Two instruments can misbehave in ways that look almost identical on a trend, yet arise from different physical causes and demand different fixes. Hysteresis and dead band are the classic pair. Both make an output lag reality, both are worse near a change in direction, and both get lumped together in casual conversation. But they are not the same thing, and calling one by the other's name sends a technician chasing the wrong repair. Separating them sharpens both the diagnosis and the cure.
Hysteresis vs Dead Band in one line: Hysteresis is a direction-dependent difference in the output reading for the same input value, depending on whether the input was approached from above or below. Dead band is a range of input change over which the output does not respond at all. In short, hysteresis gives two different outputs for one input depending on history, while dead band gives no output change for a small input change in either direction.
Hysteresis is about path. If you slowly raise an input to a given value and note the output, then approach that same value while lowering the input, hysteresis is present when the two outputs differ. The instrument's reading depends not just on where the input is now but on where it came from. Plotted as output against a rising-then-falling input, hysteresis shows up as a loop: the up-path and the down-path trace different curves, and the gap between them is the hysteresis.
Dead band is about a zone of insensitivity. It is a span of input change over which nothing happens at the output at all. Push the input a little and, until you have moved past the dead band, the output sits still; only when the input change exceeds the dead band width does the output finally begin to move. Dead band is most visible on a reversal, when the input must first cross the entire insensitive zone before the output starts responding in the new direction.
The two often coexist and can be hard to separate by eye, because both produce a lag around reversals. The distinguishing test is what happens with the same input approached from two directions versus what happens to a small input change. Hysteresis yields two different but definite outputs for one input value; dead band yields no output change until a threshold of movement is exceeded. Mechanical backlash in linkages and gears is a common source of dead band, while elastic and frictional effects in sensing elements tend to produce hysteresis.
A rising-versus-falling calibration test is the practical way to expose both. You sweep the input up through the range recording outputs, then sweep it back down through the same points, and compare. If the down-sweep readings differ from the up-sweep readings at the same input values, that difference is hysteresis, and its largest gap over the range is the specified hysteresis figure. A device with negligible hysteresis retraces nearly the same curve in both directions.
A sticky control valve shows the phenomena vividly and combines them. When a valve has stiction, a small change in the controller's demand signal produces no movement of the stem at all until the demand accumulates enough force to break the valve free, which is a dead band. Once it breaks free, the stem may jump, and the relationship between demand and position on the way in differs from the way out, which is a hysteresis loop. To an operator watching a trend, the valve seems to ignore small corrections and then lurch, and the process oscillates as the controller keeps hunting.
The reason to distinguish them is that the fixes differ. Dead band from mechanical backlash is addressed by tightening or replacing worn linkages, gears, or couplings, or by adding a positioner that closes the loop around actual position. Hysteresis rooted in friction or elastic memory in a sensing element is addressed differently, sometimes by servicing or replacing the element, sometimes by accepting it as a bounded error in the budget. Prescribing a linkage repair for what is really sensor hysteresis, or vice versa, wastes a maintenance visit and leaves the real problem in place.
On a SCADA trend, dead band and hysteresis both show up as a control point that fails to track small setpoint changes and reacts sluggishly around reversals, often accompanied by a slow limit-cycle oscillation as the loop overshoots and corrects. Because Merobix historizes the demand signal and the measured position or process value together, an engineer can overlay the two and see the characteristic lag: the response that only begins after the demand has moved a certain amount points to dead band, while a response that traces different paths up and down points to hysteresis. Trending the pair turns a vague complaint about a hunting loop into a specific mechanical diagnosis.
There is a naming collision worth heading off, because the word deadband appears in a completely different SCADA context. An alarm deadband, or a reporting deadband on a historized tag, is an intentional configuration value: a threshold that a signal must move by before a new alarm state is declared or a new value is logged, deliberately introduced to suppress chatter from noise. That is a designed software setting, not a physical instrument defect, and it has nothing to do with the mechanical dead band of a valve or linkage.
Keeping the two ideas separate avoids a real confusion in operations. If a loop hunts, the cause could be a physical dead band in the valve, which needs a mechanical fix, or it could be an alarm or reporting deadband set too tight or too loose in the platform configuration, which needs a settings change. Diagnosing one when the other is at fault is a familiar dead end. Documenting which deadband is which for each point, the physical instrument characteristic versus the configured suppression threshold, keeps the troubleshooting on the right track.
No. Dead band is a range of input change over which the output does not respond at all, while hysteresis is a direction-dependent difference in the output for the same input value. They look similar on a trend because both cause lag around reversals, but the distinguishing test is that hysteresis gives two definite outputs for one input depending on approach direction, whereas dead band gives no output change until a movement threshold is crossed.
Mechanical dead band commonly comes from backlash in linkages, gears, and couplings, and from stiction in the valve stem where friction must be overcome before the stem moves. A small change in the demand signal produces no stem movement until enough force accumulates to break the valve free. Adding a positioner that closes a loop around actual stem position is a common way to compensate.
An alarm deadband is an intentional software setting: a threshold a signal must cross before a new alarm state is declared or a new value logged, added to suppress noise-driven chatter. Instrument dead band is a physical defect where the output does not respond to small input changes. One is a configured suppression value, the other a mechanical characteristic, and confusing them leads to fixing the wrong thing when a loop misbehaves.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.