Deadband Filtering

This tutorial shows you how to use deadband filtering to reduce the number of notifications from monitored variables. Deadband filtering only sends notifications when the value changes by a significant amount, reducing network traffic and processing overhead.

Deadband filtering is essential for:

• Analog values with minor fluctuations (temperature, pressure, flow)
• High-frequency sampling where small changes aren't significant
• Reducing network bandwidth usage
• Preventing notification storms

Prerequisites

• An active connection to an OPC UA server (see Create a connection)
• A subscription configured in the connection endpoint
• Basic understanding of monitoring (see Monitoring a Single Variable)

What is Deadband?

A deadband defines a threshold below which value changes are not reported. Only when the value changes by more than the deadband threshold will a notification be sent.

Without Deadband:

Time:  0s    1s    2s    3s    4s    5s
Value: 23.5  23.6  23.4  23.7  23.5  23.6
       ↓     ↓     ↓     ↓     ↓     ↓
       All changes reported (6 notifications)

With Absolute Deadband (0.5):

Time:  0s    1s    2s    3s    4s    5s
Value: 23.5  23.6  23.4  23.7  23.5  23.6
       ↓     --    --    --    --    --
       Only first reported (1 notification)

Deadband Types

None (No Filtering)

Default behavior - all value changes are reported.

{
    "deadbandType": "None"
}

Absolute Deadband

Reports changes only when the absolute difference exceeds the threshold.

Formula: |NewValue - LastReportedValue| > DeadbandValue

Example: Absolute deadband of 0.5

• Last reported: 23.5
• New value: 23.8 → Difference: 0.3 → Not reported
• New value: 24.1 → Difference: 0.6 → Reported

{
    "deadbandType": "Absolute",
    "deadbandValue": 0.5
}

Percent Deadband

Reports changes only when the percentage change exceeds the threshold, based on the variable's EURange (Engineering Units Range).

Formula: |NewValue - LastReportedValue| / EURange * 100 > DeadbandValue

Requirements:

• The monitored variable must have an EURange property
• EURange defines the instrument's measurement range

Example: Percent deadband of 1% with EURange [0, 100]

• EURange span: 100
• Deadband threshold: 1 (1% of 100)
• Last reported: 50
• New value: 50.5 → Difference: 0.5 → Not reported
• New value: 51.2 → Difference: 1.2 → Reported

{
    "deadbandType": "Percent",
    "deadbandValue": 1.0
}

Configuration Methods

Method 1: Node Configuration

Set deadband in the Monitor node configuration panel:

1. Open Monitor node settings
2. Locate Deadband section
3. Select deadband type: Not Use, Absolute, or Percent
4. Enter Deadband Value

Method 2: Message Injection

Override deadband settings dynamically via message properties:

{
    "topic": "ns=1;s=Temperature",
    "deadbandType": "Absolute",
    "deadbandValue": 0.5
}

Overridable Properties:

• msg.deadbandType: "None", "Absolute", or "Percent"
• msg.deadbandValue: Numeric threshold value

Example: Temperature Monitoring with Absolute Deadband

Monitor a temperature sensor that fluctuates slightly:

[
    {
        "id": "inject1",
        "type": "inject",
        "name": "Monitor Temperature",
        "topic": "ns=1;s=Temperature",
        "payload": {
            "deadbandType": "Absolute",
            "deadbandValue": 0.5
        },
        "payloadType": "json"
    },
    {
        "id": "monitor1",
        "type": "OpcUa-Monitor",
        "name": "Filtered Temperature",
        "endpoint": "opcua_endpoint1",
        "subscription": "subscription1",
        "startImmediately": false,
        "samplingInterval": 100
    },
    {
        "id": "debug1",
        "type": "debug",
        "name": "Significant Changes Only"
    }
]

Behavior:

• Server samples every 100ms
• Only reports when temperature changes by ≥ 0.5°C
• Reduces notifications significantly

Example: Pressure Monitoring with Percent Deadband

Monitor a pressure transmitter with EURange [0, 10] bar:

