Intermittent NO BOTTLE Alarm on a Filling Machine
An 8-head filling machine uses a PLC with 24 VDC input cards. The bottle-present sensor at station 4 is a three-wire PNP photoelectric type wired to input X12 through a 28 m cable that shares a tray with the 3-phase motor feeders for the conveyor. The machine trips a NO BOTTLE alarm three to five times per shift, always when the filler motor starts under load. With a multimeter, the sensor output measures 23.7 V when a bottle blocks the beam and 0.9 V when the beam is clear. The PLC input LED flickers faintly during the trips. The sensor 0 V wire is landed on a terminal block, but with the machine running a meter shows 0.4 V between that terminal and the PLC input common. The coils of three solenoid valves near the PLC have no suppression devices. Maintenance has already replaced the sensor once with no change.
1. Is the sensor the most likely faulty item? Use the voltage readings and the history to justify your answer, and name the condition the 0.4 V reading points to.
Read the worked answer
- A healthy PNP sensor on a 24 V supply switches to within 1–2 V of supply and leaks well under 1 V when off. 23.7 V on and 0.9 V off are both normal.
- A sensor replaced with no change in symptoms is strong evidence against the sensor.
- The 0.4 V between the sensor 0 V terminal and the PLC input common is the key finding. The PLC input compares its terminal voltage against its own common; if the two 0 V points sit 0.4 V apart, every logic level is shifted and small transients push the input across its threshold.
- The 28 m run in the same tray as motor feeders, plus three unsuppressed solenoid coils, supplies the transient energy. The fault appears only when the filler motor starts under load, which is the moment of highest dV/dt in the tray.
- The faint LED flicker confirms short spikes rather than a steady state.
Conclusion: this is a noise and reference-ground problem with a wiring contribution, not a sensor failure.
2. Describe the measurements and temporary tests you would use to prove electrical noise coupling rather than a genuine sensor or PLC input failure. State the expected result of each test.
Read the worked answer
- Scope or fast logging at X12. Capture the input while jogging the filler motor. A genuine bad sensor gives a clean switch; noise gives spikes above the ON threshold lasting microseconds to a few milliseconds. Analog meters and averaging DMMs hide this.
- Test cable outside the tray. Temporarily run a short cable from the sensor to X12 away from the motor feeders. If the alarms stop, coupling is confirmed.
- Signal substitution. Disconnect the sensor and feed X12 from a clean 24 V source through a switch. If the input still trips when the motor starts, the noise is entering the input wiring or the common, not the sensor.
- Coil test. Disconnect the three solenoid coils one at a time. If the trips stop when one coil is removed, its unsuppressed turn-off spike is the source.
- Common-potential check. Measure 0 V bus to PLC input common and 0 V bus to earth with the machine running. Target below 0.1 V. The present 0.4 V must be explained and removed.
- Type check. Confirm the card is a sinking-input type for a PNP sensor and that the sensor supply 0 V and the card common are the same point, not two separately landed terminals.
Only after these tests would I touch the sensor again.
3. State the permanent corrective actions in priority order, and explain how you would verify the fix and what you would change if the correct filter setting would make the input too slow.
Read the worked answer
- Single reference. Land the sensor 0 V and the PLC input common on the same 0 V bus at one star point; retorque and re-measure, target below 0.1 V.
- Separate and shield. Replace the 28 m run with shielded cable, earth the shield at the panel end only, keep it at least 200 mm from motor feeders or use a metallic divider, and cross at 90° where separation is impossible.
- Suppress the sources. Fit RC snubbers or flyback diodes across the three solenoid coils, or replace the valve plugs with suppressed types.
- Terminations. Re-terminate any ferrule that shows corrosion, and check the sensor cable for a nick where it passes the tray edge.
- Input filtering. Increase the digital input filter to 10–20 ms if the process allows.
- Run two full shifts with alarm logging; zero nuisance trips.
- Re-measure the common potential and capture the input at X12 with the motor starting.
- Record the changes on the drawing so the next technician does not undo them.
Calculate the window: at line speed, the bottle stays in the beam for a known number of milliseconds. If a 20 ms filter eats too much of that window, do not simply lower the filter. Fix the coupling instead (shield, separation, suppression), or change the sensing scheme, for example a sensor with a shorter response time mounted closer to the PLC with a remote I/O block in the station panel.