Sensors and Instrumentation

Every field device answers two separate questions: what physical thing does it detect, and how does it report what it found? Sensor problems get solved fast when you keep those apart, because a device can be detecting perfectly and signaling badly, or signaling a clean number that describes nothing real.

Why it matters on the job

A control system is only as truthful as its instruments. A PLC executing perfect logic on a lying pressure transmitter will do exactly the wrong thing, confidently, and the alarm history will look normal while it does it.

Field devices are also the part of the control system that lives in the plant, exposed to vibration, washdown, heat, coating, and forklifts. They fail more often than processors do, and they fail in ways that look like other faults.

Discrete sensors: is something there?

Inductive proximity sensors detect metal by the effect it has on an oscillating field. They are sealed, have no moving parts, and are the default for detecting machine parts. Their rated sensing distance assumes a standard steel target, and other metals reduce it, so a sensor that reaches 8 mm to mild steel reaches considerably less to aluminum or brass. Check the manufacturer’s reduction factors rather than assuming the catalog number.

Capacitive proximity sensors detect almost anything, including liquids and powders through a container wall. That sensitivity is also their weakness: product buildup, moisture, and humidity change what they see.

Photoelectric sensors come in three arrangements, and the order matters.

• Through-beam puts the emitter and receiver on opposite sides of the gap. Longest range, most reliable, and needs power and alignment in two places.
• Retroreflective puts emitter and receiver in one body with a reflector opposite. Middling range and one wiring point, but a shiny target can fool it.
• Diffuse puts everything in one body and reads light bounced off the target itself. Simplest to install, shortest range, and the most affected by target color, finish, and background.

Limit switches are mechanical, and that is both their virtue and their weakness. Their state is unambiguous and easy to prove with a meter, and their contacts and actuators wear out.

How a discrete sensor signals

A three-wire DC sensor has power, common, and a switched output, and that output is either PNP (sourcing, it switches the positive supply to the load) or NPN (sinking, it switches the load to common). The input module has to match. Get it wrong and the point never turns on, with no error message anywhere.

Two-wire DC sensors wire in series with the load, which means they must pass a small leakage current even when off, to keep their own electronics alive. On a modern high-impedance input, that leakage can be enough to hold the input on permanently. The symptom is an input that never turns off, and the cause is not a failed sensor.

Analog instruments and the 4 to 20 mA loop

Analog measurements (pressure, level, temperature, flow) are usually carried as a current between 4 and 20 milliamps. Current is used rather than voltage because current is unchanged by the resistance of a long cable, so the reading does not drift with wire length or a slightly corroded terminal.

The reason the scale starts at 4 mA rather than 0 is worth knowing. It is called a live zero. A transmitter reporting the very bottom of its range sends 4 mA, and a broken wire, a dead transmitter, or a lost power supply sends 0 mA. The system can therefore tell “the tank is empty” from “I have no idea,” which a 0 to 20 mA scheme cannot.

Worked example: reading and checking a loop

A pressure transmitter is ranged 0 to 300 psi and sends 4 to 20 mA. The HMI shows 142.5 psi, the local gauge on the same tap reads 150 psi, and an operator wants to know which one to believe. Clamp a meter on the loop: it carries 11.6 mA.

Convert the current to pressure. The span is 20 − 4 = 16 mA for 300 psi.

Fraction of span = (11.6 − 4) ÷ 16 = 7.6 ÷ 16 = 0.475.

Pressure = 0.475 × 300 = 142.5 psi.

The signal starts at 4 mA so that zero milliamps can mean something different from zero pressure

That matches the HMI exactly, so the PLC’s scaling is right and the disagreement is upstream of it. The question is now between the transmitter and the gauge.

Work the other direction. If the true pressure were 150 psi, the transmitter should send:

4 + (150 ÷ 300) × 16 = 4 + 8.0 = 12.0 mA.

It is sending 11.6 mA, which is 0.4 mA low. As a fraction of span, 0.4 ÷ 16 = 0.025, so the transmitter reads 2.5% of span low, and 2.5% of 300 psi is 7.5 psi, exactly the disagreement observed.

Two candidates remain, and one test settles it. Apply a known pressure with a calibrator, or compare against a reference gauge you trust. A consistent 2.5% offset across the range is a zero or span calibration issue in the transmitter. If the offset changes with pressure, look for a plugged or partly filled impulse line.

Proving an instrument is telling the truth

The loop has three separable parts: the field device, the wiring, and the scaling in the controller. Separate them.

• Inject a known current with a loop calibrator in place of the transmitter. If the PLC then displays the value the arithmetic predicts, the wiring and the scaling are good and the transmitter is the suspect.
• Measure the actual loop current with the transmitter connected and compare it with what the process should be producing. This is the step above.
• Apply a known input to the transmitter (a pressure calibrator, an ice bath, a known level) and see what it sends.

Every one of those is a comparison against something you know. That is the whole method.

Where it bites

• A sensor that “works when you wiggle it” is a wiring fault, and it is almost always at a connector, a ferrule, or where a cable flexes.
• Instruments drift, and drift is invisible. A transmitter reading 2.5% low reports a plausible number all day. Only a comparison finds it, which is why calibration is on a schedule rather than on a complaint.
• Impulse lines fail more often than transmitters. A plugged tap, trapped gas in a liquid leg, or condensate in a gas leg all produce confident wrong readings from a healthy instrument.
• Never adjust an instrument’s calibration to make it agree with another instrument. Adjust it to agree with a traceable standard, and record what you changed.
• Check the sensor’s rated target before blaming its range. Sensing distances are stated for a specific target material and size, and a smaller or non-ferrous target legitimately reads shorter.
• A 4 mA reading is not proof of health. A transmitter can sit at 4 mA because the process is at the bottom of range, or because it has failed low. Compare against the process.

Exam relevance

Expect the 4 to 20 mA scaling arithmetic in both directions, the reason for the live zero, the difference between PNP and NPN outputs and why it must match the input module, and the three photoelectric arrangements in order of range and reliability. Expect a scenario asking you to separate a field device fault from a wiring fault from a scaling fault. The NCCER Industrial Maintenance program carries process instrumentation in its upper levels alongside PLCs, and the two subjects are examined together for good reason.