Skip to main content

Comparing Closed Loop Methods

MethodCorrectsWeaknessTypical FRC use
P onlyProportional to current errorPermanent steady-state errorQuick, rough position holds where a small error doesn't matter
PDError, damped by its rate of changeStill has steady-state errorFast-moving mechanisms where overshoot is worse than a small final offset
PIError, plus its full historyOvershoot, risk of integral windupMechanisms needing zero steady-state error, without needing to be fast
PIDError, its history, and its rate of changeThree gains to tune instead of oneGeneral-purpose position and velocity control

Every method in this table is built from the same three ingredients covered throughout this section, proportional, integral, and derivative response to error, just weighted differently depending on which terms are in play.

Where This Leads Next

A well-tuned closed loop is very good at reaching a setpoint accurately. It says nothing about what that setpoint should be doing over time, jumping straight to a final target and letting the controller fight its way there produces exactly the fast, jerky, overshoot-prone motion the gain tradeoffs in this section keep warning about. Shaping a smoother path for a closed loop to follow, instead of a single abrupt target, is the subject of Motion Profiling.