Deadband Filter

The Deadband filter rejects Data Point snapshots where the change from the previous value is smaller than a configured threshold. It computes |new_value − previous_value| and discards the snapshot if the change falls below the deadband value — suppressing noise from sensors by requiring a minimum change magnitude between consecutive readings.

Info: This is a Data Point Filter — a leaf node that sits under a Data Point’s filter chain and gates incoming snapshots before they are stored.

Overview

Analog sensors jitter. A temperature probe that reads 22.0 °C will happily report 22.01, 21.99, 22.02 between meaningful changes, filling your history with noise and triggering downstream automation for movements that don’t matter. The Deadband filter draws a band around the last accepted value and ignores anything that stays inside it.

How It Works

graph TD
    A[New snapshot arrives] --> B[Compute change = abs new − previous]
    B --> C{change >= deadband?}
    C -->|Yes| D[Accept snapshot, store, propagate]
    C -->|No| E[Discard snapshot]
    D --> F[previous = new value]

1. A new snapshot arrives at the parent Data Point.
2. The filter computes the absolute difference between the new value and the previously accepted value.
3. If the change is greater than or equal to the deadband threshold, the snapshot is accepted, stored, and propagated to child triggers and executors.
4. If the change is smaller, the snapshot is discarded and never stored.

Configuration

Field	Description	Required
deadband	Minimum change magnitude (Double) required to accept a new snapshot	Yes

Use Cases

• Reduce noisy sensor data — only persist readings that represent real movement.
• Suppress jitter in analog readings such as temperature, pressure, or voltage.
• Only log meaningful value changes, keeping your time-series history compact.

Examples

• Only record temperature changes of at least 0.5 degrees.
• Ignore pressure changes smaller than 1 PSI.
• Suppress sensor noise — only log if the value changes by 2 or more.

Related

• Data Points
• Data Point Filters
• Debounce Filter

---

Last verified: 2026-05-21