Whether the error between setpoint and actual has crossed its limit deserves its own visual cue

What it is

A deviation cue encodes directly whether setpoint–process error exceeds an allowed envelope, instead of asking the operator to subtract two numbers and judge for themselves. It must be designed separately from absolute high/low alarm, because a variable can sit inside a safe absolute range while still failing the control objective — temperature between its safety limits but far from setpoint is a control-quality problem, not a safety problem, and the two need different indication channels.

Why it happens

Loops naturally tolerate transient error and measurement noise: a disturbance or a brief overshoot after a setpoint step is normal and should not trigger a cue every time. For deviation to become actionable it must cross a threshold in magnitude, duration, or pattern — large but momentary noise should not alarm, while a smaller but persistent steady-state offset should. A separate deviation cue moves this judgment out of the operator's head and into the system, but at a cost: it must be visually distinct from absolute-state indication in colour, icon, or position, or the two channels collapse into one at the perceptual level and the operator sees "it turned red" without being able to tell whether the process left target or approached a safety limit.

Where it stops holding

Startup, a setpoint step, and manual mode routinely produce large, expected deviation. A fixed tolerance band will keep firing in these situations, generate noise, and train operators to ignore it — alarm fatigue reproduced at the deviation-band level. The tolerance must come from process characteristics and the control objective itself (a loop's normal tuning precision, the transition time a process step allows), not from a designer's guess at a number that looks reasonable. A wrong tolerance is worse than none, because it manufactures a false impression that everything is fine.

Applying it

Show the error value, upper and lower band limits, how long it has persisted, and the currently applicable mode (startup, normal, manual — tolerances can legitimately differ), coded distinctly from absolute alarms through icon shape rather than colour alone.

• Configure bands per mode or process phase rather than as one global constant.
• How to check: replay historical step responses, slow drift, and stuck-actuator episodes; count how often the cue misfires during a normal post-step overshoot and how often it fails to fire during a sustained actuator-stuck deviation. Both false-positive and false-negative counts must be measured, not judged by feel.

Related

• Same group: Y3.08.1 Co-displayed setpoint and process value · Y3.08.3 Obsolete setpoint diagnosis · Y3.08.4 Hidden control target
• Nearby: Y3.09 Manual–automatic transfer · Y2.03 Alarm priority classification
• Search terms: Deviation-band indication · process control display · industrial human factors