OPC-UA for Machine Monitoring: What Factory Managers Need to Know

OPC-UA is a protocol that lets monitoring software read data directly from your machines — cycle times, fault codes, spindle speeds, program states, temperatures — without proprietary drivers or middleware. If you’re evaluating machine monitoring platforms, you’ll see OPC-UA mentioned constantly. This post explains what it actually is, what data it gives you, which machines support it, and where the gaps are.

What OPC-UA actually is

OPC-UA (Open Platform Communications Unified Architecture) is an open standard for machine-to-machine communication, maintained by the OPC Foundation. It replaced the older OPC DA standard, which was locked to Windows. OPC-UA is platform-independent — it runs on Linux, embedded controllers, and cloud servers, not just Windows PCs.

The standard is published as IEC 62541 and has been adopted globally. It defines how machines expose their data as a structured address space — a tree of nodes that a client application can browse, read, and subscribe to. Each node has a unique identifier, a data type, and a value that updates in real time.

In practical terms: if your CNC controller or robot has an OPC-UA server running, a monitoring platform can connect to it over Ethernet, browse the available data points, and begin integrating production data without needing machine-specific protocol development. In many cases this avoids proprietary drivers and reverse engineering, though real-world deployments still often require custom tag mapping, certificate configuration, and testing before data flows reliably.

What data does OPC-UA expose?

This depends entirely on the machine and its controller. OPC-UA is a transport and data modelling standard — it defines how data is structured and transmitted, not what data a machine must provide. That said, most OPC-UA-enabled controllers expose some combination of:

  • Machine state: running, idle, stopped, faulted, in alarm
  • Program information: active program name, program state, cycle count
  • Axis and spindle data: positions, speeds, loads, torque values
  • Alarms and events: fault codes, warning conditions, protective stops
  • Production counters: parts produced, cycle times, reject counts
  • Temperature data: spindle temperature, coolant temperature, zone temperatures (particularly relevant for injection moulding)

The depth varies. A Siemens Sinumerik 840D with the OPC-UA option enabled will expose detailed spindle load, axis positions, active tool data, and program block information. A simpler PLC-based machine might expose only basic run/stop state and a handful of process variables.

The important point for factory managers: OPC-UA gives you the data the machine already knows about itself. You’re not adding sensors or estimating from power draw — you’re reading the same values the machine’s own controller is working with.

Which machines support OPC-UA?

This is where it gets uneven. Some controllers ship with OPC-UA as standard. Others require a paid option. Others don’t support it at all.

CNC controllers:

  • Siemens Sinumerik 840D sl / Sinumerik ONE: OPC-UA server available, but requires a separate licence (Siemens part number 6FC5800-0AP67-0YB0) and SINUMERIK Operate V4.8 or later. Not free, but well-documented and exposes rich CNC data including spindle loads, axis positions, and active program information.
  • Haas NGC: Haas controllers use MTConnect as their primary monitoring protocol, not OPC-UA. MTConnect is a read-only, XML-based standard that’s simpler than OPC-UA but widely supported in the CNC world. If your shop runs Haas machines, your monitoring platform needs to support MTConnect — or you need a gateway.
  • FANUC: FANUC’s CNC controllers use FOCAS (FANUC Open CNC API Specification) as their native data interface. FANUC does offer an OPC Server application (FASOPC) that bridges FOCAS data to OPC, but it’s a paid add-on running on a separate PC. For FANUC robots (as opposed to CNCs), OPC-UA communication conforming to the OPC Robotics companion specification is included as standard on current controllers sold in Europe and the UK — check with your regional FANUC distributor for other markets.
  • DMG MORI: Their CELOS controllers integrate OPC-UA and MTConnect. Coverage is generally good on newer machines.

Industrial robots:

  • ABB: IRC5 and OmniCore controllers don’t use OPC-UA as their primary monitoring interface. ABB provides Robot Web Services (RWS) — a REST API accessible over HTTP. This is what RoboVigil uses for ABB connectivity. OPC-UA is available via add-on options requiring an additional software licence and edge hardware, but RWS is more direct for monitoring purposes.
  • Universal Robots: UR cobots don’t use OPC-UA as standard although there is a third party URCap available at extra cost. But UR robots do provide three native TCP interfaces — RTDE (Real-Time Data Exchange) at up to 500Hz, the Dashboard Server for high-level commands, and the Primary Interface for program and safety state. These are open, well-documented, and free. RoboVigil connects natively to all three at once for a complete dataset.
  • KUKA: KUKA.Connect and OPC-UA options are available on newer controllers.
  • FANUC robots: OPC-UA with the OPC Robotics companion spec is now standard on current controllers in Europe and the UK, exposing robot position, program state, alarms, and I/O signals. Availability may vary by region.

