How to Separate a Sensor Fault from an RECD Performance Problem

An alarm appears: “RECD differential pressure LOW.” It could mean the filter has cracked and exhaust is bypassing it, which is a real performance problem with compliance implications. Or it could mean the impulse line to the pressure transmitter is blocked, the cable is damaged or the sensor has drifted. That’s a sensor problem, and the Retrofit Emission Control Device itself may be fine.

The response to each is completely different. Treat a sensor fault as a device failure, and you may take a generator out of service, call in the supplier and order parts for nothing. Treat a device failure as a sensor fault, and you may run for weeks with a damaged filter, unaware that particulate is passing straight through.

This guide gives a practical method for telling the two apart: the questions to ask, the signatures that distinguish sensor faults from real problems, simple field tests, and worked examples.

The Core Idea: Real Problems Are Physically Consistent

The most useful principle is simple:

A real RECD problem changes several related measurements in ways that make physical sense together. A sensor fault usually affects one measurement, often in a way that doesn’t fit with anything else.

For example, a genuinely blocked filter raises ΔP across the RECD and raises engine back-pressure and often comes with failed regenerations. A faulty ΔP transmitter shows high ΔP while back-pressure stays normal and regeneration history looks ordinary.

Most diagnosis comes down to checking whether the readings tell a consistent story.

Four Questions to Ask First

Before touching any equipment, review the data and ask:

1. Does the reading make physical sense?

• Is the value within the sensor’s range, and within what is physically possible?
• Does it make sense for the DG’s state? For example, ΔP should be near zero when the engine is stopped.

2. Does it agree with related measurements?

• Does back-pressure move in the same direction as ΔP?
• Do inlet and outlet temperatures agree when no regeneration is active?
• Does the engine’s own exhaust temperature fit?

3. How did it change?

• Gradually, over days or weeks, which is typical of soot or ash build-up or slow sensor drift
• Suddenly, in a single step, which is typical of a crack, a seal failure, a wiring fault or a blocked impulse line
• Erratically, jumping around, which is typical of loose connections, electrical noise or a failing sensor

4. Does it respond to changes in operating conditions?

• When load changes, does the reading change as expected?
• During and after a regeneration, does it behave as before?
• A reading that doesn’t respond at all to load changes is very likely a sensor or impulse-line problem.

Signatures: Sensor Fault vs Performance Problem

Signature	Points to a sensor fault	Points to an RECD performance problem
Value frozen exactly, even as load changes	✔ (stuck sensor, blocked impulse line, comms issue)
Value at range limit (e.g. 0 or full scale) or error value	✔ (broken wire, open circuit, failed sensor)
Non-zero ΔP with DG stopped and cold	✔ (zero offset, condensate in impulse line)
Erratic spikes and drops with no operating cause	✔ (loose connection, noise, failing sensor)
Only one measurement abnormal; related ones normal	✔
ΔP and back-pressure both rise steadily over weeks		✔ (soot/ash build-up)
ΔP drops suddenly; back-pressure also drops; downstream opacity rises		✔ (crack, seal failure, bypass)
Repeated regeneration failures alongside rising ΔP		✔ (regeneration problem; possibly heater or duty cycle)
High outlet temperature with high ΔT, while controller shows regeneration or heater active		✔ (real thermal event)
Abnormal readings appear across several sensors after a known event (e.g. engine work)		✔ (changed exhaust or installation)
Problem appears right after a sensor, wiring or software change	✔ (installation, scaling or configuration error)

Use the table as a guide, not a verdict. Field checks confirm the diagnosis.

Field Tests That Settle the Question

Test 1: The zero check (pressure sensors)

With the DG stopped and cooled, differential pressure and back-pressure sensors should read zero, within tolerance.

• Reads zero: zero is fine; any problem is elsewhere
• Reads a steady offset: zero drift or condensate in the impulse line
• Reads full scale or error: wiring or sensor failure

This takes seconds and should be the first step for any pressure alarm.

Test 2: Compare with a portable reference

With the DG running steadily, connect a calibrated handheld manometer to the same tapping points, using test ports or tees, or a reference thermometer probe through a spare port.

