PIDSnap

Noisy PV and valve chattering

A valve that constantly makes small movements is wearing itself out for no control benefit. The question is whether it is responding to real process movement, to measurement noise, or to noise the controller itself is amplifying.

The distinction matters because the fixes are completely different, and the most common response — adding filtering — makes two of the three cases worse in the long run.

Derivative action amplifying measurement noise Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over. high low PV SP Output PIDSnap time
Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over. The output is noisier than the measurement, not quieter. Every wiggle on the process variable appears larger on the output. The valve is being asked to chase turbulence. On a flow, pressure or speed loop this is almost always derivative action that should not be there.

Compare the noise on the output against the noise on the measurement

This is the whole diagnosis in one step. If the output is noisier than the process variable, the controller is amplifying — which on a fast loop means derivative action almost every time.

If the output is about as noisy as the measurement, the controller is faithfully responding to what it is being told, and the question moves to whether the measurement is telling the truth.

If the output is quieter than the measurement, the controller is already filtering and the noise you are seeing is in the process or the instrument.

Real process noise versus instrument noise

Real process noise has a physical cause you can usually name: turbulence past a flow element, boiling, mixing, a reciprocating machine upstream. It changes with process conditions.

Instrument noise does not. Electrical pick-up, a failing transmitter, a loose connection, or blocked impulse lines produce noise whose character stays the same regardless of what the process is doing. Compare a period at high rate against one at low rate — if the noise is identical, suspect the instrument.

What causes it, most likely first

Derivative action on a noisy measurement

most common tuning can fix this

Derivative multiplies rate of change, and noise is nothing but rate of change.

How to confirm it
Output noticeably noisier than the process variable.
What to do
Set rate to zero on flow, pressure and speed loops.
Derivative action amplifying measurement noise Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over. high low PV SP Output PIDSnap time
Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over.

Gain too high on a noisy loop

common tuning can fix this

Proportional action amplifies noise too, just less dramatically than derivative does.

How to confirm it
Output noise scales with the gain setting.
What to do
Reduce proportional action, or deal with the noise at source.

Blocked or gas-locked impulse lines

common tuning cannot fix this

A very common source of genuinely noisy differential pressure measurements.

How to confirm it
Noise character unchanged across different process conditions.
What to do
Blow through the impulse lines and check for trapped gas or sediment.

Operating at the bottom of the instrument range

common tuning cannot fix this

Transmitter accuracy is quoted against full span, so a loop running in the bottom tenth of its range works with a signal whose error is a large fraction of the value.

How to confirm it
Compare the normal operating point against the calibrated range.
What to do
Re-range the transmitter around the real operating envelope.

Questions that come up

Why not just increase the transmitter damping?

Because damping adds lag, and lag limits how well the loop can ever perform. It makes the trend look better while making the control worse. It is the right answer only when the noise is genuinely faster than anything the loop needs to respond to, and even then it should be a deliberate choice with a known cost.

Related

Last reviewed 2026-08-01.

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