Fiber Optic Blanking Sensor: What It Means and How to Set It Up
A fiber optic blanking sensor is not automatically able to ignore a physical area. First distinguish detecting a blank or passing part from evaluating a light-level window, gating a signal in time, or excluding a spatial zone. The right setup depends on which of these functions your machine actually needs.
What does “blanking” mean in your application?
Start by describing what must be detected and what must be ignored. Detecting a stamped blank is a production task; ignoring an area is a sensing function. A product name alone does not establish that one device can do both.
| Your requirement | Function to look for | What it does not establish |
|---|---|---|
| Detect a blank or ejected part | An optical event when the part crosses the sensing path. | Which area can be ignored, or whether the part fully cleared the machine. |
| Accept a received-light range | A signal window with lower and upper thresholds. | The location of the object causing that light value. |
| Evaluate only part of a cycle | A time or position gate in the sensor or controller. | Rejection of an unwanted object that appears inside the gate. |
| Ignore a fixed physical region | Documented spatial blanking, selective channels or a suitable optical arrangement. | Detection inside the excluded region. |
Ask the supplier to show the relevant manual function and demonstrate it with your fixture and target. “Teach-in,” “matrix,” and “multiple outputs” are not substitutes for that explanation.
Can a fiber optic sensor really ignore an expected area?
Only when the complete configuration supports that behavior. Teaching a normal light value can tolerate a repeatable optical condition, but it does not automatically remove a physical region from detection.
A light-level window is not a spatial mask
A useful documented example is Banner’s DF-G2: its Window SET mode places upper and lower thresholds around a presented light condition. This is a band of received-light values, not a drawing of an excluded area. The output can be configured for conditions inside or outside that band.
If an amplifier receives only one combined-light value, that value alone cannot identify where the light was blocked. A fixture at one position and an unwanted object elsewhere can produce indistinguishable readings. Adjusting the window cannot recover information the optical arrangement never measured.
Expected obstruction
An upper path is blocked.
Unwanted obstruction
A lower path is blocked.
Documented blanking still has detection limits
A manufacturer’s blanking function may be tied to a field edge, object height or a particular optical layout. SICK’s RAY26 conveyor-blanking instructions illustrate this: for the RAY26P-xxxxx5 variant, the example combines a 3 mm blanked height with a 5 mm minimum detectable object, requiring an object approximately 8 mm high. That is a photoelectric array example, not a fiber-amplifier specification or proof of arbitrary spatial masking.
For your setup, mark both the ignored region and the remaining detection region on a drawing. Then check whether the smallest required target can travel entirely through the ignored region. If it can, change the geometry or detection method; a successful teach will not close that gap.
A practical alternative is to route a narrow optical path clear of the fixture, provided every required target still crosses it. Where target paths vary, separately evaluated channels or a suitable imaging system may be needed.
Which fiber arrangement should you try first?
For opaque parts that reliably cross a known path, start by evaluating through-beam sensing. Opposed emitting and receiving fibers detect the reduction in transmitted light, rather than relying on reflection from the part’s surface.
Match the optical coverage to the part’s possible path
A narrow beam suits a tightly controlled crossing point. A slot fiber holds the opposed optics in a fixed geometry when the part fits through the opening. An array can cover a wider path, but a small target may block only a small share of the received light. Check minimum detectable object performance at the actual head separation, amplifier mode and worst target position—not just the array’s overall width.
Do not infer individually selectable channels from an array-shaped head. Confirm whether the amplifier sees a combined signal or separate spatial information. Also test the full useful field; a center-position demonstration is not evidence of coverage at its edges.
Use reflective sensing when access is one-sided
A reflective head sends and receives light from the same side. It may fit where opposed heads cannot, but the returned signal depends on the target surface and geometry. Evaluate the darkest, shiniest and most tilted acceptable parts, along with the exposed background. If those states overlap, first improve the optical arrangement rather than repeatedly narrowing the threshold.
Include the fiber cable in the installation check. Use the specified bend radius, protect it from pinching and moving tooling, and keep the sensing faces accessible for cleaning. The head, cable and amplifier work as a system; changing one can change the available signal.
How should you set up the signal and output?
Establish a repeatable optical difference first, then select the decision mode and verify the electrical output. Teach-in sets a parameter; it is not a production acceptance test.
- Define the good and fault states. Write down what the sensor must distinguish: clear path, passing part, missing part, persistent blockage, or an object in an excluded region. Decide which judgments belong in the controller.
- Observe those states in the final geometry. Record signal ranges across representative part positions, finishes and motion. A saturated reading or overlapping good/fault ranges is a reason to revisit the optics, not to declare the teach successful.
- Choose the documented mode. Use a threshold for a separable light change; use a signal window when an inside/outside range is genuinely the criterion. Spatial exclusion and cycle gating need their own supported functions. Follow the exact amplifier’s teach sequence.
