A control loop is the simplest powerful idea in engineering — and also the idea that almost nothing in modern automation works without. Your thermostat is a loop. Your car’s cruise control is a loop. The current controller running a traction motor at 10 kHz is the same loop, only faster. The flux observer estimating rotor position from voltage and current is also the same loop, just observing instead of acting. Once you see this — that “loop engineering” is the substrate underneath most of the machines you’d describe as “controlled” — the field becomes much smaller and much more navigable than the textbook taxonomy suggests.
This field guide is the working tour. It is the natural companion to the Digital Control reference (which covers what changes when the loop is implemented on a microcontroller) and to the Optimal Control field guide (which covers what changes when you have a cost function to minimize). This one is about the anatomy — the universal block diagram, the role of each block, the rules of thumb that keep a real loop stable in the presence of noise, delay, and a plant that never matches its datasheet.
What it covers
Twelve sections, about twenty-two minutes of reading.
§ 1 — Measure → compare → correct → repeat. The universal loop in its simplest form. The setpoint, the error, the controller, the plant, the sensor. Every later section is a deeper look at one of these blocks.
§ 2 — Two interactive labs. The hero figures. Drag the setpoint and watch the controller respond. Tighten the proportional gain and watch the response speed up — then keep tightening it and watch the oscillation start.
§ 3 — The plant — what you don’t control. The system being controlled. Why every loop spends most of its life fighting the plant’s quirks (lag, nonlinearity, disturbances). The plant model and why it’s always wrong.
§ 4 — The controller — the part you design. PID and why it’s still the right answer 80% of the time. State-feedback for when PID isn’t enough. The handful of more exotic options (LQR, MPC, adaptive) and when they earn their keep.
§ 5 — The sensor — what you trust. Sensor noise. Sensor delay. Sensor bias. The four-rule discipline that catches the failures sensor-vendor datasheets glide past.
§ 6 — Comparators & error math. What “comparing setpoint with measurement” actually means in code. Wrapped angles. Saturation arithmetic. The wraparound that catches every embedded engineer once.
§ 7 — Tuning rules of thumb. Ziegler-Nichols, Cohen-Coon, the relay test. The practical methods that actually work on real hardware and converge in a single afternoon.
§ 8 — Stability — the one property that decides everything. What it means for a loop to be stable. The phase-margin / gain-margin lens. The Nyquist plot. The intuition that comes before the math.
§ 9 — Noise, delay, and the disturbance rejection trade-off. What you give up when you push the bandwidth higher. The Bode integral. The fundamental limit nobody warned you about.
§ 10 — Nesting loops. The cascade pattern. Inner current loop, outer speed loop, outermost position loop. Why this works and the rules for inner-loop bandwidth versus outer-loop bandwidth.
§ 11 — Practical rules that survive contact with hardware. The dozen field observations. Don’t tune in simulation only. Always log the error signal. Watch for integrator windup. The handful of habits that compound.
§ 12 — Further reading. The texts that actually teach this well. Åström & Murray. Doyle, Francis & Tannenbaum. The control-engineering blogs worth a subscription.
Read it
The field guide lives at its own URL with a warm-paper editorial layout — blue for measurement, orange for control action, green for plant. Twelve sections plus two interactive labs.
← Back to Autonomy