[
    {
        "id": "inject1",
        "type": "inject",
        "name": "Monitor Pressure",
        "topic": "ns=1;s=Pressure",
        "payload": {
            "deadbandType": "Percent",
            "deadbandValue": 2.0
        },
        "payloadType": "json"
    },
    {
        "id": "monitor1",
        "type": "OpcUa-Monitor",
        "name": "Filtered Pressure",
        "endpoint": "opcua_endpoint1",
        "subscription": "subscription1",
        "startImmediately": false,
        "samplingInterval": 500
    },
    {
        "id": "debug1",
        "type": "debug",
        "name": "Significant Changes Only"
    }
]

Calculation:

• EURange: [0, 10] → Span: 10
• Deadband: 2% → Threshold: 0.2 bar
• Reports only when pressure changes by ≥ 0.2 bar

Dynamic Deadband Adjustment

Adjust deadband based on operating conditions:

// Function node: Adaptive deadband
const operatingMode = flow.get("mode");
let deadbandValue;

if (operatingMode === "startup") {
    // More sensitive during startup
    deadbandValue = 0.1;
} else if (operatingMode === "normal") {
    // Less sensitive during normal operation
    deadbandValue = 0.5;
} else if (operatingMode === "shutdown") {
    // Very sensitive during shutdown
    deadbandValue = 0.05;
}

msg.deadbandType = "Absolute";
msg.deadbandValue = deadbandValue;

return msg;

Choosing Deadband Values

Absolute Deadband Guidelines

Temperature (°C):

• High precision: 0.1 - 0.5
• Normal: 0.5 - 2.0
• Low precision: 2.0 - 5.0

Pressure (bar):

• High precision: 0.01 - 0.1
• Normal: 0.1 - 0.5
• Low precision: 0.5 - 1.0

Flow Rate (L/min):

• High precision: 0.5 - 2.0
• Normal: 2.0 - 10.0
• Low precision: 10.0 - 50.0

Percent Deadband Guidelines

General Purpose:

• High sensitivity: 0.1 - 1.0%
• Normal: 1.0 - 5.0%
• Low sensitivity: 5.0 - 10.0%

By Application:

• Critical control loops: 0.1 - 0.5%
• Monitoring: 1.0 - 2.0%
• Historical logging: 2.0 - 5.0%

Combining Deadband with Sampling Interval

Use both features together for optimal efficiency:

{
    "samplingInterval": 1000,
    "deadbandType": "Absolute",
    "deadbandValue": 0.5
}

Process:

1. Server samples every 1000ms
2. Server compares to last reported value
3. Server reports only if difference > 0.5

This reduces both server-side processing and network traffic.

Example: Multi-Variable with Different Deadbands

Apply different deadbands to different variables:

// Function node: Configure monitoring with variable-specific deadbands
const variables = {
    temperature: {
        nodeId: "ns=1;s=Temperature",
        deadbandType: "Absolute",
        deadbandValue: 0.5,
        samplingInterval: 1000
    },
    pressure: {
        nodeId: "ns=1;s=Pressure",
        deadbandType: "Percent",
        deadbandValue: 2.0,
        samplingInterval: 500
    },
    flowRate: {
        nodeId: "ns=1;s=FlowRate",
        deadbandType: "Absolute",
        deadbandValue: 5.0,
        samplingInterval: 2000
    }
};

// Create separate messages for each variable
const messages = [];
for (const [name, config] of Object.entries(variables)) {
    messages.push({
        topic: config.nodeId,
        samplingInterval: config.samplingInterval,
        deadbandType: config.deadbandType,
        deadbandValue: config.deadbandValue
    });
}

// Send to multiple Monitor nodes
return [messages];

Monitoring Deadband Effectiveness

Track how much deadband filtering reduces notifications:

// Function node: Count filtered vs reported
const context = flow.get('deadbandStats') || {
    sampled: 0,
    reported: 0,
    lastValue: null
};

context.sampled++;

// This message represents a reported value (passed deadband)
context.reported++;
context.lastValue = msg.payload;

flow.set('deadbandStats', context);

// Calculate efficiency
const filterRate = ((context.sampled - context.reported) / context.sampled * 100).toFixed(1);

msg.stats = {
    sampled: context.sampled,
    reported: context.reported,
    filtered: context.sampled - context.reported,
    filterRate: filterRate + '%'
};

return msg;

Deadband vs. Rate of Change

Deadband is different from rate-of-change filtering:

Deadband:

• Filters based on magnitude of change
• Independent of time
• Reduces notifications during stable periods

Rate Limiting:

• Filters based on frequency of notifications
• Independent of magnitude
• Ensures maximum notification rate

Consider combining both for comprehensive filtering.

