Machine vision event sequencing for deterministic inspection

A reliable inspection sequence starts with a physical event, not a chain of software messages. The challenge is deciding which actions need bounded timing, which must follow conveyor position, and which can wait for supervisory logic. This guide helps you divide those responsibilities, calculate a timing budget, and commission the sequence without assuming that any particular controller or programming interface provides the required performance.

Choose the timing architecture before the hardware

Machine vision event sequencing should put position-critical and exposure-critical actions in a hardware path with documented timing bounds. That path might use a dedicated timing controller, camera timing outputs, or suitable PLC hardware. The device label matters less than its verified capabilities.

Keep recipe management, inspection reporting, operator commands, and production coordination in supervisory logic. A network command can prepare an acquisition without being the signal that determines its exact instant.

A PLC with suitable hardware functions can provide precise timing. Conversely, a separate controller is not automatically deterministic because it is dedicated to vision.

For an integrated processing platform, the standalone vision controller selection guide explains the broader selection decision. The vision controllers category identifies that equipment class, but built-in lighting outputs alone do not establish encoder-sequencing capability.

Define an event contract for every device

Before wiring, write down what each signal means. “Trigger” is incomplete: it could mean a detected part, accepted acquisition, exposure start, or completed inspection.

Use an event table that separates the requested action from evidence that it happened.

Specify active edge, signal polarity, minimum pulse width, rearm conditions, and behavior while busy. Also define startup states: an output should not pulse merely because a connection appears or a recipe loads.

Electrical compatibility is a separate requirement. Check voltage levels, output topology, input current, isolation, common references, and connector assignments using current vendor documentation.

The camera GPIO wiring guide covers electrical fault prevention. If a valid electrical pulse is ignored, use the trigger source and line configuration checks before changing the sequence logic.

Use encoder position for travel and time for device behavior

A fixed delay represents a fixed distance only while speed remains constant. Encoder-based scheduling follows measured travel, making it the usual choice between an upstream sensor and an inspection position.

Use time-based delays for actions whose physics are time-dependent: pulse duration, settling intervals, or a documented device response. Often the correct sequence combines both methods.

Worked example: a changing conveyor speed

Assume an inspection point is 150 mm downstream from the sensor. The calibrated encoder scale is 0.05 mm per count.

The positional offset is:

Scheduling at the captured encoder count plus 3000 counts preserves the nominal travel offset, provided the encoder represents actual part movement. It does not correct belt slip, loose parts, or an incorrect scale factor.

The offset inspection sensor timing guide develops the sensor-to-camera relationship. Use encoder calibration and position-reset troubleshooting when the calculated location changes after restarts or counter operations.

Account for acquisition latency

Reaching the camera trigger position is not necessarily reaching the exposure position. If exposure starts after a fixed delay, the part continues moving during that interval.

For example, a hypothetical 100 µs delay at 1000 mm/s corresponds to 0.1 mm of travel. At the example encoder scale, that equals 2 counts. Compensation must use the relevant speed and documented delay, rather than assuming that one fixed count correction suits every operating condition.

Under acceleration, a constant-speed correction is only an estimate. Validate the complete operating envelope or use a timing architecture that accounts for the motion profile.

Align usable light with actual exposure

A controller output edge, electrical LED current, and useful optical output are different events. Design around the interval when the required illumination actually reaches the part.

For a global exposure, a short pulse must fall within the interval when the intended pixels integrate light. With rolling readout, rows may integrate at different times; a short flash can illuminate only part of the image unless the camera provides a suitable common exposure interval.

As an illustrative timing target, consider a 100 µs exposure containing an 80 µs usable light interval. Centering that interval leaves 10 µs at each end. Those margins must accommodate relevant edge uncertainty, optical rise and fall, and camera timing variation. They are not automatic allowances supplied by a controller.

The optical rise-time measurement guide explains why an electrical pulse measurement cannot establish the useful light interval. For moving parts, also check the exposure and motion-blur relationship.

Prefer a documented camera exposure signal when it represents the interval you need. Otherwise, derive camera and lighting outputs from a common timing event and measure their relationship.

Budget variation separately from fixed delay

Deterministic does not mean instantaneous. A repeatable fixed delay can often be compensated; unpredictable variation consumes the available inspection tolerance.

Build a worst-case budget from documented bounds or measurements with a stated test scope. Separate sensor switching variation, input filtering, event capture, output scheduling, camera response, and optical timing. Avoid counting the same delay twice.

Suppose a hypothetical design allocates 0.05 mm to encoder quantization and capture uncertainty. Three remaining timing contributions have bounds of 20 µs, 15 µs, and 10 µs.

At 1000 mm/s:

If the permitted positional error is 0.2 mm, this allocation leaves 0.105 mm for other effects. That remainder must cover relevant mechanical uncertainty, sensor geometry, and uncompensated behavior. It is not spare accuracy established by calculation alone.

