Seven network mistakes that stop production lines.

Plant networks grow line by line, often by different contractors. The result is a network nobody designed.

The common mistakes

1. A flat network. Controllers, cameras, laptops and guests share one broadcast domain, so a fault anywhere affects everything.
2. Office-grade switches on the floor. Dust, heat and vibration shorten their life, and they lack industrial protocol support.
3. Redundancy that was never tested. A ring that has not been broken on purpose may not recover as designed.
4. No IP plan. Duplicate addresses and undocumented devices cause strange faults.
5. Uncontrolled remote access. Vendors connecting through shared passwords and modems.
6. No separation from office IT. One infected laptop can reach controllers.
7. No documentation. When the person who built it leaves, the knowledge goes too.

What good looks like

A defined architecture with zones by function, industrial switches in redundant rings, a plan for addresses and VLANs, a boundary with the office network and controlled vendor access.

Fixing it safely

Start with discovery: map what is connected. Then plan changes for shutdown windows, with rollback procedures, test plans and a pilot ring.

Validate

Measure failover time, loss and latency. Keep the test records as part of the documentation pack.

Why it pays

Network faults often hide behind “the PLC dropped out”. A well-designed network removes a whole class of unexplained stoppages.