3-ph PM motor spinning all by itself?

[arvidb]

I've just tried my servos in torque control mode, and I'm wondering a bit about their behaviour.

This is a 200 W 3-ph PM servo. I tried it without load.

If I set the torque reference to 0 (or very close), I can of course turn the motor by hand. It has considerable cogging. (Is this normal? This is the only servo motor model I've played with.)

But if I give the rotor a spin, it then continues to spin at about 1000 RPM all by itself! This it does no matter what direction I spin it. It is very easily stopped with just a finger though. I guess this has something to do with the cogging, but am not sure. Anybody experienced this before?

Arvid

[Al_The_Man]

If they are 3ph brushless (not true AC sinusoidal) , cogging at low rpm is normal. The normal way to control a servo in the torque mode is the motion controller has to have control via feedback generaly in the form of an encoder to close the torque loop, otherwise runaway condition is likely. That is why it is usual to only enable the drives after the motion control has come up and running and is not in a fault condition, otherwise the same thing can happen.
Al

[arvidb]

I believe they are AC sinusoidal. There is no feedback so I would expect the torque reference to function something like applying a DC voltage to a DC servo? Or is it more like feeding it with a constant current supply?

With cogging I mean that the shaft is alternatingly difficult to turn or turns (almost) by itself a few times during one revolution. Like the steps of a stepper but far fewer. (I can understand why this happens but I'm not sure how strong it should be or if the term "cogging" is correct.)

What I find strange is that the motor is at a standstill, but then takes off at 1000 RPM just because I give it a spin by hand? And in any direction (without changing the torque reference?)

Arvid

[Al_The_Man]

Wether BLDC or AC sinusoidal there has to be some feedback at least for commutation, both require the amplifier drive unit to know the position of the armature. BLDC use hall effect or equiv. and can be located on the encoder disc if one is present, AC sinusoidal often have a resolver, this gives 2 phases back to the controller which produces the third artificially, some AC have an encoder disk, which is required for initial position on power on, after that it keeps track of rotor angle. But this only provides the commutation angle for the motor, after this the motor should be under true torque command from the position (motion) controller which feeds the correct digital or analogue signal into the amplifier with feedback from an encoder or other feedback device.
What motors are they?
Al

[arvidb]

The motor is a Yaskawa SGM-02A3. They use encoder feedback only, with the third (index) track having some smart design enabling the driver to find exact rotor position in just a fraction of a turn. The driver BTW is a Yaskawa SGDA-04AS.

Arvid

[Al_The_Man]

Looking at the online manuals they look like quite impressive drives, which can be configured as digital, analogue or step/dir control with simulated encoder output to motion controller, how are you controlling them for testing? I assume you have the correct cables for the motor PLG connected? It appears all the commutation is done in the drive and just requires the control signal.
Also they can be used in the speed or torque mode, I assume the reason they are running away is that if you do not have an intelligent controller with the simulated encoder output connected. If you try them in the speed mode they may stay stable but would depend on how the trye of control input is configured.
Al

[arvidb]

I think you have been looking at the SGDAH drives (Sigma II series). The ones I've got are the older Sigma I, which can only do analogue speed or torque control.

I am just using an adjustable +/-10 V source for control, just trying to get a feel for the torque control mode. I'm building a servo controller which will be controlling these (Progress on my servo controller), so I wanted to get a hands-on feel of the torque control mode. It works about as expected, except what I described above.

Also, when set to speed control mode, the motors behave as expected (maybe a bit too much speed fluctuations than I like due to the cogging, but this may be because I haven't found the correct PI parameters in this mode).

Arvid

[Al_The_Man]

The drives I looked at were AC sinusoidal which should be very smooth down to zero speed, if the earlier ones are BLDC then cogging is normal at low speed.
If you have very sloppy feel to it when you try and turn the shaft then that is probabally a tuning issue.
Al

[arvidb]

They call them AC servos in the manual, and write "Control Method: PWM (sine wave driven)".

How do you define cogging? I thought this was some kind of magnetic effect in the motor. When turning the motor by hand, unconnected, I can feel the cogging as loose "steps", a few per turn (~4-8). If I short the motor leads for dynamic braking, the "cogging" becomes permanent - i.e. resists movement not in steps, but regardless of the rotor position. The "feel" of the reistance is the same, just that it's only in certain positions in the turn when it's because of the cogging.

Wouldn't the uneven movement of a BLDC motor at low RPM be because only a few discrete directions are possible for the magnetic field in the motor? Not because of cogging?

Just trying to understand cogging, not sure I've got it right.

Thanks for all the help Al!

Arvid

[Al_The_Man]

They call them AC servos in the manual, and write "Control Method: PWM (sine wave driven)".

How do you define cogging?
Just trying to understand cogging, not sure I've got it right.
Arvid

Although there are a few forms of PWM, a common one is PWM with variable frequency-variable amplitude and although the voltage wave form looks square the resultant current in the winding (inductance) takes on a sine wave appearance.
These are closer to true Sinusoidal than 6 step trapezoidal BLDC.
Cogging may not be the right term to use, but at low speed a BLDC rotates in coarse steps, much like a stepper, but more so. A sinewave driven should not.
Al