Device Connectivity
EmberNet talks to industrial equipment over the protocols that equipment already speaks. This page covers what's supported, how to register a device, and the handful of choices that matter when you do.
Protocols
OPC UA
The primary protocol for SCADA and automation systems: PLCs, DCS, and HMI integration.
Servers on the local network are discovered automatically, and you browse the server's address space to pick the tags you want rather than typing addresses. Where the server supports subscriptions, use them: the server pushes on change instead of answering a poll every second, which is easier on both sides.
Security modes and policies are fully supported, and historical reads work against OPC UA historians.
MQTT
Lightweight pub/sub for IoT sensors, gateways, and custom edge applications.
You can connect to an external broker or use AnvilMQ, the platform's own. Topics and payloads map onto EmberNet metrics, QoS 0, 1, and 2 are all available, and Sparkplug B payloads are decoded natively.
Use TLS on 8883. A broker reachable in the clear on a plant network is worth avoiding.
Modbus TCP/RTU
The workhorse for PLCs, RTUs, power meters, motor drives, and VFDs. TCP over Ethernet and RTU over RS-232 or RS-485 are both supported.
You configure coils, discrete inputs, input registers, and holding registers directly, with integer, float, string, and bit-level types. One connection can poll several devices.
Modbus has no concept of data typing on the wire, so register mapping is where mistakes hide. Verify a few known values against the device before trusting a map.
SNMP
Network switches, UPS systems, managed PDUs, and environmental sensors. Both v2c and v3 are supported, along with custom MIBs, trap reception, GETBULK for efficient polling, and network discovery.
Setting up the device side is covered in SNMP Configuration.
BACnet/IP
Building systems: HVAC, lighting, access control, and fire panels. Objects are discovered automatically, change-of-value subscriptions avoid needless polling, and writes support the BACnet priority array.
Beyond these five
Device Monitor collects metrics over eleven further industrial protocols, including EtherNet/IP, PROFINET, DNP3, S7comm, FINS, ADS, and IEC 61850.
Registering a device
From Device Monitor:
- Add the device and choose its protocol
- Enter the connection details: address, port, credentials
- Browse or define the data points to collect
- Set the polling interval and data mapping
- Test the connection
- Save and activate
Test before you activate. A device that saves cleanly and then never reports is harder to diagnose than one that failed its test with a specific error.
Organizing devices
Group devices so configuration and alerting apply to sets rather than individuals. The groupings that hold up over time are by location, by equipment type, and by tenant or business unit.
Device status
| Status | Meaning |
|---|---|
| 🟢 Online | Connected and transmitting |
| 🟡 Degraded | Connected, but some data points are unavailable |
| 🔴 Offline | Unreachable |
| ⚪ Disabled | Monitoring deliberately paused |
Degraded is the one to pay attention to. It usually means a tag mapping points at something that no longer exists, which is what happens after somebody changes a PLC program.
Mapping tags
| Field | What it holds |
|---|---|
| Source | The protocol-specific address: OPC UA node, Modbus register, SNMP OID |
| Name | A human-readable metric name |
| Unit | Engineering unit, such as °F, PSI, or kW |
| Data type | Numeric, boolean, string, or enumeration |
| Scaling | Optional linear scaling, slope and offset |
| Tags | Metadata for filtering and grouping |
Every reading also carries a quality flag: Good for a valid reading, Uncertain where the device reports doubt about its own value, and Bad where no valid reading was available. Uncertain readings are worth surfacing rather than filtering out, because a sensor that knows it's drifting is telling you something.
Practical advice
Give industrial devices static addresses or DHCP reservations. A device whose address moves looks exactly like a device that failed.
Turn on encryption wherever the protocol supports it: OPC UA security policies, MQTT over TLS, SNMPv3.
Pick polling intervals deliberately. Faster is not better; every poll costs the device something, and older equipment can be knocked over by an aggressive schedule. Where subscriptions exist, prefer them to polling entirely.
Name tags consistently from the start. Renaming a tag after a year of history is not a pleasant operation.
Next steps
- Device Monitor: the registry and health view
- Metrics & Monitoring: seeing the data
- Alerts & Notifications: alerting on it