• Installed and reference agree: the sensor is fine, so the reading is real
• They disagree: sensor, impulse line or scaling problem

See installed RECD sensor calibration: what should facility teams document.

Test 3: Inspect and clear impulse lines

Soot-blocked or condensate-filled impulse lines are among the most common causes of false pressure readings. Following the supplier’s procedure:

• Check the lines for kinks, damage or low points where condensate can collect
• Disconnect and check for blockage
• Clear them, reconnect, and recheck the reading

If the reading changes noticeably after clearing, the impulse line was the problem.

Test 4: Check the signal path

Compare the value at each stage:

• At the sensor, by measuring the 4–20 mA signal with a loop meter, or reading the sensor’s own display
• At the RECD controller display
• At the BMS or remote platform

If the sensor output is right but the BMS is wrong, the problem is scaling, mapping or communication. See RECD Modbus data: units, scaling and register mapping explained.

Test 5: Cross-check related measurements

• Compare ΔP with engine back-pressure: do they move together?
• Compare inlet and outlet temperatures at steady state with no regeneration: are they close?
• Compare RECD inlet temperature with the DG controller’s exhaust temperature: is the relationship normal?
• Check downstream opacity with a calibrated meter: has it changed? See smoke opacity vs particulate mass

Test 6: The load step

Change the load (for example with a load bank) and watch the response:

• Real ΔP and back-pressure rise and fall promptly with load
• A blocked impulse line or stuck sensor responds slowly or not at all

Test 7: Swap or substitute (if practical)

If two identical sensors are fitted, for example on two DG sets, swapping them, or temporarily fitting a known-good spare, shows whether the fault follows the sensor. Do this only with the supplier’s guidance and with the change recorded.

Worked Case 1: Low ΔP Alarm, Real Problem

The scenario is illustrative.

Symptoms: “RECD ΔP LOW” alarm on DG1 during a routine run at 60% load. ΔP 21 mbar, against a baseline of 39 mbar at that load (ΔP index 0.54). See how to compare RECD pressure readings at different generator loads.

Checks:

1. Zero check (DG stopped): reads 0.2 mbar, fine
2. Reference manometer at the same tappings, DG running: 20.5 mbar, which agrees with the transmitter
3. Engine back-pressure had also dropped compared with the usual value at this load
4. Downstream opacity had risen from a typical 0.7% to 3.8%
5. Event log: a high outlet temperature event two weeks earlier during a long light-load run, followed by a regeneration

Conclusion: a real performance problem. The pattern (low ΔP, lower back-pressure, higher opacity, after a thermal event) points to a cracked substrate letting exhaust bypass the filter. The supplier inspected the device and confirmed a crack. The substrate was replaced, and a retest was arranged. See when should an RECD be retested after changes to the DG set.

Worked Case 2: High ΔP Alarm, Sensor Problem

Symptoms: “RECD ΔP HIGH – critical” on DG2. ΔP shows 96 mbar at 50% load, against a baseline of 32 mbar.

Checks:

1. Engine back-pressure: normal for this load
2. Regeneration history: normal, with recent successful regenerations
3. Trend: ΔP jumped from 34 to 96 mbar in a single step two days ago, and hasn’t changed since, even across load changes from 30% to 70%
4. Zero check (DG stopped): reads 91 mbar
5. Impulse line inspection: the low-pressure (outlet-side) impulse line was blocked with soot and condensate

Conclusion: a sensor problem. A blocked impulse line on the outlet side traps a pressure, so the transmitter sees a large, unchanging “differential.” The frozen value, the non-zero reading at rest and the normal back-pressure all point the same way. After clearing and re-routing the impulse line with a continuous fall, ΔP read 33 mbar at 50% load. The installation was improved to prevent a repeat. See pressure sensors vs pressure switches in RECD monitoring.

Worked Case 3: High Outlet Temperature, Mixed Picture

Symptoms: “RECD OUTLET TEMP HIGH” on DG3. Outlet reads 640 °C, inlet reads 410 °C, at 70% load. The controller shows no regeneration active and the heater off.

