I/O Loop Checking: The Commissioning Procedure Nobody Teaches and Everybody Is Judged On
What loop checking is
Before a machine runs, somebody has to prove that every input the PLC thinks it has is connected to the device the drawing says it is, and that every output actually operates the thing it is supposed to operate.
That is loop checking, and on a mid-sized machine it is several hundred points. It is unglamorous, it takes days, and it is the single most common task a junior automation engineer is given in their first year. It is also how you are judged, because a missed point becomes a fault during production and everyone remembers who signed the sheet.
The two directions
Inputs: operate the field device, watch the PLC.
Physically actuate the sensor. Push the limit switch, put a target in front of the proximity sensor, press the button. Then look at the input status in the software and confirm the right bit changed. Not a bit. The right bit.
Outputs: force from the PLC, watch the field.
Force the output on from the software and go and look at the field device. The valve should move, the contactor should pull in, the lamp should light. Then force it off and confirm it goes back.
Both directions matter. An input checked only in software proves nothing about the wiring. An output checked only by watching the card's LED proves the card works, not that the valve is connected.
Why "the right bit" is the whole game
The classic fault is not a dead point, it is a crossed pair. Two sensors wired into each other's terminals both work perfectly. Every LED lights. Everything looks correct, and the machine will behave bizarrely the moment it runs, because the PLC thinks the part is at station 3 when it is at station 4.
Crossed points are only caught by checking each device against its tag, one at a time, and never by checking that "all the inputs come on".
The procedure that scales
For a handful of points you can improvise. For three hundred you need a method.
- Print the I/O schedule. Tag, description, card, channel, terminal, device type. On paper or a tablet, one line per point, with a signature column.
- Work in address order, not physical order. Walking the plant in tag order feels inefficient and prevents you losing your place. Efficiency here costs you accuracy.
- One person at each end, on the radio. Field person operates, panel person confirms the tag, then says the tag back. "IX2.4, conveyor 3 photo-eye" and not "yes, that one worked".
- Sign each line as you go. Not at the end of the sheet. Not at the end of the day.
- Mark the faults and move on. Do not stop to fix. Note it, continue, come back with a list. Stopping to chase one fault is how a two-day check becomes a five-day check.
- Re-check anything you rewired. A repaired point is an unchecked point.
Forcing outputs safely
Forcing an output makes something move. Before you force anything:
- Confirm the machine is isolated where it needs to be, and that people know what is about to happen.
- Know what the output does. Forcing a hydraulic valve is not the same as forcing a lamp.
- Force one at a time.
- Remove every force when you finish. A forgotten force left in a processor is a genuinely dangerous thing to hand over, and it is a fireable mistake on some sites.
Some platforms show a prominent indicator when forces are active. Trust it, and check for forces before handover regardless.
Analog points
Analog loops need more than "it changed". Use a loop calibrator to inject known values and check the scaled reading at several points, normally 0%, 25%, 50%, 75% and 100%. Record the actual readings, not ticks. A loop that is right at zero and wrong at span has a scaling or burden problem, and you only see that with multiple points. See 4 to 20 mA loop troubleshooting.
What the paperwork is for
The signed schedule is the evidence that the machine was proved. When something fails in month three, that document decides whether the conversation is a technical one or a commercial one. Keep it legible, keep it complete, and hand a copy over.
A punch list of everything not resolved goes with it. An honest punch list is respected. A clean sheet that turns out to be optimistic is not.
Why employers ask about this
"How would you check I/O against the schedule?" is a favourite interview question precisely because it cannot be answered from a textbook. A candidate who says "force each output and prove it at the field device, operate each input at the field device and prove it at the PLC, sign each line, never sign a point I did not personally see move" has clearly stood in front of a panel. That answer is worth more than a list of PLC brands.
Our free interview question prep tool has this one under Commissioning and fault finding, with what a strong answer contains.
Do it for real
Loop checking is learned by doing it badly once under supervision. On the Automation Engineer Program trainees join live client projects as junior engineers and check real I/O against real schedules, which is why the portfolio they leave with is the thing that gets them hired.
Related: The seal-in circuit, First 90 days on site.