Injection moulding machines:

Injection moulding is one of the sectors where OPC-UA standardisation has seen particularly strong adoption. The EUROMAP 77 companion specification defines a standard OPC-UA interface between injection moulding machines and MES systems. It was developed jointly by Arburg, Engel, KraussMaffei, Netstal, Negri Bossi, Sumitomo (SHI) Demag, and Wittmann Battenfeld. EUROMAP 83 provides the base type definitions.

In practice, this means a modern Engel or Arburg injection moulding machine with EUROMAP 77 support will expose standardised cycle data, shot weights, temperatures, mould information, and production counts over OPC-UA — and a monitoring platform can read them without machine-specific configuration.

PLCs:

  • Siemens S7-1500: OPC-UA server built in as standard from firmware V2.0 onwards.
  • Siemens S7-1200: OPC-UA server available from firmware V4.4.
  • Beckhoff TwinCAT: OPC-UA server available as a standard function.
  • Rockwell / Allen-Bradley: Rockwell’s ecosystem primarily uses EtherNet/IP. OPC-UA support has historically relied on FactoryTalk Linx Gateway or third-party servers like Kepware, though newer controllers increasingly include native OPC-UA capabilities.

Companion specifications — the standardisation layer

Raw OPC-UA gives you a transport mechanism and a data model. Companion specifications add the domain knowledge — they define what a “cycle time” means, what a “machine state” should look like, and what data points should be available for a given type of equipment.

The OPC Foundation lists over 60 companion specifications across different industries. The ones most relevant to factory monitoring:

  • OPC UA for Machine Tools (VDMA 40501-1): Standardises monitoring data for CNC machine tools — job overview, machine state, production statistics. Developed by the VDW (German Machine Tool Builders’ Association) and marketed under the umati brand.
  • OPC UA for Robotics: Defines motion control data, safety monitoring, and coordinate systems for industrial robots.
  • EUROMAP 77 / 83: Injection moulding machine data exchange, as described above.
  • PackML: Packaging machinery states and modes, based on ISA-TR88.
  • PLCopen: PLC program and variable exposure.

The reality on the factory floor is messier than the specification documents suggest. Companion spec adoption is growing but far from universal. Many machines expose OPC-UA data using their own proprietary information models, not the companion spec structure. A monitoring platform has to handle both.

The OPC-UA cost question

OPC-UA is an open standard, but accessing it on your machines isn’t always free.

Machines where OPC-UA is included at no extra cost: Siemens S7-1500 PLCs, FANUC robots (current controllers in Europe/UK), Beckhoff TwinCAT systems, and newer injection moulding machines from the major European builders.

Machines where OPC-UA is a paid option: Siemens Sinumerik CNC controllers (licence required), FANUC CNCs (requires FASOPC or a third-party OPC server), ABB robots (OPC-UA option; though RWS is available without it).

Machines where OPC-UA isn’t the right path at all: Haas CNCs (use MTConnect), Universal Robots (use RTDE/Dashboard/Primary Interface natively), older machines without Ethernet connectivity.

For a factory manager, the practical question isn’t “does my machine support OPC-UA?” — it’s “what’s the most direct way to get data out of this specific machine?” Sometimes that’s OPC-UA. Sometimes it’s a native protocol that’s cheaper, faster to deploy, and exposes richer data.

How RoboVigil uses OPC-UA

RoboVigil connects to OPC-UA servers as a client over your existing factory network. There’s no edge device, no gateway hardware, no SIM card. If your machine has an OPC-UA server accessible on the network, RoboVigil connects to it through a WireGuard VPN tunnel — encrypted, authenticated, and without opening ports on your factory firewall.

OPC-UA has its own security layer, separate from the VPN. Most production OPC-UA servers require certificate-based authentication — the client presents a signed certificate, the server checks it against its trust store, and the session is established with encryption (typically Basic256Sha256). This is good practice, but certificate setup is often the most time-consuming part of an OPC-UA deployment. RoboVigil handles this by generating OPC-UA client certificates directly within the app, so you can create the certificate, export it, and add it to your OPC-UA server’s trust store without needing separate tooling or a PKI infrastructure.

