Key takeaway: If you can’t access a PLC, you can still get reliable OEE. Three options exist : manual operator entry, IoT retrofit sensors, and Computer Vision : each with very different cost, accuracy, and rollout timelines. Computer Vision is the only option that needs zero machine integration.
Quick answer: OEE monitoring without PLC integration is possible in 2026 via Computer Vision OEE: one industrial camera per line replaces PLC tags, sensor wiring, and IT-OT integration projects. The camera tracks machine state, counts parts, and detects causes of downtime in real time. Result: full Availability + Performance + Quality data on legacy lines in 1–2 days, not the 6–12 weeks a PLC integration takes. Best fit: plants with mixed-vintage equipment where wiring every line to a SCADA system is uneconomic.
The textbook answer to OEE is "tap the PLC." The reality on most floors is messier.
A typical mid-size plant has machines from three or four decades, bought from different vendors, with three or four different controllers. Some of them have PLCs you can read from. Some have proprietary controllers locked by the OEM. Some : especially older filling lines, mechanical presses, packaging machines : have no PLC at all.
For the machines you can’t tap, you have three practical choices: manual entry, IoT retrofit sensors, or Computer Vision. The right choice depends on line speed, criticality, and how much downtime you can spend on integration.
Related deep-dives: Computer Vision OEE field guide · visual downtime verification comparison · OEE data collection methods · OEE Complete Guide.
The cheapest path: give operators a tablet at the line and ask them to log start/stop, downtime reasons, and good vs. reject counts.
Where it works: Low-speed lines where operators have idle time between cycles. Pilot runs. Plants that need OEE data for a finance audit, not for shift-by-shift improvement.
Where it breaks: High-speed lines (anything above ~30 cycles per minute). Three-shift operations where data quality drifts as fatigue sets in. Any culture where operators feel the data will be used against them.
Real cost: The software is cheap (often under €30/line/month). The hidden cost is data quality: independent audits show manual OEE entries miss 20-40% of true downtime, mostly micro-stops under 2 minutes.
Stick a sensor on the machine: vibration sensor on the spindle, current clamp on the motor, photo-eye on the conveyor. The sensor talks to a gateway, the gateway talks to the cloud.
Where it works: One sensor per machine type that’s stable in your fleet : e.g., 40 identical CNCs, all the same model. Greenfield retrofits where you’ve budgeted for cable runs and electricians.
Where it breaks: Mixed fleets (every machine type needs its own sensor profile). Food and pharma lines (washdown zones eat sensors and connectors). Older mechanical equipment where there’s no clean signal to tap.
Real cost: Sensor + gateway + install runs €400-€2,000 per machine. Multiply by 50 machines. Add an electrician for the cable runs. Add a downtime window to install. Then add 6-8 weeks for the integrator to map signals to OEE states. The software fee is the smallest line item on the invoice.
Mount a camera above the line. The camera watches the work, recognises good output, recognises stops, and computes OEE from what it sees. No PLC tap. No wires into the machine. No electrician.
Where it works: Almost any line where you can place a camera with line of sight. Mixed fleets where every machine is different. Older mechanical equipment with no usable signal. Food and pharma lines where washdown forbids in-machine sensors.
Where it breaks: Very enclosed cells where camera placement is physically blocked. Lighting environments that change minute by minute (rare). Compliance cultures that forbid recording on the floor (you can use frame-only mode without retention).
Real cost: Camera plus mount plus PoE switch is around 300 to 800 EUR per line. No machine integration time. Typical pilot live in 1 to 5 days. The software fee is the main line item, which is what you want: you pay for the intelligence, not the cabling.
This is the path Fabrico calls Computer Vision OEE, and it is the only option that scales across a mixed fleet without per-machine integration cost.
Pick by the dominant constraint on your floor.
| Constraint | Best fit | Why |
| Mixed fleet, every machine different | Computer Vision | One sensor format covers all |
| Identical CNCs, 30+ units | IoT retrofit sensors | Per-machine economics work |
| Low-speed line, small team | Manual operator entry | Cheap, fast, good enough |
| Washdown food or pharma line | Computer Vision | Cameras stay outside the wet zone |
| Older machines with no controller | Computer Vision | No tap needed |
| Need data this week | Manual entry, then Computer Vision | Get a signal now, automate later |
A 1-day pilot proves whether camera placement, lighting, and signal recognition work on your real line. No machine integration. No long contract. We bring the camera, we calibrate, you see live OEE numbers by end of day.
Book a 30-minute call to scope a pilot for one of your hardest lines.
Login
Blog
