Utility · for the week and after
Resources
Control is a century-old discipline with a deep literature; these are the entries that match this week's altitude, plus the one classic practitioner series every embedded programmer ends up reading.
References you will actually use
| Resource | What it is | Use it when |
|---|---|---|
| Wikipedia: PID controller | A thorough article: history, the math, tuning methods, limitations, industrial variants | A second pass over any term, or meeting a variant (PI-D, feed-forward) in the wild |
| Improving the Beginner's PID | Brett Beauregard's classic series: a naive PID upgraded step by step (sample time, derivative kick, windup, on-the-fly tuning) into the Arduino PID library | Seeing every refinement in this hub's class argued one at a time, in C, by the person who wrote the most-used hobbyist PID |
| Wikipedia: Ziegler–Nichols method | The classical tuning rules, their table and their known aggressiveness | Running the ladder's rung 2, or citing the method's limits in your log |
| MicroPython: time module | The reference for ticks_ms, ticks_diff and friends, wrap semantics included | Writing or debugging the fixed-period loop's timing |
| Wikipedia: Control theory | The wider field: stability, feedback, where PID sits in it | Curiosity upward: what the loop looks like with the full mathematics turned on |
Background reading, stage by stage
| After reading… | Deepen with |
|---|---|
| The loop | Wikipedia: Feedback, Bang–bang control and Hysteresis, the baseline controller and the band that tames its chatter |
| P, I, D | Wikipedia: Setpoint, Process variable and Low-pass filter, the vocabulary formalized, and the filter family your averaging belongs to |
| Tuning & the big loop | Wikipedia: Edge computing, SCADA and Programmable logic controller, the industrial names for where your two loops live |
Practice between classes
- Napkin drills. Draw the block diagram cold, five times this week, until the signal names place themselves. It is a final-exam staple and a job-interview staple in your field.
- Trace drills. Invent three ticks of PV readings and hand-compute u with given gains, Week 7's trace-table habit on the controller. The quiz's first question is the template.
- Curve flashcards. Sketch the five diagnosis shapes from memory with their one-line verdicts. Being able to see a tuning fault is the skill that transfers to every loop you will ever meet.
- One more plant. If you kept the second LED or a resistor heater idea from the kit, tune a second loop and compare gains: nothing teaches "gains belong to the plant, not the algorithm" faster.
- Project thinking. For each project idea you bring Monday, write its SP, PV, u and disturbances in one line each. An idea that fills those four blanks is a project; one that cannot is a demo.
Glossary
| Term | Meaning |
|---|---|
| Closed loop / feedback | Measure the result, compare to the goal, correct, repeat; the structure that absorbs disturbances |
| Setpoint (SP) | The target value the loop holds |
| Process variable (PV) | The measured value; the controller's only window on reality |
| Error (e) | SP − PV; signed, and the sign steers |
| Control output (u) | The actuator command; here, LED duty in percent, clamped 0–100 |
| Plant | Everything between u and PV: LED, air, room, LDR |
| Disturbance | An outside push on the plant the loop must cancel |
| Bang-bang control | Two-level control with hysteresis; oscillates by design |
| Proportional / integral / derivative | Push per present error / per accumulated error / against the error's trend |
| Gain (Kp, Ki, Kd) | Each term's volume knob; speed bought at stability's price |
| Steady-state error | The settled gap; nonzero under P-only, erased by I |
| Rise time / overshoot / settling time | The step response's speed, excess and completion grades |
| Saturation | The actuator at its limit; where the math and the hardware disagree |
| Windup | Integral inflation during saturation, repaid later as overshoot |
| Anti-windup | Any scheme that stops integrating while saturated; ours is conditional integration |
| Derivative on measurement | Differentiating −PV instead of e; same damping, no setpoint kick |
| Derivative kick | The one-cycle output spike from differentiating a setpoint step |
| Sample period (Δt) | The loop's measured heartbeat; I and D both scale with it |
| Ultimate gain / period (Ku, Tu) | The gain at steady oscillation and that oscillation's period; Ziegler–Nichols's two inputs |
| Supervisory control | A central layer adjusting local loops' setpoints; fast local, smart central |
| Edge computing | Running computation beside the physical process; the modern name for the PLC's oldest habit |