in a brush motor maintains this angle; the electronic commutation scheme must do this for a brushless motor.

To be more general, the above torque equation should be T=Kt*I*sin(A), where A is the angle between rotor and stator magnetic fields. In a properly operating servomotor control scheme, A is held at 90 degrees, and the current I is varied to meet the torque demands. If no torque is demanded, no current is applied.

However, in a stepper motor control scheme, I is held constant, and the torque varies with A. At no-load equilibrium, A is 0, so no torque is produced, even though current is being applied. So in this scheme, you are burning current at standstill. Perhaps this is what you were thinking of.

Curt Wilson
Delta Tau Data Systems

 

Look on the motor data sheets and you will find a "torque constant". Its units are lb-in/amp. Simply multiply the measured current by this constant and you will have the torque. Don't forget that there are 3 components to torque.

These are friction, work, and acceleration torques. The tension in the wire will be the measured torque less the friction and acceleration torques.

Thomas B. Bullock, President
Bull's Eye Marketing Incorporated
Industrial Controls Consulting Division
104 S. Main Street, Suite 320
Fond du Lac, WI 54935
PH: 920: 929-6544
FX: 920: 929-9344
E-mail: [email protected]

 

Daniel
With brushless DC drives the torque is proportional to the current and the torque constant of the motor. Most commercial servo amplifiers provide a dc analog output proportional to current. Thus so many output volts are equal to so many amps. Multiply this measured current from the amplifier output by the motor torque constant and you have the torque. Do not try to measure the motor current since you are looking at a sinusoidal varying square wave in
each phase of the motor.

George Younkin

 

Modern motion control drives provide access to the current Id parameter for the motor in use. Your algorithm could monitor this parameter and provide an additional input into your control loop.

Jeff Gazidis
Senergis Technologies Inc.

 

The motor current has a nearly linear relation with the torque on shaft. Normally the motor specs tell you the factor. But you better use an
intelligent drive which gives you the actual current measured internally (like B&R Acopos servo drive). Be careful of the fluctuation of the current and also the efficiency of the power train - there is a lose from motor shaft to web.

Mark

 

You seem to have it backwards and your understanding of pwm/servo motors is way off. For example, with proper commutation (field oriented vector) control of a 3 phase brushless servo, torque is proportional to current and speed to voltage. Are you thinking of ac motor off the ac line? Totally different than brushless servo motor applications.

The problem with predicting torque from measuring servo motor phase current is that the phase current waveform is complex/instantaneous due to
electronic commutation scheme. You could use an isolated current probe/scope with RMS math function to crunch the RMS value for a given
section. Another problem is that the torque constant Kt usually saturates with higher current and magnet temperatures. Since torque = Kt x Irms, the prediction would be off. Sometimes Kt is units of torque per amp peak (of the sinewave). Then you crunch the RMS and multiply by 1.41 to get the proper units.

This winding application should be done the industry proven way with either a "dancer" or load cell etc.

 

Tom,

Lots of winding apps have been done using only torque er... current feedback for tension control, at least in the world that I work in.

Another option nobody has mentioned (I thought I did, but I don't see my response posted)... is that most current loop amplifiers are controlled with motion controllers. Most motion controller provide the ability to limit the DAC output with some clamp value. You can effectively control current / torque / tension if you know the drivetrain details, Kt, roll radius and actively set the DAC clamp while keeping the velocity loop in saturation. This works well enough to wind tissue paper at high speeds in torque mode without a dancer or force transducer. This control methodology works well in a broad range of winding applications.

Ken Brown
Applied Motion Systems, Inc.
www.kinemation.com