Temperature sensor and transmitter fault tracing

 Tracing Faults in a Thermocouple / Temperature Transmitter Loop — From Sensor to PLC Card

Anyone who has chased a "wrong temperature" alarm on a kiln inlet, preheater cyclone, or cooler bed knows the frustration: the field looks fine, the transmitter looks fine, and yet the DCS/PLC shows a value that doesn't make sense. The trick isn't guessing — it's working the loop in order, ruling out one section at a time, and knowing where each type of fault likes to hide.

This is a practical, field-proven sequence for tracing a temperature loop fault from the thermocouple (or RTD) all the way to the PLC analog input card.

1. Understand What You're Actually Chasing

Before touching a screwdriver, classify the symptom. It saves hours.

Reading pegged high or low (over-range/under-range) — almost always an open circuit or a burnt-out sensor.

Reading stuck at a fixed value, not moving — dead transmitter, dead card channel, or a logic/scaling fault.

Erratic, jumping readings — loose connection, ground loop, EMI/RFI pickup, or moisture ingress.

Reading offset by a constant amount — calibration drift, wrong cold-junction compensation, or wrong range configured on the transmitter/card.

Reading drifts slowly over time — thermocouple degradation, insulation breakdown, or junction box corrosion.

Write down what the trend looks like before the fault started. A sudden jump to full-scale points at a break; a slow creep points at sensor aging or a dirty connection.

2. Start at the Sensor — Not the PLC

It's tempting to open the PLC cabinet first because it's clean and comfortable. Resist it. Start where the temperature is actually measured.

For thermocouples (Type K, J, R/S in kiln/preheater service):

Check physical condition — is the thermowell intact, is the TC pulled back or dislodged, any visible damage to the sheath?

Disconnect at the head and measure resistance across the TC leads with a multimeter (mV/resistance mode). An open circuit reads infinite resistance — sensor is dead.

Check millivolt output against a reference table for the actual process temperature (or compare with an adjacent healthy TC on a similar point). A dead-flat 0 mV or wildly wrong mV confirms sensor failure.

Inspect the TC head for moisture, dust ingress, or corroded terminals — extremely common in cement plants with clinker dust and thermal cycling.

Check that polarity is correct at the terminal block; reversed leads give a nonsensical (often negative or halved) reading rather than a clean fault code.

For RTDs (Pt100, common on bearings, gearboxes, cooler air):

Measure resistance across the element. At 0°C a Pt100 reads 100 Ω; use the standard table to sanity-check against ambient/process temperature.

For 3-wire RTDs, check all three leads for continuity and equal lead resistance — mismatched lead resistance introduces measurement error even with a healthy element.

A shorted RTD reads near 0 Ω; an open one reads infinite.

3. Check the Wiring Run and Junction Boxes

This is where most "intermittent" faults actually live, especially in a cement plant with vibration, heat, and dust.

Trace the run from sensor head to junction box to marshalling cabinet. Look for chafed insulation, pinched cable near conveyor structures, or cable runs too close to VFD/motor power cables (induced noise).

Open junction boxes and check terminal tightness — vibration from crushers, fans, and kiln drives loosens screw terminals over months.

Check for moisture or clinker dust bridging terminals — a classic cause of erratic readings that come and go with weather or washdown.

For long thermocouple extension wire runs, confirm the extension wire is the correct compensating type (matching TC type) — using ordinary copper wire beyond the head introduces a false cold junction and a fixed offset error.

Verify cable shield is grounded at one end only. Grounding both ends is a common cause of ground-loop noise showing up as jumpy readings.

4. Test at the Transmitter

If the sensor itself checks out, move to the transmitter (head-mounted or DIN-rail, e.g. E+H, Siemens, Yokogawa types commonly used in cement).

Confirm the transmitter's configured input type and range matches the sensor actually installed (TC type, RTD type, and range in °C). A mismatch here gives you a clean, stable, but wrong reading — easy to miss because nothing looks "broken."

Check loop power at the transmitter terminals (typically 24 VDC) — low voltage from a long cable run or a weak power supply causes erratic behavior under load.

Measure the transmitter's output directly with a multimeter in mA mode, in series or using a HART communicator/handheld if available. A healthy 4–20 mA transmitter should track smoothly from 4 mA (low range) to 20 mA (high range).