Do not substitute average latency for a worst-case bound. Likewise, adding standard deviations does not produce a guaranteed maximum. The full feedback-loop latency guide helps distinguish acquisition timing from later processing and actuation delays.

Keep supervision and precision timing under clear ownership

Assign one owner to each output. Avoid having supervisory software and a timing engine independently rewrite the same lighting or trigger state.

A useful division is:

• Supervisory logic selects the recipe, checks readiness, authorizes production, and handles faults.
• The timing layer captures events, tracks position, and generates scheduled edges.
• The camera reports acquisition state and supplies image identity where supported.
• Inspection software associates images with parts and returns valid, invalid, or failed-inspection status.

Recipe changes need a handshake. Load parameters, validate them, acknowledge readiness, and only then accept the next part. Define whether previously accepted parts finish under the old recipe or are cleared through an explicit recovery procedure.

The recipe-selection race troubleshooting guide covers this boundary. For supervisory data exchange, use the PLC communication integration guide without assuming that ordinary register writes provide precision trigger timing.

A process sequencer is not a safety controller. Emergency stops, protective functions, and restart prevention require the machine's compliant safety design, independent of ordinary inspection scheduling.

Handle overlapping parts and missing events explicitly

A sequence that works with isolated parts can fail when several parts occupy the sensor-to-camera distance simultaneously. Track each accepted part rather than storing only the latest sensor position.

A first occupancy estimate is travel distance divided by minimum part pitch, rounded upward. Add capacity for the part entering the interval and any downstream tracking requirements.

For example, a 150 mm sensor-to-camera distance and 40 mm minimum pitch produce 150 / 40 = 3.75 pitches. An initial design allowance is therefore 5 records for that interval, before adding downstream capacity. Validate the allowance against actual sensor behavior and queuing rules.

Define what happens when a queue fills, a trigger arrives while the camera is busy, or an image never appears. Silently discarding an event can shift every subsequent result onto the wrong part.

Use an identifier, captured position, recipe version, and acquisition status for each tracked record. The asynchronous image and result matching guide explains why arrival order alone can be insufficient.

Select supporting parts without assuming sequencer features

The catalog provides light controllers, but lighting control and general event sequencing are different capabilities. A digital interface or strobe feature does not establish encoder inputs, position comparison, event queues, or specified timing jitter.

The LS-CTL-0070 lists a strobe feature. Treat it as a lighting-stage candidate, with timing behavior and interface compatibility requiring separate confirmation.

The LS-CTL-0290 is a digital constant-voltage light controller. Its catalog description does not establish dedicated event-sequencer functions.

Use the light controller finder to narrow the electrical output and channel requirements. Check continuous and pulsed operating limits against the actual light, following the light and controller duty-cycle matching guide.

For the signal path, select cables by documented connector, pin assignment, electrical requirements, and installation conditions. The LS-CBL-0300 is a camera power and I/O cable intended for drag-chain use; suitability still depends on the connected equipment.

Use the accessories finder after confirming the device pinouts. Connector fit alone does not establish compatibility.

Commission the sequence with measurable acceptance criteria

Record the permitted event-to-exposure error, minimum part pitch, speed range, and required fault responses before testing. Acceptance should depend on those limits, not simply on obtaining a recognizable image.

• Measure the sensor edge, scheduled output, camera exposure indication, and optical pulse against a common reference where practical.
• Use suitable test equipment and approved measurement points; avoid unsafe probing or incompatible grounding.
• Test minimum and maximum speed, acceleration, stopping, and any permitted reverse movement.
• Verify minimum spacing and bursts, not just average production rate.
• Introduce missing images, busy responses, and full queues through approved test methods.
• Test recipe changes, communications loss, controlled restart, and restoration of tracking state.
• Confirm that an invalid acquisition cannot become an ordinary inspection pass.
• Preserve the validated configuration, measurement conditions, and acceptance records.

Repeat timing checks with image processing, logging, and communications active. These tasks may not affect hardware edges, but they can affect readiness, queue service, and result delivery.

Know when a dedicated timing layer is unnecessary

For a stationary part with a generous settling interval, supervisory triggering may satisfy every requirement. Adding hardware then creates extra interfaces without improving inspection quality.

A camera-generated lighting signal may also be sufficient when there is no upstream position tracking or complex sequence. Suitable PLC capture and compare hardware can be preferable when it already meets the timing and maintenance requirements.

Conversely, precise electronics cannot repair slipping parts, inconsistent presentation, or illumination geometry that hides the defect. Resolve those limitations before tightening timing tolerances.

Start the controller configuration process with your electrical requirements, event table, and measured timing budget. If the architecture remains uncertain, send those requirements through application support so the lighting stage and precision-timing responsibilities can be evaluated separately.