Tips & Best Practices

Start Conservative

Begin with smaller deadband values and increase gradually:

// Start with 1% and adjust based on results
msg.deadbandType = "Percent";
msg.deadbandValue = 1.0;

Use Absolute for Direct Measurements

Use absolute deadband when you know the acceptable tolerance:

// Temperature must be within ±0.5°C
msg.deadbandType = "Absolute";
msg.deadbandValue = 0.5;

Use Percent for Scaled Measurements

Use percent deadband for variables with EURange:

// 2% of instrument range
msg.deadbandType = "Percent";
msg.deadbandValue = 2.0;

Verify EURange for Percent Deadband

Check if the variable has EURange before using percent deadband:

// Read EURange first
const readMsg = {
    topic: "ns=1;s=Pressure.EURange",
    attributeId: "Value"
};

// If EURange exists, use Percent; otherwise, use Absolute

Consider Instrument Accuracy

Set deadband larger than instrument noise:

// If sensor accuracy is ±0.1, use deadband ≥ 0.2
msg.deadbandValue = 0.3;

Monitor Performance

Regularly check notification reduction:

// Log statistics periodically
setInterval(() => {
    const stats = flow.get('deadbandStats');
    node.log(`Filter effectiveness: ${stats.filterRate}`);
}, 60000); // Every minute

Common Pitfalls

Deadband Too Large

Symptom: Missing important changes

Solution: Reduce deadband value

// Was: 5.0 (too large)
// Now: 1.0 (better)
msg.deadbandValue = 1.0;

Deadband Too Small

Symptom: Still getting too many notifications

Solution: Increase deadband value

// Was: 0.01 (too small)
// Now: 0.5 (better)
msg.deadbandValue = 0.5;

Percent Deadband Without EURange

Symptom: Deadband doesn't work

Solution: Use absolute deadband or add EURange to variable

// If EURange not available, use Absolute
msg.deadbandType = "Absolute";

Conflicting with Alarm Thresholds

Symptom: Alarms not triggered

Solution: Ensure deadband is smaller than alarm thresholds

// If alarm at 80°C, use deadband < 1°C
msg.deadbandValue = 0.5;

Troubleshooting

No Notifications Received

Cause: Deadband too large or value not changing enough

Solution:

1. Reduce deadband value
2. Check if value is actually changing
3. Verify monitored variable is correct

Too Many Notifications

Cause: Deadband too small or not configured

Solution:

1. Increase deadband value
2. Verify deadband is being applied
3. Check sampling interval

Percent Deadband Not Working

Cause: Variable doesn't have EURange property

Solution:

1. Verify EURange exists using Read node
2. Switch to absolute deadband
3. Contact server administrator to add EURange

Unexpected Notification Timing

Cause: Misunderstanding deadband behavior

Clarification:

• Deadband compares to last reported value, not last sampled value
• Initial value is always reported regardless of deadband

Performance Impact

Notification Reduction

Typical reductions with proper deadband configuration:

• No deadband: 100% of samples reported
• Small deadband (0.5%): 60-80% reduction
• Medium deadband (2%): 80-95% reduction
• Large deadband (5%): 95-99% reduction

Server-Side Filtering

Deadband is processed by the OPC UA server, reducing:

• Network bandwidth usage
• Client-side processing
• Message queue size

Best Practice

Deadband + Sampling Interval = Maximum efficiency

{
    "samplingInterval": 1000,
    "deadbandType": "Absolute",
    "deadbandValue": 0.5
}

Next Steps

• Monitor Array - Apply deadband to multiple variables
• Monitor JSON Structure - Deadband with structured data
• Monitor Events - Event-based monitoring

See Also

• OPC UA Specification - Data Change Filter
• OPC UA Specification - Deadband
• EU Information