Diagnose Before Retuning

The Core Idea

Poor loop performance does not always mean the controller gains are wrong. Diagnosis uses the target, measurement, output, and independent observations to distinguish a tuning problem from a limit, delay, disturbance, or measurement problem.

SP (setpoint) is the target temperature. PV (process variable) is the measured temperature. CV (controller output) is the controller's output command to the heater, shown as a percentage. PID means proportional, integral, and derivative control; PI uses proportional and integral action without derivative. The controls P, I, and D adjust those three actions.

What you will learn

  • Use SP, PV, CV, and configuration evidence to choose a first action.

  • Separate tuning, noise, actuator capacity, direction, and measurement problems.

  • Verify a correction before making another gain change.

Why it matters

Retuning the wrong cause can waste time or make the response worse. Identify what the evidence supports before changing a gain.

Five calls about a poor loop

You receive five commissioning cases. Cases A–D use the same positive-gain heater: 20 °C start, 15 s lag, 5 s dead time, and 0–100% output. Read the target and settings for each case; one case requests a different temperature.

For each case, choose a case button and do the following:

  1. For A–D, press Run and let the supplied settings run to 120 simulated seconds. The simulator speed changes playback time, not the process. Compare the measured temperature (PV) and heater command (CV) with the target (SP). Case E has field readings instead of a simulator; use those readings to calculate the temperature.

  2. Select a Primary problem and a First action, then press Check diagnosis. A correct pair marks that case with a check. Complete all five cases to finish the activity.

  3. Optional: compare a correction. You don't need to save a trace to answer the diagnosis. For A, B, or D, press Save this trace after the run finishes, make a deliberate change, press Restart, and run again for 120 seconds. The saved run stays on the same graph as dashed purple lines: compare those with the new temperature and output. The lines overlap until the new run differs. Switching cases clears the saved trace.

Choose the first action from evidence

Start with the operating request and signal validity. Confirm the engineering units and scaling, compare a suspect measurement with an independent observation, and check whether the actuator has room to respond. A tuning change cannot create missing heating capacity or make an incorrect temperature indication trustworthy.

Next check the control direction. On this heater, raising output raises temperature. Correct negative feedback therefore needs reverse action: with SP unchanged, falling PV should raise CV. A cooling process may need direct action; the process response determines the choice.

Only after those checks should you compare gains. Name the problem in the trace, make one deliberate change, restart from the same conditions, and finish the observation window.

Verify the proposed correction

After diagnosing Case A, try reducing P from 3 to 2 and I from 0.3 to 0.12, then restart and compare with the saved run. For Case B, try reducing D from 3 to 0, or changing the measurement filter from 0 to 1 s, while keeping the noise sequence and target unchanged. Check output movement as well as peak temperature.

For Case D, select the appropriate Control direction, restart, and check whether heating now moves PV toward SP. For the capacity case, state what equipment or process-load information you would investigate. For the measurement case, verify the corrected scaling at both endpoints and midpoint before trusting automatic control.

Hint — Two different reasons for a temperature gap

A command pinned at 100% and a temperature that has nearly stopped rising suggest a different problem from a command stuck at 0% while PV is below SP. Check the physical capacity and the direction setting before increasing gains.

Worked diagnosis after all five cases

A: the clean delayed loop has excessive gain for the requested response; make a fair PI comparison. B: noise and derivative action drive repeated output reversals; investigate the measurement and compare less D or filtering.

C: the 140 °C request exceeds the heater's approximately 120 °C maximum at 100% output; investigate available heating and load. D: direct action is inappropriate for this positive-gain heater; correct the action before tuning.

E: 13824 counts is the midpoint of a 0–27648-count input. The 0–100 °C transmitter therefore indicates 50 °C, matching the independent thermometer. The configured 0–200 °C scaling incorrectly reports 100 °C; correct and verify that conversion.

Check your understanding

For every case, record one observed fact, your diagnosis, and your first action. For a case you corrected in the simulator, compare the original and corrected traces. Explain what additional field evidence you would need before applying that diagnosis to a real loop.

Enable JavaScript to run the embedded simulator and complete the interactive exercises.