Field commissioning notes: the things you only learn by being in the cabinet

[Aior]

Office work doesn't teach you this​

Every commissioning we've done has included at least one "huh, didn't expect that" moment that wasn't in any document. After enough cells, those moments form a pattern. Below are the things we now build into our checklists, written for the engineer who hasn't been on enough sites yet to have learned them the expensive way.

Day one: walk the cell before you touch anything​

• Photograph every cabinet before you start. From outside, then with the door open, then inside any subassemblies. The "before" reference saves hours later.
• Verify the as-built drawings match the actual wiring. They don't. Not in any cell we've ever commissioned. Mark up the discrepancies on day one.
• Find the maintenance access path. Where will the operator stand? Where will the engineer stand at 3 AM? The wiring you'd choose if you accept that someone needs to access the panel quickly is different from the wiring you'd choose for the prettiest schematic.

Power-on sequence — slow, with checks​

Never apply full power to an untested cell. The sequence:

1. Verify mains voltages at the disconnect, with the disconnect off. Phases, neutral, ground.
2. Energise the disconnect. Verify voltages downstream.
3. Energise control voltage circuits one-by-one. Each circuit, check current draw against expected.
4. Energise drive bus voltage with motors disconnected. Verify drive POST.
5. Energise drives, motors connected, but with motion software disabled. Verify enable signals work as expected.
6. Enable motion at lowest speed and torque limits. Move each axis individually, manually.
7. Run automated motion at progressive speed/torque limits.

This sounds slow because it is. The alternative is finding out about the wiring error by destroying a 5 000 EUR servo on the first power-up.

Grounding — the thing nobody verifies​

Every commissioning, measure:

• Ground impedance from cabinet to building ground bond (target < 0.1 Ω)
• Voltage between cabinet ground and signal common at every PLC and drive (target < 100 mV)
• Common-mode voltage on shielded comms cables under load

If any of those are out of spec, you'll see them as intermittent comms errors, motor noise, sensor drift, or worse — and you'll spend three days chasing software bugs before realising the ground is the problem. Measure first, save days later.

Cable routing — keep separations religiously​

Power and signal cables in the same tray cause more EMC problems than any other commissioning issue we see. The rules:

• Servo / VFD power cables: separate tray, minimum 200 mm separation from signal trays.
• Crossings at 90°, never parallel runs.
• Encoder / sensor cables: shielded, shield grounded at one end (the controller end), not both.
• Profinet / EtherCAT cables: away from contactor coils — those switching transients do show up on the bus.

If the as-built routing violates these, fix it before commissioning. It's faster than debugging the consequences.

The commissioning notebook​

On every site, one engineer keeps a paper notebook. Yes, paper. In it goes:

• Every parameter changed from default, why, and the value
• Every wiring discrepancy found, with the cabinet location
• Every odd behaviour observed, even if "fixed itself"
• Setpoint values, drive tuning constants, calibration offsets — anything not in source control

The notebook gets photographed and added to the project archive at handover. The number of times we've gone back to a notebook from a cell commissioned 18 months earlier and found exactly the answer we needed is high enough that the practice pays for itself many times over.

The handover ritual​

Before you leave site:

• Run the cell in production conditions for at least one full shift, with the customer's operator
• Force at least three failure conditions and verify the cell handles them (E-stop, comms loss, safety door open during cycle)
• Walk the customer through the alarm log and the runbook for the top 5 alarms
• Hand over backups: PLC project, HMI project, drive parameters, IPC image, parameter sheet
• Schedule a 30-day check-in. Cells that have a scheduled check-in survive better than cells that don't.

One thing we always carry in the toolkit​

A USB-to-RS485 adapter, a multimeter, and a 5 m Ethernet cable. Half the field debugs come down to one of those three. The other half: a flashlight, because cabinet lighting is always inadequate.

What's your day-one ritual on a new site? Curious how others structure cold-commissioning vs hot-commissioning sequences.