How to Add OEE Monitoring to Old Machines Without Replacing Them

Old machines are not the problem

The most common reason plants give for not measuring OEE is that their machines are too old. In reality, almost any machine that runs, stops and produces parts can be monitored. Before moving into software I spent seven years installing, commissioning and maintaining instrumentation and control systems on petrochemical and FMCG plants, including Siemens S5 to S7 migrations. The lesson from that work is simple: you rarely need to change the machine. You need to pick the right signal.

Option 1: add a simple sensor

For machines with no usable controller, a small retrofit sensor is the fastest route. It does not touch the machine's control circuit at all.

• A proximity or photoelectric sensor counting each cycle, stroke or product passing a point gives you Performance and part counts.
• A current transformer clamped around the motor supply shows whether the machine is running, idling or stopped, which gives you Availability.
• Tapping the machine's existing stack light or run lamp through an isolated input is a quick way to read its state.

Option 2: read signals the PLC already has

If the machine has a PLC, it already knows when it is running, faulted or counting. A spare digital output can be configured to pulse once per cycle or to signal the run state, and wired to an IoT gateway. On older Siemens S7-300 and S7-400 systems this is often the most reliable option, because adding communication to an old CPU can be harder than wiring one extra signal.

Always agree any PLC change with the maintenance team, document it, and keep a backup of the program before and after. Monitoring must never interfere with interlocks or safety logic.

Option 3: connect over the network

Newer controllers can share their data directly. Siemens S7-1500 CPUs include an OPC UA server, many Allen-Bradley controllers can be read over EtherNet/IP, and plenty of drives, meters and HMIs speak Modbus TCP or RS-485. A gateway reads counters and status tags on a schedule and forwards them to the cloud, with no wiring changes on the machine.

Keep the connection read-only wherever possible, and put the gateway on a separate network segment so the production network is protected.

What to measure first

Start with three things per machine: a run or stop state, a good-part counter and a reject counter if one exists. That is enough to calculate Availability, Performance and Quality. Downtime reasons can be added by operators on a tablet for stops longer than a few minutes, while short stops are captured automatically.

Common mistakes to avoid

• Counting the wrong thing, such as machine strokes when one stroke makes several parts. Agree the counting rule first.
• Using an unrealistic ideal cycle time, which makes Performance look worse or better than it is.
• Noisy signals from vibration or loose wiring that create false stops. Debounce inputs and check the wiring.
• Rolling out to the whole plant at once. Prove the numbers on one line first.

Bringing it together

A mixed plant usually ends up with all three options: sensors on the oldest machines, PLC signals on mid-life equipment and network connections on the newest lines. EnicSoft OEE accepts production pulses from any of these through a single endpoint, so every line is measured the same way. If you have a line full of older equipment and want to know what it would take to monitor it, we are happy to look at it with you.