- Verify the physical state at the controller input. Record clear and blocked states, amplifier indication, electrical output and PLC input bit. Light/Dark operation and PNP/NPN wiring answer different questions. Neither tells you by itself what “part OK” means in the program.
- Review response mode and timers together. A faster mode may trade optical margin for speed. Timers can delay or stretch an output; a stretched pulse may help capture but can also obscure closely spaced events. Confirm the specified behavior and measure the resulting signal.
Isolate equipment before changing wiring or entering hazardous areas. Use qualified personnel and the machine’s approved commissioning procedure for powered tests; do not bypass guards to present a sample.
Is the part missed by the optics or by the PLC?
Compare the amplifier output with the controller’s recorded event. If the optical signal does not separate the target, investigate coverage and contrast. If a valid output pulse exists but the PLC misses it, investigate input filtering, capture method and program timing.
Keep three intervals separate: optical dwell is how long the part creates a detectable light change; output pulse width is what the input receives after sensor processing; the process gate is when the controller accepts an event. A long gate does not guarantee capture of a short pulse.
Illustrative timing example—not a machine test
At 300 cycles per minute, a cycle lasts 200 ms. A 30° acceptance interval occupies about 16.7 ms only if angular speed is uniform: 200 × 30 ÷ 360.
Now assume the amplifier delivers a 2 ms pulse, while the selected controller input requires a signal to remain stable for 3 ms. That pulse does not meet the assumed filter requirement, even if it occurs entirely inside the 16.7 ms gate.
The useful next step is to measure the pulse and verify the actual input specification. This example does not calculate press stopping time, a safe operating speed or a validated timing margin.
For a missing-part decision, no event is available to trigger the alarm: the controller must determine that the required event did not arrive by a defined deadline. Check the remaining time for processing and the required machine response. The cycle rate alone does not establish that deadline.
What should a part-out signal actually prove?
A beam interruption proves an optical change at one location. It does not necessarily prove that the right part passed, that only one part passed, or that the discharge path is now clear.
Illustrative case: a washer leaves a discharge chute
The initial idea is to accept “beam blocked” as “part ejected.” But a washer lodged in the beam could leave that state active across later cycles. The program would be evaluating a persistent state, not fresh evidence of another discharge.
A more informative process requirement is a new clear → blocked → clear sequence within the assigned cycle interval. The first transition records arrival at the sensing point; the second records departure from it. If the path is already blocked, or does not clear by its deadline, the controller should apply the defined process-fault response instead of counting a new successful passage.
This still cannot distinguish the washer from scrap that produces the same sequence. If identity, double-feed detection or complete downstream clearance matters, add evidence for that specific condition. One beam cannot automatically prove it.
Two overlapping opaque blanks may create the same interruption as one. Do not sell a presence sensor as a thickness or double-sheet measurement system without documented performance for that task.
What must the final trial demonstrate?
Demonstrate both reliable detection of required events and rejection of the disturbances you intend to ignore. Use the actual head, amplifier, mounting geometry and controller settings.
- Coverage: the smallest required target is detected across every permitted path, including boundaries beside any excluded region.
- Separation: normal variation remains distinguishable from the relevant fault states under expected contamination, vibration and surface changes.
- Capture and sequence: the controller records the required transitions at operating-speed extremes and reaches its decision before the process deadline.
- Fault handling: approved simulation or controlled tests establish the response to missing events, persistent blockage, loss of power and relevant connection faults.
Save the exact head and amplifier codes, fiber length, mounting drawing, response mode, thresholds, output logic, timers and test conditions. Recheck affected evidence after changes to tooling, target, fiber routing or replacement components. A reteach should follow an understood change, not silently absorb a developing fault.
Process detection is not personnel protection. Ordinary fiber amplifiers must not be treated as safety light curtains. Safety-related blanking requires equipment intended for that function and machine-level safety assessment and validation. A working part-out signal is not evidence of a validated safeguard.
The most useful specification is therefore not “a blanking sensor,” but a clear statement of the event to detect, the region or interval to ignore, and the evidence the machine needs before it continues.
Sources and method references
- Banner DF-G2 product manual, document 193602, Rev. E — Window SET, response modes, timers and the personnel-protection warning. The cited behavior belongs to the documented model, not all fiber amplifiers.
- SICK: RAY26 array blanking — conveyor blanking, model-specific settings and minimum detectable object-height limitations.
- OMRON: Technical explanation for fiber sensors — through-beam and reflective principles, optical paths and fiber types.
- Banner: Array and slot fiber optics — wider-field and fixed-slot arrangements for small-object applications.
The equal-path illustration and numerical and washer examples are explanatory models, not reported production tests. The hero is an AI-generated editorial illustration, not an exact-model photograph or installation drawing.