If it's a HART-capable transmitter, connect a HART communicator and pull diagnostics directly — most units flag sensor break, over-range, or electronics faults internally, saving a lot of guesswork.

For head-mounted transmitters, check for heat damage — direct kiln/cooler radiant heat cooks these units over years, causing drift long before outright failure.

5. Verify the 4–20 mA Loop Itself

With the transmitter confirmed healthy, the loop wiring from transmitter to PLC cabinet is next.

Break the loop at a convenient terminal and insert a multimeter in series (mA mode) to confirm actual current flowing matches what the transmitter is producing.

Check loop resistance — total resistance of wiring plus any load must stay within the transmitter's rated compliance voltage, or the loop can't drive full range (usually shows up as reading pegged near the top of range under certain conditions).

Inspect isolators or barriers if the loop passes through intrinsically safe zone barriers or signal isolators — a failed isolator gives a dead or frozen signal that looks exactly like a card fault.

Check for damaged or reversed polarity at any intermediate terminal block between field and cabinet.

6. At the PLC — Terminal Block Before the Card

Before blaming the card, confirm what's actually arriving at the rack.

At the marshalling/terminal block just ahead of the card, measure the incoming signal directly — mA for a 4–20 mA input, mV for a direct TC input if the card is a native thermocouple input card.

Compare this reading against what you measured at the transmitter output. If it matches, the problem is downstream (card or logic). If it doesn't match, the fault is in the wiring between transmitter and cabinet — go back to Section 5.

Check that the card's channel is wired to the correct terminal — a surprising number of "faulty" points are simply mapped to the wrong channel after a rewire or rack change.

7. Diagnosing the PLC Card Itself

Now you're at the card — this is often where people start, but should really be last.

In Studio5000 (Allen Bradley) or the equivalent Siemens TIA Portal, check the module's channel status/fault tags. Analog input cards flag open-wire, over-range, under-range, and channel-fault conditions individually — read these before assuming a hardware failure.

If it's a dedicated thermocouple input card, confirm the correct TC type and cold-junction compensation (CJC) setting per channel in the module configuration — a wrong CJC setting gives a consistent offset that mimics a sensor calibration issue.

Swap the suspect channel's field wiring to a known-good spare channel on the same card (temporarily, for test purposes) — if the fault follows the wiring, the card channel is fine; if the fault stays on the same channel regardless of wiring, the channel itself is bad.

If possible, swap in a spare card of the same model and re-terminate — the fastest way to confirm or rule out a card-level fault, especially with I&C access to spares.

Check card power supply rail voltage and card diagnostics LEDs (fault, channel status indicators) — a sagging system power supply can cause multiple channels to misbehave together, which points away from a single-channel field fault and toward the rack power supply.

8. Confirm in Logic and HMI Last

Once the raw electrical signal is confirmed correct at the card, the final check is scaling and logic.

Verify the analog scaling (raw counts to engineering units) in the PLC program matches the transmitter's configured range exactly. A transmitter range change without a corresponding logic update is a very common source of "wrong but stable" readings.

Check for any filtering, clamping, or fault-substitution logic in the program that might be holding a stale or forced value — look for forces or overrides left active from a previous test.

Confirm the HMI tag is actually pointing at the correct PLC tag/address, especially after any program changes or tag database updates.

Quick Reference — Fault Location by Symptom

Symptom

Most likely location

Reading pegged full-scale high/low

Open sensor, open wiring, or blown fuse

Reading frozen, not moving

Dead transmitter, dead card channel, or forced tag

Erratic/jumpy reading

Loose terminal, moisture, ground loop, EMI from nearby power cables

Fixed offset error

Wrong CJC setting, wrong extension wire type, wrong range configured

Slow drift over weeks/months

Sensor aging, junction corrosion, transmitter heat damage

Multiple channels faulty together

Card power supply or rack power issue, not a field fault

A Field Habit Worth Building

Keep a simple log per critical loop (kiln inlet, cooler bed, preheater cyclones) noting the last known-good mA/mV reading, the transmitter range, and the CJC/TC type configured. When a fault hits at 2 AM, that reference cuts troubleshooting time dramatically — instead of re-deriving the expected values under pressure, you're comparing against a baseline.

The underlying discipline is always the same: work from the sensor outward, confirm each section with an actual measurement rather than an assumption, and let the signal — not the symptom on the HMI — tell you where the break really is.