Slot Sensors for Small Part Detection: How to Choose and Avoid Missed Counts

When is a slot sensor the right choice for small parts?

A slot sensor, also called a fork or U-shaped photoelectric sensor, is a useful first option when a guided, opaque feature can pass between its two arms. The emitter and receiver face each other inside one housing. You do not have to align two separate sensor heads, but you still have to align the part’s path with the beam.

This suits tasks such as counting separated terminals, detecting a metal flag or checking a feature on an indexed component. It is less suitable when the part cannot enter the opening, wanders around the sensing area or fails to interrupt enough light.

For example, Banner’s published application for counting hardware falling from a bowl feeder uses an LX light array across the passage. That illustrates a useful change of approach when a single narrow beam cannot cover the drop path; it does not establish that overlapping parts will be counted individually.

This guide concerns process detection. An ordinary slot sensor is not a personnel-protection device; a safeguarding function requires appropriately safety-rated equipment and machine-level validation.

What is the smallest part the sensor can actually detect?

Use the minimum detectable object specification and its test conditions—not the slot width, switching repeatability or housing size. A part may fit comfortably inside the fork yet block too little of the effective beam to switch the output.

Read the shape and conditions behind the number

The relevant dimension is the feature presented to the beam. A rod diameter, a rectangular test piece and a thin edge are different optical targets. Orientation, position within the slot, transparency and sensitivity setting can change the result.

For instance, OMRON lists a 5 mm slot and a minimum opaque object of 2 × 0.8 mm for the EE-SX670. The rectangular test-object value is not a promise to detect a 0.8 mm diameter wire. The same datasheet’s differential-distance value describes switching behavior, not the smallest detectable object.

Do color, shine and transparency matter?

For an opaque target that adequately interrupts an opposed beam, surface color and reflected brightness are usually less important than with diffuse sensing. But clear plastic, thin translucent material and small holes may let enough light reach the receiver to prevent a clean switch.

Try a consistently opaque feature if the part has one. If that feature cannot be guided through the beam, evaluate another sensing arrangement. A laser label or a higher sensitivity setting alone does not prove that a clear or very thin part is detectable.

How much slot width and depth do you need?

Choose the opening for the part’s full moving envelope: its size plus permitted tilt, sideways movement and mounting variation. Then check where the beam sits inside that opening. Slot width, available depth and beam position are separate dimensions.

Do not assume the beam is halfway down the fork. As a documented example, Banner’s SL30 has a 30 mm-wide, 45 mm-deep opening, with the beam 10 mm from the outer edge. Use the selected model’s drawing to locate that line on your machine layout.

Mount the sensor rigidly relative to the part guide, leave access for cleaning and protect the cable from movement or snagging. Check the widest and most tilted permitted part before reducing the gap. Excessively tight clearance can turn a detection problem into contact damage or a jam.

Will the PLC capture both the part and the gap?

It must recognize two states: the interruption made by a part and the clear interval before the next one. A high switching-frequency rating does not, by itself, prove that the sensor and controller will preserve your shortest pulse or gap.

Estimate the event duration before comparing ratings

For a guided target moving at approximately constant speed, divide its dimension along the direction of motion by speed. Do the same for the minimum clear gap. With millimeters and meters per second, the numerical result is in milliseconds.

Illustrative example: a 1.2 mm feature and a 3.0 mm clear gap moving at 0.5 m/s give about 2.4 ms of interruption and 6.0 ms of clear time. The 8.4 ms pitch corresponds to roughly 119 parts per second.

Those are starting estimates, not guaranteed electrical pulse widths. Beam geometry, switching thresholds and the sensor’s operate/release delays change the signal. Compare the exact sensor and input specifications, then inspect both high and low intervals at the connected input under maximum-speed conditions.

A scan-based counter can miss a perfectly visible pulse

An ordinary cyclic input read only reports the state present when sampled. A counter instruction that detects edges in those sampled states cannot reconstruct a transition that never reached its program.

