Matrix Fiber Optic Sensor: When Wider Detection Helps—and When It Does Not
A matrix fiber optic sensor uses an array of optical paths to detect objects across a wider area than a single sensing point. It helps when parts do not follow one exact path. But wider coverage does not automatically provide finer detection, separate counts or position data: those depend on the head, amplifier, target and control system.
What does a matrix fiber optic sensor actually detect?
It detects a change in light across an arranged sensing area. Instead of forcing every part through one small spot, a matrix or array head spreads optical paths across a wider window. That is useful when a falling pin, ejected component or moving edge can arrive at different positions.
The fiber head carries light to and from the detection point. A compatible amplifier provides the light source, evaluates the received signal and produces the electrical output. The head defines where you can look; the amplifier and its settings help determine which light changes become a detection event.
Here, matrix means an industrial object-detection fiber arrangement—not a camera pixel array or a distributed fiber system for measuring temperature or strain. Suppliers also use names such as area, array and multi-point. Ask for the optical layout and output description, because these names do not guarantee the same functionality.
Wider coverage is not the same as more information
A combined switching output can tell the controller that something interrupted the field. It may not tell it where the object was, how many objects were present, or whether every position in a fixture was occupied. For example, a left-hand part and a right-hand part could produce the same ON signal.
Illustrative combined-output example
The same binary result cannot identify the position or quantity. This assumes a presence channel configured to switch for any of these interruptions.
If the task is “confirm a part passed,” this may be enough. If it is “confirm all six fixture positions are filled,” require independent channels or a documented inspection method that can distinguish those states.
Should you use opposed heads, a reflective array or a single sensing point?
Start with opposed heads when you can mount on both sides and the target blocks light. Consider a reflective array when only one side is accessible. Keep a single-point or slot arrangement when the part path is already controlled; more coverage is not automatically a better solution.
Opposed heads: look for an interruption
One head sends light across the gap to the other. For an opaque target, detection depends mainly on interrupting that path rather than returning light from its surface. A matte-black and a shiny metal part can therefore be much easier to compare than in reflective sensing. Transparent parts still need a specific trial: they may transmit or redirect light rather than create a clean interruption.
Both heads need rigid mounting and a clear view of the intended target envelope. Chute walls, guides and brackets must not mask the part's valid paths or create unintended optical paths.
Reflective arrays: look for a return from the target
Emitting and receiving fibers are on the same side. This solves an access problem, but now color, finish, angle, distance and background can change the returned signal. Test the weakest valid target and the strongest unwanted background; a successful test with a large white sample says little about a tilted dark part.
These are real configuration choices, not just naming variations. OMRON's E3X-HD documentation lists through-beam area fibers separately from reflective array fibers and gives different sensing distances for each amplifier mode. Choose the head-and-amplifier combination, not an isolated head rating.
When another arrangement is simpler: use a point fiber for a repeatable path, a slot sensor when the part can pass through a fixed opening, and a measuring array or vision system when the required result is position, dimensions or inspection data rather than presence.
Why can a wider window still miss a small part?
Coverage and sensitivity are different limits. A target can be inside the window yet change too little of the received light to cross the threshold. The relevant specification is the minimum detectable object under stated conditions—not simply the head width or number of visible optical points.
Think of a thin pin turning edge-on. Its position has not left the field, but its projected silhouette has become smaller. Near an edge, only part of that silhouette may overlap the useful sensing area. Increasing sensitivity might catch it, but also make dust, vibration or no-target signal variation harder to separate from a real event.
A documented window-size comparison
Banner's D10 Expert Small Object Counter manual, revision E, lists these PFCVA arrays at a 2% threshold offset. The values belong to that documented system and setting; they are not xsz sensor specifications.
| Array model | Detection window | Listed minimum object size |
|---|---|---|
| PFCVA-10X25-S | 10 × 25 mm | 1.5 mm |
| PFCVA-25X25-S | 25 × 25 mm | 3 mm |
| PFCVA-34X25-S | 34 × 25 mm | 4 mm |
The useful lesson is not a universal scaling formula. It is that a larger window in this series does not preserve the same minimum-object capability. Before choosing a wider head, compare its smallest-object specification with the smallest projected target you actually need to detect.
Test that target across the center, edges and corners of the allowed path, at the intended head separation and final amplifier setting. A sensitivity claim based on a centered stationary sample is not yet evidence for every moving orientation.
Can it count fast-moving parts without misses or double counts?
Only if each part produces a distinguishable event and the complete system captures it. The array must detect the part, recover between parts, and send a usable signal through the amplifier, PLC input and counting logic. Average parts per second alone does not establish that.
Check both the target time and the clear gap
For a simple constant-speed, thin sensing-plane estimate:
Use the target length along travel to estimate occupancy time, and the clear spacing along travel to estimate gap time. These are geometric estimates, not guaranteed output pulse widths. The optical field's depth, trajectory, threshold, response mode and timers can change the actual signal. A freely falling or rotating object may also have a changing local speed.