Checks:

1. Outlet temperature trend: jumped from 405 to 640 °C instantly, then stayed there
2. Back-pressure and ΔP: normal for this load
3. Visual check from a safe distance: no visible signs of overheating, no unusual smoke
4. Signal path: the controller showed 640 °C, but the BMS showed the same value only from that moment, and a loose terminal was found at the thermocouple extension cable junction
5. After tightening: outlet read 402 °C, consistent with inlet

Conclusion: a sensor fault (loose connection). A real uncontrolled burn would raise temperature progressively, usually with changes in ΔP and back-pressure, and the jump here was instantaneous. Even so, the alarm was treated as real until checks showed otherwise, which was the right call. For high-temperature alarms, safety comes first: follow the alarm matrix, then investigate. See how to design an RECD alarm response matrix for operators.

Worked Case 4: Rising Back-Pressure, Problem Outside the RECD

Symptoms: a “high back-pressure – warning” alarm on DG4 at 75% load. Back-pressure is 52 mbar, against a typical 41 mbar at this load.

Checks:

1. RECD ΔP index: 1.04, essentially normal
2. Back-pressure index: 1.27, clearly elevated
3. Back-pressure transmitter: zero check fine, and agrees with a reference manometer
4. Exhaust walk-down: the rain cap on the stack had been damaged in a storm and was partly closed, restricting the outlet

Conclusion: neither a sensor fault nor an RECD problem, but a restriction elsewhere in the exhaust. The steady RECD ΔP alongside rising back-pressure pointed downstream of the device. After the rain cap was repaired, back-pressure returned to normal. See RECD protection from rainwater and condensation.

This case is a reminder that the exhaust system has more than two possible culprits. Comparing ΔP and back-pressure indices helps locate the restriction.

Safety First, Diagnosis Second

Some alarms must be treated as real until proven otherwise:

• High outlet temperature
• High engine back-pressure

For these, follow the response in your alarm matrix (reduce load, keep clear of the device, stop if required) before diagnosing the sensor. Diagnosis can wait a few minutes. An uncontrolled burn or an engine overheating from back-pressure can’t.

For other alarms, such as low ΔP, regeneration overdue or sensor fault, there’s usually time to investigate methodically.

A Simple Diagnostic Flow

1. Is it a safety-critical alarm? If yes, follow the alarm matrix first.
2. Check the trend: gradual, sudden or erratic? Frozen?
3. Check related readings: do they tell a consistent story?
4. Zero check (pressure) or cold check (temperature) when the DG is stopped.
5. Compare with a portable reference while running.
6. Inspect impulse lines and wiring.
7. Check the signal path: sensor → controller → BMS.
8. Decide: sensor fault (repair, recalibrate, record) or performance problem (inspect device, involve supplier, consider retest).
9. Record the diagnosis, tests and outcome in the event log.

Building Automatic Diagnostics Into Monitoring

Many sensor faults can be flagged automatically by the RECD controller, BMS or monitoring platform, before anyone has to investigate by hand. Useful automatic checks:

Check	Logic	Flags
Flatline	Value unchanged for more than X minutes while DG runs and load changes	Stuck sensor, blocked impulse line, comms freeze
Range	Value outside physical or sensor limits	Broken wire, open circuit, failed sensor
Zero at rest	Pressure reading above tolerance when DG has been stopped for a set time	Zero drift, condensate
Pressure consistency	ΔP index and back-pressure index diverge sharply	Possible sensor or impulse line problem
Temperature consistency	Outlet − inlet temperature outside expected band with no regeneration active	Sensor or installation problem
Rate of change	Value jumps faster than physically possible	Loose connection, electrical noise
Response to load	Value doesn’t change after a significant load step	Stuck sensor or blocked impulse line

These checks generate sensor health alarms, separate from process alarms. That separation helps operators: a “sensor suspect” message points to instrumentation, while a “high ΔP” message points to the device. See detecting missing data in RECD monitoring records and how to compare RECD pressure readings at different generator loads.

