VII. Summary and Conclusions

The purpose of this project is to retrofit an aged but relevant experimental apparatus that is used to study magnetically actuated motion and friction. This has added to its 25-year lifetime and allow it to continue to function. The control system of this apparatus is described in this paper and in greater detail in Alex Rokhvarg’s original thesis. The Adafruit Metro M4 possesses the computational speed and precision required to function as a replacement controller. It is also selected because it has the necessary hardware and features for this application. The experimental data shows that the controller can move the payload to the user-defined destination albeit with some degradation in performance. I believe that this degradation can be addressed by retrofitting some of the components of the motion plant. Suggestions for future work are listed below.

Suggestions for Future Work

Because of the high bandwidth and accuracy of the control system, a possible future use of the apparatus could be to assess the quality of certain proximity sensors that can measure the distance to an object without physically touching it. Examples include laser rangefinders and ultrasonic transducers.
The Arduino serial monitor paired with the Matlab routine makes for a clumsy user interface. Accessing the QSPI flash memory is also cumbersome. A single GUI application might be able to replace all this. In such an arrangement, nothing can interfere with the Metro M4’s hard real-time control of the motion plant. The Metro M4 should be the master and the host computer should be the slave. The market has several GUI applications that can complement or replace the Arduino serial monitor. An alternative to this could be to improve the Matlab interface.
Another task might be to replace the linear potentiometer with a high-quality, calibrated proximity sensor. This proximity sensor will have to possess high accuracy and response speed. A non-touch sensor will remove the influence of friction force and the spring’s reaction force and attenuate, if not eliminate, the drift in the force response.
Because of the high computing power of the Adafruit Metro M4, a different controlling algorithm, such as a reduced-order observer, should be attempted. Perhaps a more accurate model of friction compensation and spring reaction can be fabricated. A future project can explore whether having a sampling period shorter than 800-μs will improve performance.

Page updated

Google Sites

Report abuse