PIDSnap

Derivative kick on setpoint change

Derivative kick is a spike in the output at the instant of a setpoint change, not the overshoot that follows minutes later. When rate acts on error, a step in setpoint is a step in error, and derivative of a step is a kick. The valve jumps. The process did not ask for it.

People see the kick, call it overshoot, and take gain out. The next setpoint change still kicks, because the gain was never the term that differentiated the step. Put derivative on the process variable and the kick goes; the useful part of rate, acting on a moving PV, stays.

Kick versus overshoot

The kick is immediate — one scan, maybe a few, at the moment SP moves. Overshoot is the PV sailing past the new setpoint after the process has started to respond. They get the same complaint on the radio. They are not the same plot.

If the output spike happens before the PV has moved, it is derivative on error, a filter that is too light, or a setpoint that was ramped as a step. It is not reset.

Derivative on PV

Most industrial blocks can apply derivative to PV rather than error. That is the usual correct setting. Setpoint ramping is the other half: if the operator (or the master of a cascade) slams SP, even derivative on PV cannot stop the proportional term stepping. Ramp the setpoint on loops that cannot take a step.

What causes it, most likely first

Derivative acting on error

most common tuning can fix this

A setpoint step is a step in error. Rate of that step is a kick.

How to confirm it
The output spikes at the SP change, before the PV moves. Turning derivative off removes the spike; the later PV overshoot may remain.
What to do
Set derivative on PV. Leave the rate number unless it was also too large.

Setpoint stepped on a loop that needed a ramp

common tuning cannot fix this

Even with derivative on PV, the proportional term steps with SP. On a heater or a composition loop that step is an upset.

How to confirm it
The spike is proportional-sized, not a needle, and remains with rate at zero.
What to do
Ramp the setpoint. Do not take all the gain out to hide operator steps.

No filter on a noisy PV with rate on

common tuning can fix this

Derivative on PV still amplifies noise. A noisy flow or a bouncing level makes the output chatter continuously, which operators then describe as kicking on every change.

How to confirm it
The output is busy even when SP is still. Rate is not zero.
What to do
Filter the PV or take rate off a noisy loop. Flow almost never wants derivative.

Questions that come up

Is this the same as overshoot after a setpoint change?

No. Overshoot is the PV going past the new setpoint. Kick is the output spike at the moment of the change. Fix kick with derivative on PV. Fix overshoot by looking at gain, reset, and whether the loop was wound up.

Related

Last reviewed 2026-09-10.

Run a free PIDSnap check on this loop

Two photos, thirty seconds, no signup. The pre-flight check finds the structural problems that no amount of tuning will fix — before you touch the controller.

Start a free check