3-Phase Brushless DC Motor Loses Torque 1

[lmc82]

Hi,

I've searched this thread and can't find an answer to the following question, so here goes.

We have a camera pan and tilt gimbal with direct drive 3-phase brushless DC motors mounted on each axis. We have the control system setup so that the camera is pointing to one spot regardless of the motion of the base (stabilisation). We have added a slight drift to the azimuth stabilisation control (makes the following problem repeatable); movement is approximately 2 electrical degrees every 5 mins. When we start the system up and turn on stabilisation, everything works fine.

Now the problem is this. We excite the base with a vibration profile for approximately 10mins while the gimbal is stabilising. The PWM duty cycle applied to the azimuth motor varies between 10 and (close to) 100%. Once the vibration profile is turned off, we leave the gimbal stabilising. Very slowly the control system increases the PWM duty cycle to the azimuth motor, which increases current until the overcurrent protection is tripped. The output torque of the motor drops as the PWM duty cycle is increased. So it looks like the control system is increasing the duty cycle because the motor is some how losing output torque.

The commutation is trapezoidal, with complementary PWM applied to the low side gate (e.g., the high side gate of Phase A is switched on/off, the low side gate of Phase A is turned on/off (opposite to the high side to avoid shoot through), with the low side of phase C always on). We have also tried trapezoidal without complementary PWMing (i.e. letting the free wheeling diode handling the recirculating current instead of the gate). The gates are IGBTs.

The reasons I can think of for losing torque are:
1. The control system is applying power to the wrong phase. This doesn't seem to be the problem since it doesn't occur when the system is cold.
2. There is a mechanical problem with friction that only occurs after vibration. I'm about to confirm this isn't the problem, however I really don't think this is the case.
3. The motor core is saturating (or something?) since we are essentially not moving and PWMing one phase, which is increasing the back EMF and reducing torque. However I would expect a drop in current (or am I thinking about this incorrectly), but the current increases to the same level as it would be with maximal torque.

Any ideas or suggestions would be greatly appreciated.

Cheers,

Leon

 

[burnt2x]

Hard to understand. There's no 3-phase DC motor, AFIK!

 

[lmc82]

The motor has 3 phases and no brushes, the commutation is done electronically. You have a set of switches for each phase (IGBT, MOSFET, etc), connecting the phase to either DC Bus or Ground. Just Google "Brushless DC Motor". The "3-phase" has nothing to do with "DC", the motor has 3 phases.

It definitely isn't a mechanical problem; that was one of the first things anyone thought of (prior to me looking at it).

We are going to sinusoidal commutation to see if that improves the situation.

 

[MikeHalloran]

How do you remove the I^2*R heat?


Mike Halloran
Pembroke Pines, FL, USA

 

[lmc82]

The motor is interference fit to the base of the gimbal, which is basically a large heat sink.

It does look like it could be temperature related, given that the problem doesn't occur before applying the vibration profile. The information on the motor is a bit sparse in terms of derating due to temperature. How would increased motor temperature affect the torque output of the motor?

Update: Sinusoidal commutation seems to reduce the severity of the problem, however it doesn't eliminate it completely.

 

[lmc82]

Oh, and we have placed thermocouples on the motor housing and seen no significant temperature rise.

 

[MikeHalloran]

I'd be inclined to look at reducing the heat input to the motor, e.g. by using PWM on the 'full on' coil and reducing the on time of the active coil where that's possible, or just reducing the supply voltage until the system can't quite maintain position, and then jack it up a little.

If you're using a commercial control, I'd look closely at the data sheet to see if it has another loop or other means to limit heating when the motor isn't moving much.

... or add a fan to cool the motor.

I'd surely start collecting motor temperature data....

Mike Halloran
Pembroke Pines, FL, USA

 

[lmc82]

Thanks Mike,

It really does seem to be a temperature issue at this stage. Unfortunately we can't get a decent temperature reading from the motor given the motor arrangement. The thermocouple we currently have is placed on metal work attached to the motor.

I think we need to get some more information from the motor manufacturer, however given their large size (and our very small volume of purchases) it is a bit difficult getting information out of them in a timely fashion.

I'm assuming your thought is that the coil temperature has increased to such a point that the permanent magnets field strength is reduced, causing the control system to apply a greater PWM duty cycle, which cause the coils to heat up even more?

Cheers,

Leon

 

[LionelHutz]

Possibly an instability in the control loop causing it to go unstable, quickly fluctuating between positive and negative torque so the motor still appears to be stable?

 

[sreid]

In order to get negative torque you must switch from phase A and C conducting to phase A and B conducting.

Most Motors are desiged for Sine wave comututation. The motor torque will drop as you approach a Hall transition if you are six step commutation.

 

[iop995]

What is mechanical gear ratio motor-camera? If it's 1 (direct connected) I think, in this case, controller try to eliminate error position and it's not possible with gear ratio 1. Check motor datasheet and get or estimate stall torque (at so low angles, DC motor is similar to a step motor - and torque lower at very small angles). Try to add a gear to system and also, try to modify (reduce) integral coefficient of controller. Adding gear will reduce speed so if motor speed may not be increased, need to change it.