Once connected, Robovigil discovers OPCUA servers on a network and their machine IDs. You choose the machine and RoboVigil then browses the OPC-UA address space. On a real machine, that address space can contain thousands of nodes — a Siemens Sinumerik or a well-configured PLC might expose several thousand endpoints across drive data, PLC variables, diagnostic channels, and internal system parameters. Most of those aren’t useful for monitoring. RoboVigil uses a combination of namespace filters, known protocol maps for common controller types, and AI-driven classification to identify the nodes that matter — machine state, alarms, production counters, fault data — and ignore the noise. If the automatic selection doesn’t precisely match your needs you can manually configure which nodes to monitor and how they map to RoboVigil.

The selected data streams to the cloud in real time. When something goes wrong — a protective stop, a fault code, a machine entering an unexpected state — RoboVigil pushes an alert to your phone with the fault data and, if an IP camera has been added, a snapshot of the machine.

The approach is protocol-agnostic at the platform level. OPC-UA is one of several protocols RoboVigil supports. For Universal Robots, it connects via RTDE and Dashboard Server. For ABB IRC5 robots, it uses Robot Web Services. For machines publishing data over MQTT (including Sparkplug B), it subscribes directly to the broker. For FANUC robots, OPC-UA is the current connector path.

This matters because most factories don’t run a single brand of equipment. A typical shop might have Siemens-controlled CNC machines, a couple of UR cobots, an ABB cell, and some PLC-driven auxiliary equipment. A monitoring platform that only speaks OPC-UA will cover some of those machines. One that speaks OPC-UA, MQTT, RTDE, and RWS covers all of them.

OPC-UA vs the alternatives

OPC-UA isn’t the only way to get data off the factory floor. Here’s where each protocol fits:

OPC-UA — Best for: Siemens PLCs and CNC controllers, injection moulding machines with EUROMAP support, multi-vendor environments where you want one protocol to cover as many machines as possible. Weaknesses: can be expensive to enable on some controllers, overkill for simple use cases, address space browsing can be complex.

MTConnect — Best for: CNC machine shops, especially Haas and Mazak. Read-only, XML-based, simpler than OPC-UA. Weakness: limited outside CNC machining, no built-in security layer.

MQTT — Best for: lightweight telemetry, IoT sensors, edge-to-cloud data streaming, machines with custom PLC programs that publish status to a broker. Weakness: no standard data model — you get whatever the publisher sends. Sparkplug B adds structure but adoption is still patchy.

Native protocols (UR RTDE, ABB RWS, FANUC FOCAS) — Best for: getting the richest possible data from a specific machine brand without paying for OPC-UA add-ons. Weakness: each is brand-specific, so you need a monitoring platform that supports multiple native protocols.

What to ask before you choose a monitoring platform

Rather than asking “does this platform support OPC-UA?” — which many will answer yes to — ask:

  1. What protocols does it support beyond OPC-UA? If your factory runs UR cobots, ABB robots, Haas CNCs, and Siemens-controlled machines, you need more than OPC-UA.
  2. Does it require hardware on my factory floor? Some platforms need edge gateways or cellular modems to bridge the connection. Others connect directly over the network. Hardware adds cost, deployment time, and maintenance.
  3. What happens when my OPC-UA server doesn’t follow a companion spec? Most real-world OPC-UA implementations use proprietary node structures. Can the platform handle custom tag mapping? How long does this take to configure and who does it?
  4. How does it handle OPC-UA security? OPC-UA supports signed and encrypted sessions with certificate-based authentication. Does the platform generate its own client certificates, or do you need to manage that separately? And beyond OPC-UA’s own security model, how does the platform connect your factory to the cloud — VPN tunnels, firewall rules, and transport encryption all matter.
  5. What does it cost per machine, all in? OPC-UA licences on some controllers, edge hardware, platform subscription fees, per-user charges — the total cost can be significantly higher than the platform’s headline price suggests.

Getting started

If you want to see what OPC-UA monitoring looks like in practice, the RoboVigil demo factory runs 22 OPC-UA simulators covering different machine types. Download the app on iOS, Android, or the Amazon Appstore and browse the demo data — machine states, alarms, production metrics, camera feeds — to see what your factory could look like.

If your machines already have OPC-UA servers running, you can connect them to RoboVigil and be monitoring the same day. No dedicated edge hardware or industrial gateway required. £150 per machine per month, everything included.