Dual loop motion control with one encoder
Inside Machines: Motion control system example shows how mechanical stiffness can improve by a factor of 100. Dual loop control, often used to improve the secondary feedback device, is simpler when second encoder functionality is integrated in the motor.
Dual loop control is often used to improve the performance of a motion control system. Although this appears complex, overall system cost and complexity needed to reach the desired level of precision may often be reduced significantly. In the example system, mechanical stiffness was improved by a factor of approximately 100 by the use of a secondary feedback device.
For a single loop lead screw system, the feedback is usually located on the back of the motor. Each element in the system between the sensor and the load adds uncertainty.
- The motor mounted feedback is blind to torsion of the shaft with load.
- The coupler between the motor and the lead screw distorts under load, causing error between the motor shaft and the lead screw.
- The lead screw itself can also exhibit torsion under load, varying with both load and position of the nut with respect to the driven end of the screw.
- The lead screw itself typically has periodic error which can be very expensive to minimize, and the nut backlash minimization can add load as well as wear to the system.
- Finally, the thrust bearings' accuracy also directly affects the system.
The result is a system in which many elements can degrade the performance, and tightening the system (reducing the difference between where motion should be and where it is) can be expensive. As the feedback only senses the back shaft of the motor, much of what happens after that is unknown to the control system, so it cannot be corrected.
Adding a second feedback sensor (the optical track shown in Figure 2) that measures close to the end effector of the system can significantly improve performance and reduce the cost. Looking at the above system again, most of the error contribution is inside the secondary control loop; this error can be measured and corrected. With a good gain margin, the system error can be reduced nearly to error of the secondary feedback sensor. Rolled lead screws can replace ground lead screws, with their thread pitch variation, both cumulative and periodic, compensated by the use of the additional feedback sensor. Differential thermal expansion coefficients of various elements also can be compensated.
For effective dual loop stiffness, a good system bandwidth is still needed. Elements of the system should be selected to minimize lost motion. The tighter system reduces the corrective action required of the control system just to hold a stationary position. Other periodic errors, such as lead screw pitch variation and thermal effects, do not affect system stability and are well tolerated, but if significant may require more control effort. Within the loop, the errors typically are reduced by approximately the gain of the system, so a high-gain system is desired.
Velocity feedback provides the phase stability for the system, as the position feedback is more than 180 degrees behind the phase of the supplied torque (180 degrees is the most optimistic). This velocity feedback is typically taken from as close to the motor to allow sensing and control of the energy being put into the system while lost motion or backlash is being corrected.
The Ibex Engineering lead screw system tested uses a 0.1 micron secondary feedback, direct driven by a hybrid servo with integral resolver (Figure 1) to provide 32,000 count resolution at the motor, corresponding to 0.25 micron per count. The closed loop system holds position within +/- 2 counts when adding a 20 Newton (5 lb) load. The motor corrects the compliance in the system under this load by moving approximately +/-225 counts (2.5 deg) when the load is applied. With the same system operating single loop, the motor will hold steady within +/-2 counts as a 20 Newton (5 lb) load is applied in either direction, while the linear scale shows a movement of approximately +/- 120 counts (12 microns). Thus a properly implemented dual loop system excels in compensating lead screw tolerances, and in significantly stiffening the resulting system under load.
Integrating motor, resolver
Resolvers determine position by operating as a variable transformer between the excitation coils and the sensing coils. The coupling varies, typically sinusoidally, as the rotor component is rotated. The excitation in a typical resolver is either brought in via brushes to a coil within the rotor, with the sensors on the stator, or via a separate excitation stator that is coupled to the rotary transformer on a portion of the rotor, which then excites the rotor windings opposite to the stator sense windings. This configuration requires multiple rotor and stator components for sensing position, all of which need to be added to the motor itself. In addition to the added components, volume, and weight, a separate sensor excitation drive must be provided, and some degree of magnetic isolation is needed to prevent the motor pulse width modulation (PWM) noise from getting into the resolver sensing circuits.
With typical single coil excitation, dual coil sensing, the excitation sine wave passes through zero volts twice per drive cycle, causing "blind" periods for the resolver when it must use filtering techniques to estimate what the motor is doing while the resolver is in this excitation zero crossing region. These techniques can create delay and loss of phase margin within the system.
How an integrated encoder helps
Integrating the resolver within the motor overcomes many of these obstacles by reusing the motor magnetic paths and PWM drive to provide excitation to the position sensing coils. The motor ripple current, which results from the PWM drive, causes a time varying flux component with a high-frequency component. The degree of overlap between the rotor and stator teeth modulates the effective reluctance of the paths through different sets of stator teeth. The ripple flux is thus directed variably to different sections of the sensor coil, approximately resulting in sine and cosine signals when the sensed voltages are properly sampled, synchronized with the motor PWM drive. As the sine and cosine signals are updated every PWM cycle, the delay is minimal and does not degrade system phase margin.
The electrical cycle of the sensor matches that of the motor. With the sensor built on the same magnetic design as the motor, the signal is automatically aligned and is used for commutation without needing phasing adjustments. When using a high-pole count hybrid servo motor shown, there are 50 electrical cycles per revolution. With 640 positions per cycle, the resulting resolution is 32,000 counts per revolution.
The sensor consists of a robust polyamide flex circuit holding the sense coils, which is added to the stator of the motor. The sensor can be made to exceed the temperature and other environmental capabilities of the motor, so that the resulting system has the environment rating of the motor, with a footprint matching the motor.
- Donald Labriola, PE, is president, QuickSilver Controls Inc. Edited by Mark T. Hoske, content manager, CFE Media, Control Engineering, firstname.lastname@example.org.
www.controleng.com/archives under October has 2 more photos and additional information about the motion system described and the integrated motor and encoder.
- Dual loop control can improve the performance of a motion control system.
- System cost and complexity required for the desired level of precision can be reduced.
- A 100-fold increase in mechanical stiffness of the motion system was achieved with a secondary feedback device integrated in the motor.
What motion system designs can you improve by using dual loop control?
This online extra has 2 more photos and links for more information.
See related motion control feedback articles below on position control without a separate encoder and about a robotic dolphin below.