When to Call the Supplier

Involve the RECD supplier when:

• Checks point to a real performance problem, such as a crack, blockage, regeneration failure or overheating
• A sensor fault keeps recurring after repair
• The diagnosis is unclear after the basic checks
• The device needs to be opened, cleaned or have components replaced
• The problem may affect warranty or compliance

Share the trend data, the checks already done and their results. That shortens the supplier’s diagnosis and often avoids an unnecessary visit. See RECD service response commitments.

Common Sensor Problems on RECD Installations

Problem	Typical symptom	Prevention
Soot-blocked impulse line	Frozen or very slow pressure response	Correct routing, periodic blow-through
Condensate in impulse line	Zero offset; readings vary with temperature	Continuous fall back to exhaust; avoid low points
Leaking fitting	Reading too low	Leak checks during maintenance
Thermocouple junction failure	Open circuit; reading jumps to error value	Quality sensors; vibration protection
Wrong or damaged extension cable	Steady temperature error	Correct cable type; protect from heat
Loose terminals	Erratic or stepped readings	Torque checks; vibration-resistant terminals
Transmitter overheated	Drift or failure	Mount away from radiant heat
Scaling or mapping error	Wrong values in BMS only	Loop checks after any software or hardware change
Moisture in connection heads	Intermittent faults, especially in monsoon	Sealed heads; drip loops

Documenting the Diagnosis

Whatever the answer, record:

• The alarm or symptom, with time and values
• The checks done and their results
• The conclusion: sensor fault or performance problem, and why
• The action taken (repair, recalibration, inspection, replacement)
• Any follow-up (retest, interval change, installation improvement)

Over time, these records show patterns. A sensor that faults every monsoon, or impulse lines that block every few months, both point to an installation fix. See what should an RECD event log record.

Related Reading

• Installed RECD sensor calibration: what should facility teams document?
• Pressure sensors vs pressure switches in RECD monitoring
• How to compare RECD pressure readings at different generator loads
• RECD temperature sensor placement: what should be measured?
• RECD maintenance guide

Frequently Asked Questions

What’s the quickest check when an RECD pressure alarm appears?

Look at related readings and the trend first. If ΔP has jumped in one step and is frozen, while back-pressure is normal, suspect the sensor or impulse line. When the DG stops, do a zero check.

Can a sensor fault hide a real RECD problem?

Yes. A blocked impulse line can freeze ΔP at a normal-looking value while the filter loads up or cracks. That’s why flatline detection, regular zero checks and cross-checks with back-pressure and opacity matter. See detecting missing data in RECD monitoring records.

Should operators diagnose sensor faults themselves?

Operators can do the safe, simple checks: reviewing trends, comparing related readings and noting conditions. Impulse line work, wiring checks and calibration should be done by trained technicians, following the supplier’s procedures.

How do I know if low ΔP means a cracked filter?

Look for supporting evidence: lower back-pressure at the same load, rising downstream opacity, and a recent thermal event or regeneration problem. If the installed sensor agrees with a reference manometer and these signs are present, a crack or bypass is likely. Arrange an inspection.

Is it worth fitting redundant sensors?

For critical sites, a second, independent measurement of the most important values, such as an independent back-pressure switch or a second temperature sensor at the outlet, makes diagnosis much faster and provides a backup alarm. For smaller sites, a portable reference instrument used during checks is usually enough.

Can a sensor fault and a real problem happen at the same time?

Yes, and it’s worth keeping in mind. A thermal event can crack a substrate and damage a nearby sensor or cable at the same time. If a sensor fault appears right after an abnormal event, repair the sensor, then check the device as well.

Do sensor faults affect compliance?

A sensor fault doesn’t change the RECD’s actual emissions, but long periods of unreliable monitoring weaken your evidence that the device was working. Fix faults promptly and record them.

Diagnose Before You Dismantle

Most RECD alarms can be sorted into “sensor” or “device” within an hour, using trends, cross-checks and a few simple tests. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA, and our service team can support diagnosis on our installations. See our certifications, or contact our team.