Configuring Ignition Tag History Deadband Style for Step Trends

When an Ignition report chart draws sloped lines between logged values, the tag history is being handled as analog. The sloped line appears even when Sample Mode is set to on-change, and you usually want a stepped trace that holds each value until the next change. Float and Double tags default to the Analog Deadband Style, and the historian and chart interpolate between analog points. Change the Deadband Style on the tag to Discrete, then check the trend on data that was recorded after the change.

What is the chart actually showing you?

Look at the trend first. Compare the sloped trace with the value the tag held at each moment. On-change logging stores a row only when the value moves by more than the deadband. With Analog style, the chart joins two stored points with a straight line. For a value that jumped instantly, such as a setpoint, a mode value, or a recipe parameter, that line shows intermediate values the tag never held. With Discrete style, the chart holds the last stored value flat and draws a vertical edge when the next row arrives. This is the stepped trace most engineers expect from a setpoint.

"Point to point" can mean either shape, so settle which one you want before you change anything:

• Stepped (sample-and-hold): the tag represents a value that changes instantly, such as a setpoint, count, state, or operator entry. Use Discrete.
• Straight lines between samples: the tag represents a continuous process variable, such as a temperature, pressure, or level. Keep Analog, which is the correct model for it.

Why does an on-change Float tag trend as a ramp?

Sample Mode and Deadband Style are separate settings that do different jobs. Sample Mode decides when the historian evaluates the tag. Deadband Style decides how the historian filters stored values and how queries reconstruct the values between stored rows.

The Deadband Style property has three options:

• Auto resolves to Analog for Float and Double data types and to Discrete for the other types.
• Analog treats the value as continuous. Its deadband compression evaluates the shape of the trend, and retrieval interpolates linearly between stored points.
• Discrete stores a value when it leaves the deadband around the last stored value. Retrieval holds that value until the next stored row.

A Float setpoint therefore inherits analog behaviour by default, so an on-change setting alone does not produce steps. Tuning does not fix a model mismatch. The chart renders what the query returns, and the query returns what the Deadband Style tells it to reconstruct.

Which signal in the chain is producing the wrong shape?

Trace the signal from the tag through the historian and the report query to the chart before you change any settings.

How do you switch the tag to discrete trending?

1. In the Designer Tag Browser, open the tag editor for the Float or Double tag that should trend as steps.
2. Under History, confirm History Enabled is true and the correct storage provider is selected.
3. Change Deadband Style from Auto or Analog to Discrete.
4. Check the Deadband and Deadband Mode settings. For a setpoint, use a deadband of zero or one smaller than the smallest meaningful change, so every real change is stored.
5. Keep Sample Mode set to on-change for event-type values. Set a maximum time between samples if you need a periodic confirmation point during long flat periods.
6. Save the tag. Apply the same change to any UDT definition that drives the tag, or a later UDT edit will overwrite the instance setting.
7. In the report, open the Tag Historian Query data source. Use raw or as-stored retrieval instead of an interval with an average aggregation, because averaging across a step recreates a ramp.

How do you verify the trend is correct?

1. Write a known sequence to the tag, for example three distinct values a few minutes apart, and record the times.
2. Preview the report over a time range that contains only data recorded after the setting change.
3. Each value should run flat and change with a vertical edge at the recorded write time. There should be no diagonal between two different values.
4. Compare against a Designer Easy Chart or Power Chart on the same tag. If the Designer chart shows steps but the report does not, the problem is in the report query or chart settings, not the tag.
5. Check that the number of stored rows matches the number of writes. Too few rows means the deadband or minimum-time setting is filtering changes.

What goes wrong after the change?

• Old data still ramps. Test on data recorded after the edit before you conclude the change failed. Trending history that spans the change can mix the old and new behaviour.
• Discrete applied to real process variables. A temperature logged as discrete with a coarse deadband shows false plateaus and sudden jumps. Apply Discrete only to values that actually change in steps.
• UDT overrides. An instance-level change silently reverts when the parent definition is edited. Make the change in the definition.
• Report aggregation. Interval sampling with averaging makes steps look like ramps regardless of the storage style. Check the query before you blame the tag.
• Renderer-only fixes. Changing the chart's line style hides the symptom in one report. Other trends and history queries still interpolate until you change the tag setting.

FAQ