DeviceNet

What is DeviceNet?

DeviceNet is a digital fieldbus protocol that carries the Common Industrial Protocol (CIP) over a CAN bus, so that sensors, actuators, motor starters, drives and I/O modules can communicate with a PLC or other controller over a single cable that carries both data and 24 V power. The protocol was introduced in 1994 by Allen-Bradley, now part of Rockwell Automation, and has been managed since 1995 by ODVA, originally the Open DeviceNet Vendors Association. DeviceNet is standardised internationally in IEC 62026-3 (current edition 2014, adopted in Europe as EN 62026-3) and runs in thousands of machines and conveyor systems, particularly where Rockwell controllers are used. In new installations it has largely been replaced by IP, but the installed base remains large.

🧠 What does DeviceNet do?

DeviceNet connects the field components at the bottom layer of a machine to the controller. Instead of wiring every sensor to an I/O card with its own conductor, all devices hang off one main cable (the trunk) with short branches (drops). The result is less cabling, faster installation and per-device diagnostics.

Its main functions:

• One cable for data and power - round DeviceNet cable has five conductors: a signal pair, a power pair for 24 V DC and a drain (shield) wire
• Cyclic I/O exchange - inputs and outputs are refreshed in real time, with priority settled by CAN arbitration
• Configuration and diagnostics over the network - parameters and status information can be read without extra wiring
• Controller-to-device and peer-to-peer - besides the classic master/slave model, devices can also exchange data directly
• Address checking - at start-up each node checks whether its address (MAC ID) is already in use, so duplicate addresses do not go unnoticed

Because the application layer is CIP, DeviceNet uses the same object model as IP. An engineer who knows CIP configures devices on both networks in the same way, and a controller can route messages between the two. If you want to use DeviceNet data higher up in the organisation, for example in an MES or historian, you usually expose it through the controller or an edge gateway with OPC UA, because DeviceNet itself lives only at machine level.

🔧 How does DeviceNet work technically?

DeviceNet has a layered structure. The physical and data link layers come from Classical CAN as defined in ISO 11898; the upper layers are CIP.

Every device has a MAC ID from 0 to 63, so a network holds at most 64 nodes, including the scanner. New devices usually ship set to address 63. The MAC ID is embedded in the CAN identifier, so CAN arbitration automatically decides which message wins priority.

DeviceNet has two kinds of messages:

• I/O (implicit) messages - time-critical process data without protocol overhead. There are four variants: polled (the scanner queries each device in turn), bit-strobe (one 8-byte broadcast message with one bit per node, to which the devices configured for it respond), change of state (a device transmits only when something changes) and cyclic (transmission at a fixed interval).
• Explicit messages - request/response messages for configuration, parameters and diagnostics that are not time-critical.

Most simple devices use the Predefined Master/Slave Connection Set, a simplified set of connections in which each device belongs to one master. Devices that support the Unconnected Message Manager (UCMM) can additionally set up connections dynamically, for example for peer-to-peer traffic.

📏 Which cable lengths and power values apply?

The chosen baud rate determines the maximum length. The higher the speed, the shorter the network, because every CAN bit has to reach the whole network within one bit time for arbitration to work.

Typical design values from the ODVA and Rockwell installation guidelines:

• Termination - a 121 Ω resistor (1%, at least 1/4 W) at each end of the trunk, across CAN_H (white) and CAN_L (blue). A measurement between those two conductors should read roughly 50–60 Ω.
• Drops - at most 6 m per drop, and the sum of all drops must not exceed the value in the table.
• Power - a 24 V DC supply; the voltage at every node must stay between 11 and 25 V. A thick trunk may carry 8 A per segment, a thin trunk about 3 A, and a 6 m drop only 0.75 A.
• Grounding - the shield and the V− conductor are grounded at one point only, usually at the power supply.

Example: a packaging line with 30 devices and 40 m of drops cannot run at 500 kbit/s, because only 39 m of drops is allowed at that rate. At 250 kbit/s the same line fits.

🏭 Where is DeviceNet used?

DeviceNet is found mainly in discrete manufacturing and machine building:

• Automotive - body shops and assembly lines with many pneumatic valves and proximity switches
• Packaging machines - communication with motors, sensors and valve terminals
• Conveyor systems - I/O control of conveyors, sorters and warehouse systems
• Motor control - motor starters, variable frequency drives and control cabinets where DeviceNet replaces conventional wiring
• Machine safety - with CIP Safety on DeviceNet (also called DeviceNet Safety), signals from emergency stops and light curtains up to SIL 3 according to IEC 61508 travel over the same network; the first products appeared in 2005

In Rockwell environments a scanner module, such as the 1756-DNB for ControlLogix, connects the network to the controller. Configuration is done with RSNetWorx for DeviceNet and the I/O is used in Studio 5000. Vendors such as Omron also offer DeviceNet devices.

