If a machine on your line stopped right now, how long would it take before the right person knew? For most plants, the honest answer is "too long." Andon is the system built to fix exactly that — and it has been proving itself on factory floors since the days of the original Toyota Production System.

What does "andon" mean?

The word andon (行灯) is Japanese for a paper lantern — a light. Toyota borrowed it for the shop floor to describe a simple idea: make problems visible the instant they happen. In a lean plant, an andon is any signal — classically a colored light — that shows the state of a line and calls for help when something needs attention.

Andon is one pillar of jidoka (autonomation, or "automation with a human touch"): the principle that a process should stop and flag a problem the moment quality is at risk, rather than pass a defect downstream. The andon is how that stop-and-signal happens in the real world.

How an andon system works

Every andon system, from a single light to a plant-wide platform, follows the same three steps:

  1. Signal. An operator presses a button, pulls a cord or taps a keypad to raise a call — machine down, material needed, quality check, safety issue.
  2. Notify. The call routes to the people who can act — a supervisor, a maintenance tech, a material handler — via a light, a board, or a message to the device they already carry.
  3. Log. The system records the call: what type, which station, and how long it took to answer and resolve. That log is what turns andon from a doorbell into a management tool.

The magic is in the routing and the record. A good andon doesn’t broadcast to everyone (which everyone learns to ignore) — it reaches the right responder, escalates if no one answers, and leaves an audit trail behind.

The parts of an andon system

  • The trigger — a wireless button, a pull cord, or a multi-button keypad at the station where the operator works.
  • The andon board — a floor display showing every active call, color-coded by station and status, so the whole team can see the state of the line at a glance. (We cover this in detail in Andon Board Explained.)
  • Andon lights / stack lights — the classic red/amber/green tower that signals running, warning, or stopped at the machine itself.
  • Responder notification — modern systems push the call to a smartwatch, phone or tablet so the responder gets it wherever they are, not just if they happen to be looking at the board.
  • Analytics software — the layer that logs every call and turns response times into KPIs by line, shift and call type.

Why andon matters

Andon pays for itself in three places. First, downtime: the faster the right person responds, the less production you lose to a stopped line. Second, quality: catching and flagging a defect at the source stops it from being built into hundreds of units. Third, visibility: once every call and response is logged, you can finally see where your line loses time — the recurring fault, the slow shift, the station that always waits — and manage the pattern instead of firefighting each incident.

Traditional wired andon vs. modern wireless andon

Classic andon is hard-wired into the line. It works, but it is expensive to install, disruptive to change, and effectively frozen once it is in — moving a station or adding a call point means an electrical project. That cost is why a lot of plants still run without any andon at all.

Wireless andon removes that barrier. With Destec’s Smart Call ANDON, the buttons, boards, lights, watches and phones all communicate over RF — so you can add andon to an existing line in days, with no rewiring and no IT project, and move or expand it whenever the line changes. Every call still routes to the right responder and logs to Bell Cloud for response-time analytics. Destec has been building this kind of wireless signaling since 2006, across 1,600+ customers and 16,000+ establishments.

Getting started with andon

You don’t have to wire the whole plant to find out if andon works for you. The lowest-risk way to start is a single line: put wireless call points at a few stations, a board at the supervisor desk, and watch the response-time data for a couple of weeks. If it moves the number, you scale it. If it doesn’t, you’ve risked nothing.