Getting Data From Old Machines Without Replacing the PLC
November 11, 2026 · 6 min read · Etzal Group Team
A factory wants to know why line three produces less than line two. The machines on line three are from 2009. They run reliably and their controllers have no Ethernet port. One supplier proposes replacing the control system at a cost that ends the conversation. There are cheaper ways. Getting data from old machines rarely requires replacing anything, and the first useful numbers can come from a sensor that never touches the PLC.
Start with the question, not the data
Before choosing a method, write down what you want to know. Typical first questions:
- How many hours did the machine actually run this shift?
- How many times did it stop, and for how long?
- How many good parts did it make?
- How much energy did it use?
These four need very little data: a running signal, a part count, and a power reading. They do not need the controller's internal variables. Many retrofit projects stall because they try to read everything before deciding what for.
Method 1: external sensors
Add your own sensors and leave the machine's controls alone.
- Current clamp on the main motor. The current pattern separates off, idle and under load. This alone gives running time and stoppages.
- Photoelectric or proximity sensor on the output. Counts parts or cycles.
- Energy meter on the supply. Kilowatt hours per shift and per part.
- Vibration and temperature sensors on bearings and motors. Early signs of wear.
- Stack light sensor. A small device that reads the machine's existing signal tower: green, amber, red.
Advantages: no change to the machine, no risk to the control program, no need for the original manufacturer, installation in hours. It works on any machine of any age.
Limits: you see what the machine does, not why. A stop is visible, and the alarm code behind it is not.
Method 2: tapping existing signals
Most machines already have electrical signals that carry meaning: a relay that closes when the cycle completes, a lamp output for a fault, a counter pulse. These can be read in parallel with a small input module, using isolation so the monitoring device cannot disturb the circuit.
Advantages: more meaning than external sensors, such as a real "cycle complete" or "fault" signal, still without touching the program.
Limits: needs the electrical drawings and someone qualified to work inside the panel. Each signal must be identified and wired.
Method 3: a protocol gateway
Many older controllers have a serial port and speak an industrial protocol such as Modbus RTU, or a manufacturer's own protocol. A gateway connects to that port, reads selected values, and publishes them in a modern form such as MQTT or OPC UA.
Advantages: access to what the controller knows: counters, setpoints, temperatures, alarm codes, recipe numbers.
Limits:
- You need to know which memory addresses hold which values. That comes from the program documentation, the manufacturer, or careful investigation.
- The port may already be used by an operator panel.
- Older controllers answer slowly. Ask for too much, too often, and you can affect the panel's responsiveness.
- Some proprietary protocols need a specific gateway or licence.
Configure the gateway to read only. A monitoring system has no reason to write to a controller.
Method 4: adding a module to the controller
Some controllers accept an added communication module, such as an Ethernet card. This can give clean access to the data, and may require a small change to the hardware configuration or program.
Advantages: reliable, fast, supported by the manufacturer.
Limits: the module may be discontinued or expensive for older models, the change needs someone who knows that controller family, and it must be done during a stop with a backup in hand.
How to choose
| Situation | Sensible first method |
|---|---|
| You want running time, stops and counts | External sensors |
| You need a true cycle or fault signal | Tapping signals |
| You need alarm codes, setpoints, process values | Protocol gateway |
| The controller is modern enough and supported | Communication module |
| No documentation and no manufacturer support | External sensors |
| Machine under warranty or a service contract | External sensors, and ask the supplier first |
Many plants combine them: external sensors on every machine for a common baseline, and deeper connection on the few machines where the reasons for stops matter most.
Reasons for stops still need people
Whatever the method, the data shows that a machine stopped at 10:42 for eleven minutes. It often cannot say whether that was a material shortage, a changeover or a jam. A simple panel or tablet at the machine where the operator picks a reason from a short list closes that gap. Keep the list to eight or ten reasons. Long lists get "other".
Safety and network rules
- Read only. Monitoring never writes to controllers.
- Separate networks. Gateways sit between the control network and the monitoring network, and the control network is not exposed to the office or the internet.
- Back up the PLC program before connecting anything to the controller.
- Qualified people in the panel. Tapping signals means working near live circuits.
- Do not interfere with safety circuits. Emergency stops and guards are never rewired for monitoring.
- Test during a planned stop and watch the machine through several cycles after.
A pilot in four weeks
- Week 1. Choose one line. Write the questions. Walk the machines and note motors, outputs, signal towers and controller types.
- Week 2. Install external sensors and a gateway on three to five machines.
- Week 3. Collect data, compare with production records, correct thresholds for off, idle and running.
- Week 4. Review with the production team. Which numbers surprised them? Which machines deserve a deeper connection?
A pilot like this costs little and tells you where further investment will pay.
When replacing the controller is right
Sometimes it is: spare parts are no longer available, failures are frequent, the process needs functions the old controller cannot provide, or safety standards require it. That decision should rest on those reasons. Wanting data is not, on its own, a reason to replace a working control system.
We design retrofits of this kind in our IoT and Control Systems work, usually starting with a short pilot on one line. To discuss your machines, contact us.
Summary
Write the questions first. For running time, stops, counts and energy, external sensors are enough and touch nothing. Tap existing signals for true cycle and fault states. Use a read-only protocol gateway when you need what the controller knows. Add operator input for stop reasons. Keep networks separate, back up programs, and pilot on one line before deciding how deep to go.
Frequently asked questions
Can a machine with no network port be monitored?
Yes. The simplest method does not touch the controller at all: a current clamp on the motor cable shows whether the machine is off, idle or working, and a sensor on the output counts parts. This gives running hours, stoppages and output counts, which is enough for availability and performance figures.
Is it risky to connect to an old PLC?
It can be if done carelessly. Reading data over a serial or network link is normally safe, and writing to the controller is not something a monitoring project should do. Use read-only access where the device allows it, keep the monitoring network separate from the control network, and test during a planned stop. Take a backup of the PLC program first.
How much does it cost to retrofit monitoring on one machine?
It varies widely with the method. External sensors with a small gateway are the cheapest and can be installed in hours. A protocol gateway reading the controller costs more in hardware and engineering time. The largest cost is usually not the device but the work of deciding which signals matter and making the data useful.
Related service
WhatsApp AI Chatbot
We build, train, and run a WhatsApp AI agent that answers customers 24/7, qualifies leads, and follows up.
