Sensor for Conveyor Belt Applications: Counting, Presence and Position Checks

Which sensor should you try first on a conveyor?

For separated opaque cartons or packages, start with a photoelectric sensor. A through-beam arrangement is a strong first trial when both sides are accessible; retro-reflective sensing needs wiring on only one side but still needs an opposite reflector. When neither an opposite receiver nor a reflector can be installed, consider a background-suppression model.

The useful choice is the sensing arrangement, not just the housing shape or advertised range. Use the conditions below to narrow a trial—not to approve a model without testing.

Ordinary diffuse sensing may be sufficient when target color, distance and background remain controlled. Do not assume that extra features fix weak optical contrast. Alignment and usable signal margin matter before communication options.

These are process-detection choices. An ordinary product sensor is not a personnel-protection device; conveyor guarding requires a separately designed and validated safety function.

Are you checking presence, counting items or locating an edge?

Define the event in one sentence before selecting the sensor: “Count one sealed pack as its front edge crosses this point,” for example. A presence signal, a count and a position trigger can use similar hardware but require different evidence.

Presence needs a dependable occupied state

For a transfer zone, the question may simply be whether a product is there. Establish which product positions must register as occupied and which belt, guide or background conditions must remain clear. A broad detection zone can be useful here, even when it would be too ambiguous for accurate edge positioning.

Counting needs one identifiable event per unit

First define the unit: bottle, multipack, carton or carrier. Then choose one transition and ensure the signal can reset between units. A counter that increments every PLC scan while the input is active counts scan cycles, not products. Reverse travel can also count the same item again unless direction or tracking logic handles it.

Position needs a repeatable feature and timing

A registration trigger should use a defined edge rather than a flap that can bend or a surface that moves sideways. Sensor switching variation, mechanical guidance and controller/actuator timing all affect where the machine acts. At variable speed, a fixed time delay represents different travel distances; consider encoder-based tracking when the task requires distance-based action.

Product presence also does not prove belt motion. A motion check needs feedback from the relevant moving belt or roller; a motor-run command alone cannot reveal every stopped or slipping conveyor condition.

Why can a conveyor miss counts even at a low item rate?

Because the brief clear gap may be much shorter than the time a product occupies the beam. The sensor and controller must recognize both states. Average products per second alone does not describe the shortest signal they must capture.

Estimate the occupied time and the clear time separately

For constant speed, a narrow beam and a simple opaque target crossing squarely, use the following first estimates. Measure length along the direction of travel—not across the belt.

These are geometric estimates, not guaranteed electrical pulse widths. Beam size, switching threshold, object angle and sensor operate/release delays change the observed transitions. Measure the actual output at the fastest speed and shortest separation.

Check the capture method, not just the PLC scan time

Separate sensor response, input pulse qualification and program capture. A fast sensor cannot recover a gap removed by filtering. Conversely, a valid pulse reaching the input may still need hardware counting or event capture if cyclic program sampling cannot reliably observe it.

Siemens' S7-1200 documentation explains that short high or low pulses can be removed by input filtering, and that pulse catch operates after that filter. A high-speed feature therefore does not automatically bypass an unsuitable filter. Check the manual for the actual CPU, input channel and configuration before changing settings.

Pulse catch is not automatically a multi-event counter, either: the documented S7-1200 function reads only the first pulse when several arrive within one scan. Choose counting or interrupt handling that can retain every required event.

Do not blindly add all delays into one universal pass/fail number. Nor should you extend every pulse: an off-delay that bridges the gap can merge counts. Keep sufficient noise rejection while preserving the required transitions, then verify the complete chain under operating conditions.

Where should you mount the sensor to get one count per product?

Choose a guided section where each product presents one continuous feature and a repeatable gap. For cartons, a consistent side-wall height is often easier to interpret than loose top flaps, handles or cutouts. For tiny parts, ensure the beam intersects the smallest valid target across its permitted movement.

Leave room for both the target and the clear state

A guide rail, belt edge or bracket must not remain in the sensing path after the product leaves. With background suppression, test the nearest background and farthest product positions, including belt movement. With a reflector, align for a strong return rather than stopping at the first point where the output switches.

Balluff's conveyor-selection discussion highlights this distinction: a switching indication can coexist with poor optical alignment and reduced margin against contamination or mechanical movement. Secure the bracket, restrain the cable and leave access for cleaning and replacement.

Do not try to fix an unobservable gap with software

If two touching products present one continuous silhouette at the selected beam height, a faster sensor cannot reveal a boundary that does not change its optical input. Separate the products mechanically, detect a genuinely distinct feature, or evaluate a suitable profile/vision system. If products overlap across lanes, one beam can also merge simultaneous passages; independent lane sensing may be necessary.

Some touching shapes still expose a usable neck or top-profile gap. Test that feature over every allowed orientation rather than assuming that touching always means uncountable—or that a camera automatically solves the problem.

What changes for clear bottles and shiny packaging?

Clear objects may transmit most of the beam, while shiny surfaces may return light in an unexpected direction. Do not qualify either using only a matte carton. For transparent products, trial a dedicated clear-object mode; for glossy opaque packaging, evaluate the specified polarized retro-reflective arrangement or another method suited to the surface.

Use the actual production variations: empty and filled bottles, rotated seams, applied or missing labels, clear and printed film, wet surfaces and product tilt. Teach against the intended stable background or specified reflector, then check both product-present and clear states without repeatedly retuning for each favorable sample.

A transparent bottle detected by its label is not evidence that the unlabeled bottle is detectable. Likewise, a sensor that reliably detects the outer shrink wrap may be checking a multipack, not counting its individual containers.

Does a continuously blocked sensor mean the conveyor is jammed?

Not by itself. A blocked beam can mean a jam, normal accumulation, an intentionally stopped product or an obstructed optical path. Interpret duration together with the conveyor's operating state and expected flow.

For a jam trial, define when blockage is abnormal: expected movement, maximum permitted occupied time, accumulation mode and the downstream-ready condition. Include long products and normal start/stop sequences. If speed varies, a single fixed blockage timeout may mistake slow normal passage for a fault.

Also decide how to distinguish an empty running belt from lost product detection. One permanently clear input cannot make that distinction alone; expected upstream arrivals or other process feedback may be needed.

What should you verify before accepting the conveyor sensor?

Run a known quantity of representative products and compare physical passages, sensor output and the controller's recorded events. Begin with a slow controlled pass, then test the fastest speed and shortest gaps. Follow the site's safe commissioning procedure; do not reach into a moving conveyor or bypass guarding to adjust the sensor.

Indicator LEDs are useful for slow checks, but not a substitute for a waveform or controller trace when the suspected event lasts only milliseconds. Where electrical measurements are needed, use appropriately qualified personnel and equipment.

Repeat the trial across target sizes, finishes, lateral positions, cleaning conditions and normal restarts. Define the allowable count error or position tolerance before the test; a single successful pass does not establish long-term reliability.

Record the final model, reflector or fiber unit, mounting dimensions, teach settings, output mode, input filter, capture logic and tested conditions. For washdown or temperature exposure, evaluate the complete installed assembly using the sensor protection-rating guide. This record makes a replacement reproducible instead of starting the adjustment process again.

A good conveyor sensor choice is the simplest arrangement that separates the required events reliably—and preserves those events all the way to the machine's decision.