Why the different response of two Simulink models?

Here is a follow-on question to this original question:

How can I remove the few peaks before the overshoot, so I only have the overshoot peak and then it directly transition into s... - (mathworks.com)

I have two different models of a buck converter (see attached). In the first, the feedback summing junction has a negative sign forming the error signal and the PWM comparator after the PID uses "<". In the second, the feedback summing junction has a positive sign, and the comparator uses ">=". I fully expected the two models to produce the same response, but the second model has over twice the overshoot. Saturation limits of the PID made no difference. Can anyone provide a simple explanation of why the different response?

Answer 1

[Sabin]

The two structures are not the same. A PID controller is designed to take an error signal as input, which is the difference between a setpoint (the desired value) and a process variable (the measured value). The controller then calculates the output required to minimize this error by adjusting a control element. This is what the first structure is doing. In the second structure, the PID will try to minimize the measurement-setpoint, as the measurement starts from 0 it means the controller will see as input -setpoint and will try to drive it to 0. The controller will overcompensate or undercompensate because it is not receiving a proper error signal which in this case leads to increased overshoot. However, in a some situations the controller will have stability issues.