How We Actually Pick Industrial Protocols for a Brownfield Line

Modbus is dead. Right? Tell that to the 1998 filling line that's still running three shifts a day. Modbus works because it's simple: map registers, move bits. The Modbus Organization calls itself the steward of the world's most widely adopted industrial communication protocol. So the real question isn't which protocol is newest. It's which one gets your line running without a rip-and-replace budget.

This is for controls engineers, integrators, and plant managers who inherited a mix of vendors and now have to make the network talk. We've done this on brownfield lines where a new skid shows up with PROFINET, the old PLC speaks Modbus TCP, and the corporate dashboard wants OPC UA. Here's how we decide.

1. Inventory what's already bolted to the floor

Before any PowerPoint, we walk the line with a clipboard. Every drive, sensor, valve island, and HMI gets a row: vendor, model, existing protocol, and whether it has an Ethernet port. Why? Protocols are layered: Modbus, PROFINET, and EtherNet/IP connect field devices and PLCs, while OPC UA and MQTT carry data securely to SCADA, MES, and cloud systems (OPC Foundation). If 80% of your devices already speak one fieldbus, that protocol wins unless there's a hard technical reason to abandon it. We don't switch a working PROFIBUS segment just because a vendor brochure mentions TSN.

2. Match the protocol to the motion requirement

Not all industrial Ethernet is equal for motion. If you're doing camming and electronic gearing on a packaging line, you need deterministic cycle times. EtherCAT was built for short cycles (≤ 100 µs) and jitter ≤ 1 µs (EtherCAT Technology Group), and it uses distributed clocks to keep axes synchronized. PROFINET offers isochronous real-time (IRT) for motion, while standard real-time (RT) is fine for cell control. Sercos goes further with configurable cycle times from 31.25 µs to 65 ms and synchronization accuracy much less than 1 µs (Sercos International). If your axis is a simple indexing table, you don't need that. Pick the cheapest protocol that meets the worst-case update rate, not the fastest one on the datasheet.

3. Check the installed base before betting on a niche

We look at what the market already runs. PROFIBUS & PROFINET International reports more than 68 million PROFIBUS nodes, over 69 million PROFINET nodes, and more than 51 million IO-Link nodes installed worldwide as of the end of 2023 (PROFIBUS & PROFINET International). That matters for spare parts, training, and finding an electrician at 2 a.m. EtherNet/IP runs CIP over standard Ethernet and dominates Rockwell/Allen-Bradley shops in North America. If your maintenance team knows Studio 5000, forcing them onto a protocol they've never seen is a management failure, not a technical upgrade.

4. Use gateways and native translation instead of replacing devices

This is where we save the most money. CIP provides native translation services for Modbus, HART, and IO-Link devices (ODVA Common Industrial Protocol). That means a new EtherNet/IP PLC can read an old Modbus flow meter without a protocol converter box. We've commissioned lines where the only change was a gateway and a few configuration files. Do not let a vendor sell you a full device replacement when a $400 gateway does the job.

5. Separate the control layer from the information layer

The biggest architectural mistake we see is trying to push production data over the same protocol that runs the servo. Keep them separate. OPC UA is the vendor-neutral, platform-independent standard with built-in encryption and authentication that serves as the backbone for IIoT and IT/OT integration (OPC Foundation). MQTT is standardized as ISO/IEC 20922 and is designed for constrained machine-to-machine environments with small code footprint and minimal bandwidth (ISO/IEC 20922 MQTT Standard). We use OPC UA for controller-to-SCADA and MQTT for cloud telemetry. Mixing them into one flat network invites broadcast storms and security holes.

6. Bake in security from the first cable pull

We treat security as a protocol selection criterion, not a later project. CISA warns that many legacy ICS devices use outdated operating systems and older protocols that lack encryption or authentication, and identifies brownfield deployments as a challenge unique to ICS (CISA Industrial Control Systems). If you're buying new hardware, ask for CIP Security or OPC UA encryption out of the box. ISA/IEC 62443 defines requirements for asset owners, product suppliers, and integrators, so use it as your procurement checklist. A protocol without authentication is a liability you'll pay for later.

7. Plan for TSN without betting the plant on it

Time-Sensitive Networking promises guaranteed packet transport with bounded latency and low jitter through IEEE 802 networks (IEEE 802.1 TSN Task Group). We like it. We also know that most brownfield devices will never support it. Our move is to specify TSN-capable switches on new projects so the infrastructure is ready, but keep the device layer on proven protocols. You can migrate later without re-pulling cable.

What can go wrong: the hidden gateway tax

Gateways are not free. We once added three Modbus-to-EtherNet/IP gateways to a line and spent two weeks chasing a timing issue because the gateway added 20 ms of latency to a safety-related interlock. It worked in the lab. It tripped the line during commissioning. The lesson: never put a gateway in a safety loop unless the gateway itself is certified for that function. Use Safety over EtherCAT (FSoE) or CIP Safety where the standard supports it, and keep safety on dedicated hardware. If a vendor says their gateway is "transparent," ask for the jitter specification in writing.

Bottom line

The single best move is to inventory your installed devices and pick the protocol that already has the largest footprint on your floor, then use native translation and gateways to bridge the rest. Do not replace working hardware to chase a trend. The global industrial automation market was valued at about $184.43 billion in 2025 and is projected to reach $326.48 billion by 2032 (Maximize Market Research), so vendors will keep pushing new protocols. Your job is to keep the line running, not to win a standards argument.

Sources

• OPC Foundation - https://opcfoundation.org/
• PROFIBUS & PROFINET International (PI) - https://www.profibus.com/technology/
• ODVA Common Industrial Protocol - https://www.odva.org/technology-standards/key-technologies/common-industrial-protocol-cip/
• CISA Industrial Control Systems - https://www.cisa.gov/topics/industrial-control-systems
• IEEE 802.1 TSN Task Group - https://www.ieee802.org/1/pages/tsn.html
• Maximize Market Research - https://www.maximizemarketresearch.com/