What to implement for gaining experience in control loops?

[atferrari]

I intend to gain hands-on experience in control loops (of the PID type). At this stage, op amps to be used.

I am looking for suggestions (the easiest / simplest / cheapest / quickest) on two concrete points:

1) a "plant" that I could implement on my bench.

2) way of tracking ( eventually recording) set point - system output - error - time. If things go not too fast, I guess I could do it manually but not really sure.

After reading a lot, starting with Bob Pease, I concluded that something related to temperature control or motor speed are the best candidates. Have read about a pendulum but sounds complex to start with so I discarded it already.

Suggestions appreciated.

[orion242]

1) a "plant" that I could implement on my bench.

Resistor with a bit more power than its rated with a thermocouple or thermistor strapped to it.

[T3sl4co1l]

Audio amplifier -- practical application, can be very fast (good for oscilloscope measurements, and potentially getting into more involved concerns like stray inductance and its effect on loops and stability).

Switching supplies

Filters, frequency response -- it all goes on the Bode plot after all

etc.

Tim

[dom0]

A large part of analog control loop design is relating bode plots to their real world effects and vice versa - deducing specific points of a bode plot from a measured real world phenomenon - the reason why many people stare at step responses and tweak things with them.

[Kalvin]

Also a simulator, like freely available LTSpice, can be handy. You can build a PID controller and generate different stimulus (sine, step, pulse signals) with an ease, and put different loads (resistive, capacitive, inductive, combinations of R L C) and see how the control loop behaves. Simulators allow paramteric sweeping, which is a nice feature to study "how changing this parameter will affect loop stability or loop behaviour". Of course, building something is the ultimate breadboard where the theory meets practice.

Ps. Simulator should be considered only as a fast calculator, not a substitute of intelligence. Simulator is a good example of a SISO device ie. s**t in - s**t out. So, if the simulation model is bad, the results will be bad - that is typically easy to detect. If the simulation model is good, the results are typically fine - and the user is happy. Worst yet, it the simulation model is wrong and you do not know that, you may still get results which look good until you build the circuit and face the reality. Simulators are nice to have, but one shouldn't trust them too much.