Application Overview
An injection molding machine ejects a transparent polystyrene-type part (similar optical density to clear disposable cutlery) that falls free out of the mold. A sensor mounted 4-6 in. (100-150 mm) below the ejection point must confirm each part has cleared the tool. If no part is detected within the expected window, the machine cycle is inhibited to prevent a stuck part from being crushed on mold close.
Three characteristics drive the sensor choice:
- Target is translucent, not opaque. A clear part passes 80-95% of incident light in a standard diffuse or non-polarized retroreflective setup, so a conventional sensor sees almost no signal change.
- Target is in free fall. The beam is broken for tens of milliseconds only. Sensor response time and PLC input capture become the limiting factors, not sensing range.
- Target position varies. Parts tumble and drift laterally as they drop. The detection zone must be wider than the mechanical drop scatter, or the sensor will report false "no-part" faults.
Sensing Method Selection
Sensing method matters far more than range at these distances. Compare the realistic candidates:
| Method | Behavior with clear plastic | Verdict for this application |
|---|---|---|
| Diffuse (proximity mode) | Returns very little light from a transparent surface; excess gain collapses. Detection depends on a specular glint that appears only at certain part angles. | Not recommended for tumbling clear parts. |
| Standard retroreflective (non-polarized) | Beam passes through the part twice with only a small attenuation. Also false-triggers on shiny mold or machine surfaces, which mimic the reflector. | Marginal. Requires very high sensitivity, prone to drift. |
| Polarized retroreflective (clear-object variant) | Emitter light is polarized; only the corner-cube reflector rotates the plane of polarization back to the receiver. Attenuation and slight depolarization from the clear part are both detected. Purpose-built clear-object models add high excess gain and tight hysteresis. | Preferred. Single-side mounting, single wiring run, reflector on the opposite side of the drop chute. |
| Through-beam (opposed) | Highest excess gain of any style; a clear part still produces a measurable dip, but a general-purpose through-beam pair may saturate the receiver and ignore it. Clear-object through-beam models with adjustable gain work well. | Good alternative when the drop chute is wide and both sides are accessible. |
| Ultrasonic | Independent of optical transparency; detects the acoustic boundary. | Backup option, but response time (typically slower than photoelectrics) and beam spread make free-fall capture less reliable. |
Because the required range is short, filter catalog selections to roughly 0.05 m - 1 m sensing distance and start with the clear-object detection family rather than the general-purpose photoelectric list. Vendors publish a dedicated clear-object sensor category precisely because the excess-gain curve and hysteresis are tuned differently from standard models.
Light Emission: Visible Red vs Infrared
| Emission | When to use |
|---|---|
| Visible red | Default choice. The visible spot lets you aim the beam at the reflector and confirm alignment across the drop path without instruments. Use it unless visible light is a problem. |
| Infrared | Use only where a visible light point must be avoided - operator glare concerns, or vision/camera systems that must not see stray illumination. IR also penetrates some dirty or dusty environments better. |
| Laser (where offered) | Tight, well-defined spot for small parts; less tolerant of lateral drift in a drop chute. |
For a part-drop detector under a molding machine, visible red, polarized retroreflective is the practical starting point. Alignment is done by eye at commissioning and re-checked in seconds after any maintenance.
Free-Fall Timing Math
Verify the sensor and the PLC can actually catch the event. Free-fall velocity at the beam plane:
v = sqrt(2 * g * h) with g = 9.81 m/s^2
h = 0.10 m (4 in.): v = sqrt(2 * 9.81 * 0.10) = 1.40 m/s
h = 0.15 m (6 in.): v = sqrt(2 * 9.81 * 0.15) = 1.72 m/sBeam-break duration for a part of vertical dimension L crossing at velocity v:
t_break = L / v (ignoring beam diameter, which slightly increases t_break)
L = 0.030 m at 1.72 m/s -> 17.4 ms
L = 0.050 m at 1.72 m/s -> 29.1 ms
L = 0.100 m at 1.72 m/s -> 58.1 msDesign rules that fall out of the math:
- Sensor output response time must be an order of magnitude below
t_break. Typical photoelectric response times in the sub-millisecond to few-millisecond range are adequate for a 17 ms event; confirm the response time on the specific model's datasheet. - Mount the sensor as far below the mold as the guarding allows only if you need lateral spread coverage; the closer to the mold, the slower the part and the longer the beam break. Mounting nearer the top of the 4-6 in. window increases
t_breakand improves capture margin. - A standard PLC input filter of 8-12 ms plus a 10-20 ms scan can straddle a 17 ms pulse. Use a fast/pulse-catch input, a hardware latch, or a sensor with an output pulse stretcher.
Wiring and Output Configuration
| Parameter | Selection guidance |
|---|---|
| Output type | Match the PLC input card. PNP (sourcing) for sinking DC input modules common on IEC-style hardware; NPN (sinking) for sourcing input modules. Mismatch is the most common commissioning failure. |
| Output mode | Light-ON vs Dark-ON. For a beam-break part detector, Dark-ON gives a positive signal when the part interrupts the beam. Many clear-object models offer a selectable or dual output. |
| Termination | Pre-wired cable (fixed length, fewer failure points) or M8/M12 quick-disconnect (faster field replacement under a molding machine where access is poor). QD is preferred for maintainability. |
| Housing | Choose an IP-rated housing appropriate for mold release spray, oil mist, and washdown near the drop area. |
| Reflector | Use the corner-cube reflector specified by the sensor manufacturer for polarized operation. A generic reflective tape will not rotate polarization correctly and will kill the polarized function. |
Mounting and Commissioning
- Mount the sensor and reflector on opposite sides of the drop path, square to each other, 4-6 in. below the mold parting line and clear of the ejector stroke.
