What Should an Instrument Loop Check Include Before Startup?
An instrument loop check procedure before commissioning should trace each signal from the field instrument, through its wiring and PLC configuration, to the displayed value and associated output. It should define the expected range, units, signal direction and response, then record the observed results and any discrepancies. Before startup, reconcile every temporary force, simulation and bypass; plan safety-related validation separately with a qualified assessor.
What should be agreed before testing starts?
Start with an approved test scope and a clear boundary around equipment that could move, energise or release material. Agree who controls the test, who observes the field device and who can authorise a change in plant state.
The procedure should identify the required permits, isolations and operating conditions under the site's arrangements. Do not improvise these conditions at the panel.
For each loop, establish:
- The instrument tag and physical location.
- The current loop sheet, wiring drawing and configuration references.
- The intended measurement range and engineering units.
- The PLC input or output channel and corresponding software tag.
- The operator display, alarm and associated control response.
- The test method, acceptance tolerance and evidence required.
- Any response that cannot be exercised during this test.
Use the agreed functional requirements to define expected behaviour. If those requirements are unclear, resolve them before deciding whether the loop passes. Our guide to what a functional design specification should contain explains where these decisions belong.
What belongs on the loop sheet?
Use the loop sheet as a record of the complete path being checked. It should let another person identify what was tested, where the test signal entered and which parts remain unverified.
| Record | What to include |
|---|---|
| Identity | Instrument tag, service, location and drawing revision |
| Field configuration | Measurement range, units and configured signal type |
| Wiring path | Terminals, cable references, junctions and controller channel |
| PLC interpretation | Raw input reference, scaling, units and relevant status handling |
| Operator indication | Display location, tag, units and expected precision |
| Associated response | Alarm, permissive, command or final element covered by the test |
| Test evidence | Applied value, observed values, tolerance and result |
| Closeout | Discrepancy reference, retest result and restoration confirmation |
Record the test equipment identity and the evidence required by the project's calibration arrangements. A tick beside the instrument tag should never be the entire result.
How do you trace the signal from the field?
Begin by matching the physical instrument to its tag and service. Confirm that its configured range and units match the approved documents.
Then follow the documented path through field terminals, intermediate connections and the assigned controller channel. Compare the drawing with what is installed, and record differences instead of silently accepting them.
State exactly where the stimulus enters the loop. Applying a process stimulus at the instrument, injecting an electrical signal at a terminal and substituting a software value are different test scopes. Treat a test as evidence only for the path it actually exercised.
Where the scope requires an end-to-end check, arrange a suitable field test method under the approved procedure. If access or plant conditions prevent it, record the missing coverage and the required follow-up.
Do not label an entire loop complete because the display changed during a software simulation.
How should range, scaling and units be checked?
Choose test points across the approved operating range, including intermediate values. Agree the points and acceptance tolerances before observing the results.
At each point, record the applied stimulus, the PLC's interpreted engineering value and the operator display. Include a local indication where one forms part of the scope. Compare each result with the same approved range and units.
Illustrative example: suppose a pressure loop is configured so that a 4–20 mA input represents 0–10 bar. The following values describe that example's intended mapping; they are not a universal instrument configuration or acceptance tolerance.
| Applied signal | Expected PLC value | Expected display |
|---|---|---|
| 4 mA | 0 bar | 0 bar |
| 12 mA | 5 bar | 5 bar |
| 20 mA | 10 bar | 10 bar |
Record actual readings alongside these expectations. Agree how display rounding will be assessed so that a rounded display does not become the only acceptance evidence.
Also check that the range suits the intended service. Matching configuration entries is insufficient if the approved range itself is unresolved.
How do you verify signal direction and the associated output?
Write the expected direction explicitly: when the measured quantity rises, what should rise, fall, activate or stop?
Check that direction at each relevant stage. Separate the measured value from alarm state, controller demand, output command and equipment feedback. For a switch input, record both physical states and their intended meanings.
For an analogue output, define the relationship between the command, transmitted signal and final element response. Where movement is authorised, record the observed position or action and compare it with the requirement.
A displayed command is evidence of the command value. Require separate evidence for physical response where that response is within scope.
If the output cannot be exercised safely, mark that part as deferred. Identify the remaining test and its release condition rather than accepting an assumed result.
Which alarms and abnormal conditions belong in the check?
Select abnormal-condition tests from the approved design and device documentation. Define the expected indication and response before introducing a test condition.
Depending on the loop, the agreed scope may include:
- Measurement alarms and their reset behaviour.
- Invalid or unavailable signal indication.
- Loss of communication for a networked instrument.
- Command and feedback disagreement.
- The specified response when a valid measurement returns.
Use an approved method appropriate to the equipment and plant state. Do not assume every device or input channel represents a failed signal in the same way.
Keep the boundary with safety-related validation explicit. Plan that work with a qualified assessor, including the required evidence and authorisation. The machine retrofit safety-assessment guide covers planning that assessment alongside the wider project.
How should discrepancies and retests be recorded?
Record the expected result, observed result and exact test condition. “Scaling wrong” is less useful than a record showing the applied value and the readings at each stage.
Give each discrepancy an owner, an agreed correction and a closure condition. Distinguish among a failed result, an incomplete test and a document mismatch.
After correction, repeat the affected checks and any other checks whose results depend on that change. Retain both the original observation and the retest evidence.
Use a shared discrepancy record so that unresolved items remain visible at the startup decision. Our guide to managing a controls commissioning punch list explains how to assign and close those items.
What must be restored before startup?
Maintain a temporary-change register throughout testing. Include PLC forces, substituted values, simulations, inhibited alarms, bypasses, temporary wiring and changed settings.
Before releasing the tested scope:
- Reconcile every register entry against the actual equipment and software state.
- Remove test forces, substitutions and temporary connections.
- Restore the approved configuration and operating mode.
- Confirm live field values and equipment feedback.
- Verify that alarms and responses disabled for testing are restored.
- Record remaining restrictions and who controls them.
Do not infer restoration from a normal-looking screen. Require explicit evidence that temporary test conditions have been removed. Any retained exception needs a separate authorised arrangement and must remain visible in the startup decision.
What common mistakes leave gaps?
Checking only the display. Require evidence at the field device, controller and associated response points covered by the scope.
Checking only the range endpoints. Include agreed intermediate points and assess the complete configured mapping.
Accepting undocumented differences. Resolve drawing and configuration discrepancies through the project's change process.
Treating loop checks as all commissioning evidence. Keep calibration, functional testing, safety-related validation and startup authorisation as separately defined requirements.
Closing deferred tests as passed. State what remains untested and when it must be completed.
Frequently asked questions
Is a loop check the same as calibration?
Define them separately in the test plan. Require calibration evidence for measurement accuracy, and loop-check evidence for the installed signal path and specified response. Do not accept one record as proof of both unless its documented scope covers both.
Can testing begin from the PLC?
Yes, where that is the agreed scope. Record the starting point and identify the field instrument and wiring checks still required before accepting the complete loop.
Who should accept the completed record?
Name the required reviewers in the commissioning plan. Acceptance should consider observed results, discrepancy closure, deferred work and restoration evidence. Plan safety-related acceptance with a qualified assessor.
When is a loop ready for startup?
Release it only when the agreed tests pass, startup-blocking discrepancies are closed, temporary test conditions are reconciled and the responsible people accept the record. A completed loop sheet supports that decision; it does not replace the plant's startup authorisation. For drives, see our VFD commissioning checklist, and for the records that follow startup, see how to keep instrument calibration records that support plant data.