Illustrative example: a comfortable rate can hide a short gap
Assume a 4 mm target travels at 1.2 m/s with a 6 mm clear gap. The geometric times are 3.33 ms occupied and 5.00 ms clear; the regular stream is 120 parts/s.
Now reduce only the gap to 1 mm. The occupied time remains 3.33 ms, but the clear interval falls to 0.83 ms. If the receiving input requires a valid 1 ms inactive interval, the actual electrical waveform must be checked: the geometric gap already flags a separation risk. That 1 ms requirement is an assumption for this example, not a standard PLC value.
A faster amplifier alone may not solve the problem. The input filter, event capture and machine's part separation must also support the short interval.
Do not use a timer to hide an unresolved counting problem
Two touching or overlapping parts can look like one uninterrupted object. One bouncing part can cross the field twice. A pulse stretcher can make a brief event easier to capture, but can also reduce the remaining clear interval before the next event. Filtering may suppress a disturbance while also suppressing a real closely spaced part.
Compare physical parts with recorded events before changing timers. Use qualified, safely arranged signal capture when events are too fast for indicator lights. Record minimum active and inactive durations at the sensor output and controller input, then check the input filter, capture mode and program logic. Do not equate a visible LED flash with a correctly counted part.
What must the amplifier and controller do with the signal?
They must produce and interpret the information your task requires. Confirm head compatibility, the chosen response mode, output type and channel purpose before wiring. NPN or PNP describes the electrical interface; light-on or dark-on describes which optical state activates it.
Do not assume two outputs mean two independently observed zones. For example, Banner's D10 counter uses a counting output and a separate health output. Neither is a beam-position map. If a supplier proposes several independently arranged tips, ask how each tip's status reaches the controller.
A published application: detection plus machine-cycle context
Yan and colleagues' 2024 cassava-planter study used an opposed matrix fiber sensor, amplifier and rotary encoder. The sensor observed cut stems passing below the cutting mechanism; the encoder supplied cycle position. The controller interpreted optical pulse timing within the expected cutting window to identify missed discharge or sticking.
The transferable lesson is that “light changed” becomes a useful machine judgment only when its timing and expected sequence are understood. This is a third-party research example, not an xsz sensor installation or a performance guarantee for another part stream.
Process detection is not personnel protection. An ordinary matrix fiber sensor must not be substituted for a safety light curtain. Monitoring parts and safeguarding people require different device evidence and system validation.
What should you verify before choosing and installing a system?
Specify the target envelope, required result and timing first. Then evaluate one defined head, fiber route, amplifier configuration and receiving input together. This keeps a successful demonstration from being mistaken for proof of the final installation.
- Define the hardest target and the full path. Provide the smallest projected size, material, transparency, rotation, lateral movement and mounting access. State whether the machine needs one presence result, an individual count or independent position information.
- Record the actual configuration. Keep complete head and amplifier model numbers, sensing distance, response setting, teach method, threshold, timers and output assignment. Verify the fiber connection and supply/input requirements from the matching manuals; do not assume a familiar connector means compatibility.
- Set up without creating unintended motion. Follow equipment isolation procedures before mounting or changing connections. Fix the heads rigidly, keep the window accessible for cleaning, and route fibers within their specified bend and flex limits. A fixed-installation bend rating does not establish suitability for repeated movement.
- Challenge detection and release. Test the weakest target positions, fastest local speed, smallest gap and relevant no-target conditions. For a reflective system, include the background and surface variations. Recheck after guards close and the fiber route reaches its final position.
- Investigate any mismatch at the right stage. If received light barely changes, inspect optical coverage, alignment and target presentation. If the amplifier switches but the PLC does not count, investigate the electrical signal and input acquisition. If one physical part creates several events, examine bounce and waveform behavior before changing filters.
Finish with a saved baseline: mounting dimensions, settings, representative samples, operating conditions and the comparison between actual parts and recorded events. Agree on the trial duration and acceptable missed/double-count performance for the application instead of adopting an arbitrary universal margin.
The selection rule is simple: choose a matrix fiber system when wider optical coverage resolves an uncertain target path and the documented output supplies the information you need. Confirm smallest-object detection and event separation independently; window width cannot stand in for either.
Sources and method references
- Banner D10 Expert Small Object Counter — manual 146132, revision E: pages 3–4 document the array comparison, 2% threshold condition and counting/health outputs; page 4 also states the personnel-protection restriction.
- OMRON E3X-HD specifications — Area Detection Models: separate through-beam and reflective heads, with sensing distances tied to operating mode.
- OMRON photoelectric sensor principles: transmitted-light interruption versus reflected-light detection.
- Banner array and slot fiber optics: array coverage and fixed-slot configuration examples.
- Yan et al. (2024), cassava-planter monitoring study: sections 2.2.1–2.2.3 and Figure 2 support the described application and machine-cycle interpretation.