High-Accuracy Closed Loop Force Feedback

Myostat Motion Control, Ontario, Canada

Hand-driven press assemblies, which require a specific amount of force applied for a given amount of time, present challenges when attempting to accurately measure or replicate results in production or quality control processes. In contrast, servo-driven digital presses are able to provide accurate, consistent, and repeatable results while outputting trackable data. By controlling position and current, the servo-driven digital press allows for stable, accurate, and repeatable control.

Fig. 1 – Feedback from the force transducer is fed back into the motor to close the loop for force control.

Myostat was approached by an integrator for a large North American electronics manufacturer looking to produce cost-effective servo-driven presses. A force gauge, specifically the Mark-10 Series 5 (www.mark-10.com), was certified by the customer and pre-selected as the preferred gauge. The system also needed to be designed within current hardware specifications, so the Cool Muscle CM1-C-17L30C-RT3 integrated motor and force gauge were fixed.

While open loop servo presses may provide adequate accuracy (±1N), and are still used as a cost- effective solution for many applications, they rely on multiple variables such as motor current, speed, acceleration and control gains to be set in order to achieve accuracy at a defined target force. This use of multiple variables makes this option an inappropriate choice for applications with multiple force targets.

For more precise applications that are able to run to multiple force loads and are programmable with no control gain changes, a higher accuracy solution is required. Additionally, in order to allow for higher cycle time and throughput, the servo must be able to switch seamlessly between position and force control.

The chosen force gauge offers two options for feedback: analog and serial. The analog option outputs the raw signal from the load cell. The signal is amplified to a 0 to 1V signal over the calibrated range of the force gauge. The advantage to the analog option is that the signal is in real time, allowing the controller to update every 1ms (time period of control loop). However, with this option, the signal has a slight non-linearity and does not directly match the force gauge display which the user can zero. In addition, the resolution is limited by the microcontroller's analog input (10bit) of the integrated servo system.

Serial communications allows the gauge display value to be streamed out at up to a 4ms time interval. Utilizing this option ensures that the serial data matches screen data and allows the motor to write directly to the gauge's serial port, allowing the gauge to be controlled by the servo, and minimizing user interaction. Full gauge resolution is available for this option; however, 4ms data streaming is slower than the ideal 1ms interval, which matches the integrated servo's force control loop frequency.

After considering the options, the serial port was chosen for implementation. A PI controller was utilized for the force feedback control loop and feedback was supplied over the serial port in 4ms intervals. In this design, the error between the force gauge signal and the target feeds into the micro-controller's motion generator through the PI controller. The motion generator generates a target speed within the servo motor, which finally translates to a position change on the force gauge. The quick response time of the servo positioning system translates to accurate force control. The integrated nature of the motor and the ability of the motor to directly read the force gauge signal reduced complexity within setup, testing, and assembly.

While designing a complete solution from scratch would have resulted in improved performance, the result would have been more costly and would not have met the required time schedule. This implementation produced very accurate results where realizing the full resolution of the gauge is possible. For example, a 250N gauge can give an accuracy of 0.05N and a single set of controller gains offers a full range of target forces from 0.05N to 250.00N. While the 4ms feedback rate limits positioning speed and acceleration, it was well within the specifications for this particular application. The motor is able to detect if force is applied too early to within 0.25mm, which allows for termination and part rejection by the system before a load is applied that could cause damage to the test product.

Potential upgrades to the design could include higher frequency and real-time feedback of the gauge signal. As an example, a PWM signal directly to the motor's micro-controller's internal high speed counter that carries the force value would be more ideal. While 250Hz feedback frequency is adequate, this solution would match the control loop's 1KHz frequency and, would create realtime feedback, eliminating the delay in generation, transmission, and parsing of data.

This technology was done by Myostat Motion Control, Ontario, Canada. For more information, visit http://info.hotims.com/40435-167.