🔄 How does DeviceNet compare with other fieldbuses?

DeviceNet and CANopen share the same CAN layer, but their application protocols are not interchangeable. Their Ethernet successors are IP and ProfiNET respectively. See also the overview of fieldbuses.

⚠️ What are the limitations of DeviceNet?

• Low speed - at most 500 kbit/s and 8 data bytes per CAN frame; for motion control and large data volumes, industrial Ethernet based on Ethernet is a better fit
• Limited cable length - at 500 kbit/s a maximum of 100 m with thick cable
• Linear topology - trunk with drops; no star or ring structure without extra hardware
• Sensitive to installation errors - missing terminating resistors, overlong drops and voltage drop are the classic causes of faults
• Phase-out - vendors are focusing on IP, so new DeviceNet devices are becoming scarcer and spare parts deserve attention in lifecycle management

🔐 How do you secure a DeviceNet network?

DeviceNet has no authentication or encryption: any device on the bus can send messages and change parameters. CIP Security, the ODVA extension with TLS and DTLS, is defined for EtherNet/IP and not for DeviceNet. Protection therefore has to come from the surroundings:

1. Inventory - include DeviceNet segments, scanners and gateways in your asset inventory; they are invisible to IP-level network scans.
2. Shield the gateways - the real attack path runs through the controller or an EtherNet/IP-to-DeviceNet gateway, because CIP can route messages between networks. Place these in a separate zone following the zones and conduits model of IEC 62443.
3. Restrict physical access - the bus runs through cabinets and machines; locked cabinets and controlled service ports stop anyone from plugging in a laptop with a CAN interface.
4. Segment and monitor - apply network segmentation to the EtherNet/IP network above and monitor changes to controller programs.

For Dutch organisations that fall under the Cybersecurity Act, in force since 15 August 2026, these legacy systems also belong in the risk assessment.

❓ Frequently asked questions

Is DeviceNet the same as CAN?

No, DeviceNet uses CAN only for the physical and data link layers. On top of that, DeviceNet adds the Common Industrial Protocol, device profiles, addressing and power specifications. Every DeviceNet device is therefore a CAN device, but an arbitrary CAN device does not speak DeviceNet.

How many devices can be on one DeviceNet network?

A DeviceNet network supports at most 64 nodes with MAC IDs 0 to 63, including the scanner. If you need more devices, you use a second scanner and a second network. In practice the supply current and the total drop length also limit the number of devices.

Is DeviceNet still used?

Yes, DeviceNet still runs in many existing machines and production lines, especially in the automotive and packaging industries. ODVA continues to maintain the specification and conformance testing. For new projects, however, vendors and end users almost always choose EtherNet/IP.

How do you migrate from DeviceNet to EtherNet/IP?

A common approach is phased: first a gateway or linking device that makes DeviceNet devices reachable over EtherNet/IP, then replacing devices during maintenance or after failures. Because both networks use CIP, the object model and many parameters stay the same. Plan the migration together with the replacement of scanners and controllers, and test the I/O mapping in advance.

What is the difference between thick, thin and flat DeviceNet cable?

Thick round cable is intended for long trunks up to 500 m and may carry up to 8 A. Thin round cable is more flexible and is used for drops and short trunks up to 100 m. Flat cable has four unshielded conductors and is mounted quickly with insulation-displacement taps.

Can DeviceNet perform safety functions?

Yes, with CIP Safety, also known as DeviceNet Safety, emergency stops, light curtains and safety controllers up to SIL 3 can communicate over the same DeviceNet network. The protocol treats the network as a black channel and detects errors in the end devices themselves. This does require certified safety devices and a safety controller.

📌 In summary

In the 1990s DeviceNet brought the CIP object model to the CAN bus and made it possible to connect dozens of field devices with one cable for data and power. The specification sits with ODVA, the standard is IEC 62026-3 and the limits are clear: 64 nodes, 125–500 kbit/s and a maximum of 500 m. For owners of existing installations the focus is now on correct maintenance, spare parts, securing gateways and a well-planned migration to EtherNet/IP.

DeviceNet is ODVA’s CAN-based CIP fieldbus that connects sensors, actuators and drives to a controller with one cable for data and power, and that will remain in service alongside EtherNet/IP in many plants for years to come.