Pulse catch, an interrupt and a high-speed counter are different capture methods. Check which function the actual input channel supports. For example, Siemens documents that S7-1200 pulse catch operates after the input filter and reads only the first pulse when several occur within one scan. Enabling that latch is therefore not equivalent to counting every event.

Select a supported counting path and filter setting that preserve both states. If pulse extension is used, check that it does not merge neighboring events. Do not use a universal “sensor delay + scan time” calculator as a production pass/fail test; the actual capture architecture matters.

Why can one part create two counts—or none?

A basic counter counts selected signal edges, not physical part identities. If one part interrupts the beam twice, it may produce two accepted events. If two parts pass without a detectable clear interval, they may appear as one.

Choose a feature that makes one clean interruption

Consider a forked terminal moving sideways through the beam. At the tips, the light may be blocked by one tine, clear through the opening, then blocked by the other tine. If the beam instead crosses a continuous shank, that same part can produce one uninterrupted blocked interval. This is an illustrative geometry change, not a tested customer result.

Holes in washers, slots in stampings, part bounce and reverse travel can cause similar extra transitions. Change the sensing position or stabilize the passage before trying to solve every duplicate with software delay.

Do not hide an inconsistent feed with a long delay

A debounce or lockout may suppress unwanted edges, but it can also discard the next legitimate part when spacing shrinks. Define the shortest valid part-to-part interval and the longest possible disturbance first. Where those intervals overlap, improve the mechanics or use additional position information rather than assuming a single timeout can distinguish them.

For touching, nested or overlapping parts, create separation upstream or evaluate a sensing system that can resolve the required distinction. Increasing sensitivity cannot make an absent optical gap appear.

Is the sensor missing the part, or is the input missing the signal?

Compare the optical response, physical output and controller record. This separates a target or mounting problem from an electrical or timing problem before you replace the sensor.

First confirm output type and operating logic

PNP/NPN identifies the electrical output arrangement and must match the receiving input. Light operate/dark operate identifies the optical state that activates it: for a conventional opposed fork, light operate is active with a clear beam; dark operate is active with an interrupted beam. They are independent choices.

Check the complete model suffix, supply range, connector pin assignment and the input’s required levels. Do not assume that two similar fork housings have interchangeable wiring. An indicator may show received light or output state, depending on the model; a visible blink is not proof that a short pulse reached the PLC.

Use qualified personnel and suitable instruments for electrical measurements. Isolate hazardous motion before adjusting the sensor, guides or wiring; perform energized measurements only under an approved safe procedure.

What should a sample trial prove before you commit?

The trial should show that every permitted part produces the intended event through the complete sensing and counting chain. A hand-passed sample or catalog frequency alone is not enough.

1. Geometry remains valid. Test the smallest feature, widest envelope and extreme allowed positions with the final bracket and guide. Check that the target crosses the beam without contact.
2. Both optical states remain distinct. Include representative finishes, material batches and foreseeable contamination. Record the sensitivity or teach settings and the clear/blocked behavior.
3. The shortest events reach the controller. At maximum speed and minimum spacing, record high and low pulse durations at the input and verify the selected filtering and capture method.
4. The physical count and recorded count agree. Use a known quantity and representative sequences, including starts, stops and permitted movement variation. Record the sample count, duration, conditions and any errors—not just “passed.”

Set the acceptance criteria for the application before testing. A run without errors supports the conditions tested; it does not establish a universal zero-miss guarantee.

For a useful supplier discussion, send a dimensioned part/path sketch, actual samples, maximum speed, minimum gap, available opening, environmental conditions and the exact PLC input or counter model. Ask for the proposed sensor’s drawing, minimum-object test conditions, output specification and timing data.

The practical choice is the fork that fits the motion, detects the critical feature and delivers a countable signal—not simply the narrowest opening or fastest headline specification.