- Angle the optical axis 10-15 degrees off perpendicular to any flat machine surface behind the drop path so specular returns do not reach the receiver.
- Align the visible red spot on the center of the reflector. Confirm the stability/margin indicator LED shows a strong margin with no part present.
- Set sensitivity: with the beam clear, back the gain down from maximum until the output is still stable, then hold a sample part statically in the beam and confirm the output switches. Repeat with the part rotated to its worst-case orientation (thin edge to the beam) - this is the true worst case, not the flat face.
- Re-verify with the part held at the far left and far right of the possible drop scatter. If the thin-edge or off-center case fails, move to a clear-object through-beam pair or widen the beam with a larger reflector.
- Blow off the lens and reflector, then confirm the margin indicator still reads healthy. Add a scheduled lens-cleaning task - mold release film on the optics is the dominant long-term drift mechanism.
PLC Logic for Part-Drop Verification
Do not read the sensor bit directly in the cycle interlock. Latch it, window it, and require a fresh event every cycle.
// Pseudo-logic, one shot per mold cycle
MoldOpen_Complete -> RESET PartDetected_Latch
-> START DropWindow_TON (PT = 2000 ms)
Sensor_DarkON (fast input / pulse catch)
AND DropWindow_TON.TT -> SET PartDetected_Latch
DropWindow_TON.DN AND NOT PartDetected_Latch
-> SET NoPart_Fault -> INHIBIT MoldClose, alarm operator
PartDetected_Latch AND DropWindow_TON.DN
-> PERMIT next cycle- Open the detection window only after the mold is fully open and ejection has started, so a swinging robot arm or falling flash outside the window cannot satisfy the interlock.
- Size the window (2000 ms above is a starting value) from the measured worst-case time between the ejector-forward signal and the part clearing the beam. Verify on the actual machine.
- If the mold is multi-cavity and every cavity must clear, one beam-break is insufficient - count edges or use a light grid across the full drop width.
- Add a beam-blocked fault: if the sensor stays dark for more than a few seconds, a part is hung in the chute. That condition must also inhibit mold close.
Failure Modes to Design Around
| Symptom | Likely cause | Fix |
|---|---|---|
| Intermittent missed parts, random cavities | Part passes edge-on; too little attenuation at that orientation | Increase gain, switch to clear-object through-beam, or add a second beam at 90 degrees |
| Detection degrades over days/weeks | Mold release and plastic dust film on lens/reflector | Scheduled cleaning; select a model with a margin/stability output and alarm on low margin |
| Output chatters with no part present | Specular reflection from a polished machine surface acting as a second reflector | Use polarized retroreflective, angle the axis off perpendicular, mask the background |
| Sensor LED flashes but PLC never sees the bit | Input filter or scan time longer than the beam-break duration | Move to a fast input, enable pulse-catch, or use the sensor's output pulse-stretch/off-delay setting |
| Sensor works on bench, fails on machine | NPN/PNP mismatch or shared 24 V supply noise from the heater/drive circuits | Confirm output type against the input card; supply the sensor from a clean 24 V rail with separate routing from heater and drive cables |
FAQ
What type of photoelectric sensor detects clear plastic parts?
A polarized retroreflective sensor from a clear-object detection family is the standard choice. The polarizing filter plus a corner-cube reflector lets the sensor see both the small attenuation and the depolarization caused by transparent plastic, which a diffuse or non-polarized retroreflective sensor cannot resolve reliably.
Should I use infrared or visible red light for clear object detection?
Use visible red as the default - the visible spot makes alignment to the reflector a two-minute job and lets you re-verify aim after maintenance by eye. Choose infrared only when a visible light point must be avoided, such as near camera or vision systems or where operator glare is a concern.
How fast is a part falling 6 inches out of a mold, and can the sensor keep up?
From rest, v = sqrt(2 * 9.81 * 0.15) = 1.72 m/s at 150 mm of drop. A 50 mm tall part breaks the beam for about 29 ms, which is well within typical photoelectric response times; the real risk is the PLC input filter and scan time, so use a fast/pulse-catch input or an output pulse-stretch setting.
Can I use a photoelectric part-drop sensor as a machine safety device?
No. A general-purpose photoelectric sensor is a process interlock, not a safety-rated component. If the function protects personnel or forms part of a rated safety function, implement it with safety-rated devices and architecture validated against the machine's applicable functional safety standard.
Do I need NPN or PNP output?
Match the PLC input module: PNP (sourcing) sensors drive sinking DC input cards, NPN (sinking) sensors drive sourcing cards. Confirm the card's common polarity on the wiring diagram before ordering - output-type mismatch is the most frequent cause of a sensor that lights its LED but never registers in the PLC.