Where should an inspection station sit, and how does it reject?

Where on the line should a station sit?

Straight after the step that can cause the fault, while the fault is still visible. Fill level and caps are checked after the capper; labels and date codes after the labeller and coder; counts at the case packer; case and pallet labels at palletising. A foil seal is checked before the cap covers it, and an empty bottle before it is filled. The line map on the applications page shows the usual positions.

The chosen spot also has to work physically. It needs space for the camera, the light and a shroud, with room to clean them; products presented the same way every time, or guides that make them so; room to fit a reject; and access to the PLC. Where one of these is missing, the proposal says what has to change on the line.

How does the line know which product to reject?

The camera and the reject are usually some distance apart, with other products in between. The station sends its result to the PLC over digital outputs or an industrial network; HIKROBOT lists TCP/IP, Modbus, serial, UDP and EtherNet/IP among the protocols VisionMaster supports (HIKROBOT catalogue 2025 Q4). The PLC then holds that result until the product reaches the reject.

• Time delay: the reject fires a set time after the result. This works on constant-speed lines where products do not slip, and goes wrong when the line slows, stops or restarts.
• Encoder tracking: the PLC follows each product by conveyor distance in a shift register, so tracking stays right through speed changes and stops, as long as products move with the belt. Where they slip or bunch up, a sensor at the reject re-identifies them.

Counting triggers against results is a simple check that no product went without a result.

Which reject device fits the product?

The reject is chosen for the product, the gap between products and the line speed, not for the camera.

Rejected products go into a lockable bin with a bin-full sensor, so they cannot find their way back to good product.

How do you know the reject worked?

Firing the reject does not prove the product left. A confirmation sensor at the chute, in the bin or just after the reject point checks that the failed product left within an agreed window. If it is still on the line, or a good product was removed instead, the PLC raises an alarm or stops the line. An air-pressure switch and the bin-full sensor complete the picture.

Food safety audits look at this closely. BRCGS guidance on clause 6.2.4 describes testing online label and print verification equipment with incorrect packs, checking the detection, the reject or line stop, the authorised restart and the mis-read or no-read alarm (BRCGS Issue 8 FAQ). Chapter 6 covers these tests.

What happens on power loss or a station fault?

Power interruptions, including load shedding and short dips, stop a line in an uncontrolled way. The camera, the PLC and the reject may come back at different times, and whatever was between the camera and the reject is no longer tracked. A fail-safe design treats that as unknown:

• Products in the tracking window are rejected on restart, or the operator clears that section before the line is released.
• Until the station reports ready, the PLC holds the line or rejects every product that passes the camera.
• A UPS lets the vision PC or controller shut down cleanly, so recipes and settings survive.
• Before production restarts, the active recipe and air pressure are checked and a short challenge with test packs is run, and the event is logged.

The same thinking covers no result in time, a camera offline, low air or a failed confirmation: each response is agreed with production and quality and tested at FAT and SAT. This is general control practice, not a certified safety function. The PLC and reject integration guide goes through signals, tracking and faults in detail.

Track the physical item through stops and restarts

The camera location must expose the defect before packaging or handling hides it. The reject location must provide enough travel time for the complete decision and actuator response while maintaining reliable product identity. A timer can be adequate under controlled motion; changing belt speed, slippage, accumulation and stops may require encoder-based or other explicit product tracking. The integration design must state its assumptions.

Worked example — hypothetical: In a hypothetical start-stop test, a known faulty pack is left between the camera and reject. The conveyor is stopped, then restarted with a good pack immediately behind it. The expected outcome is removal of the faulty pack without losing the good one, or a documented controlled hold if tracking cannot be preserved. Save the tracking event record rather than relying only on the screen’s pass/fail icon.

A reject device is not automatically a machinery safety function. Guarding and safe maintenance access belong to the machine risk assessment. In food premises, R638 provides the local hygiene context for installation and cleaning; it does not specify a universal reject delay or make the camera safety-rated. (South Africa: food premises and transport hygiene regulations R638, 2018) (EMVA: GenICam)

Keep the original image together with the active recipe and the reference judgement when investigating a disagreement. Change one controlled setting at a time and re-run the relevant challenge samples. Once a correction works, verify the neighbouring requirements too: improving contrast for one feature can hide another. Record the revised configuration and the evidence for release so maintenance does not have to reconstruct the decision from memory.