Mechanical

Although the cart was not complex mechanically, a number of realizations that came later in the project would have been quite helpful early on in the construction of the system. The first realization was that the pendulum should be about one meter long. Initial work on the cart assumed that the pendulum would only be about one foot long, but it turned out that a longer pendulum weighted at the end is much easier to balance. Also, the pendulum needed to be stiff. Any give in the pendulum along the direction of motion would create problems in the control circuit. Additionally, testing the motor speed with the available voltage early on would have avoided a number of headaches associated with changing the supply voltage and pulley ratio late in the project. Also, the angle at which the cart couldn't recover should have been more conservatively estimated to avoid having to remachine the stops on the pendulum bracket. Furthermore, the traction requirements of the high torque motor used should have been more carefully considered. Additional weight over the shafts as well as wheels with better traction would have helped immensely.

Circuitry

Although the PD control circuit was built and tested early on, it would have been better to integrate the controller with the motor early on. The addition of the motor and motor control circuit lead to the rise of a series of difficulties including burning out transistors and op-amps as well as losing the bidirectional motor control. Swapping LM10 op-amps for more robust 741 chips alleviated the burn out problem and the addition of fly-back diodes solved the issue of losing bidirectional motion.Additionally, using an on/off switch to connect and disconnect the motor rather than reconnecting the lead to the breadboard and creating a spark virtually eliminated the burning out issue.

Overall

Although the project did not end up functioning entirely according to plan, once the issues mentioned above were addressed, the performance of the cart improved by leaps and bounds. Although the project was started rather early on, many difficulties arose from not connecting the electrical and mechanical aspects of the project until late in the project timeframe. Had these issues been identified and addressed earlier, the final performance of the cart would have been far superior to the clumsy operation it exhibited when presented. It should not take much work to improve the cart to make a properly functioning demonstration of PD control.