RECD Alarm Returns After Reset: What Evidence Is Missing?

Every plant has a reset button, and every plant has a story about it. The RECD alarm appears, the operator presses reset, the light goes out and the set carries on. Three days later the alarm is back. The operator resets it again. After the third time, the maintenance lead asks the question this article is about: what do we know now that we did not know before the first reset?

Usually the honest answer is “nothing”, and that is the problem. A reset is a controller instruction that clears a latched signal. It does not repair anything and it does not explain anything. What it can do, if used carelessly, is remove the very information that would have explained the alarm: the values at the moment of the trigger, the order of events, the counters, the alarm state itself. The alarm returns because the cause was never touched. The next reset removes the evidence once more.

This guide treats the returning alarm as an evidence problem. First we look at what a reset actually removes. Then we build an evidence ledger, a one-page record that captures the facts a reset would erase. We set out a reset ladder that says when a reset is reasonable and when it is not, show how to read the interval between returns, and walk through a worked scenario. The guidance is general. Your controller’s manual and your supplier’s reset procedure always take priority.

What a reset really does

The word “reset” covers several different actions, and people use them interchangeably. Learn which one your controller offers.

Acknowledge (silence). Stops the horn or flashing lamp. The alarm condition remains and is still recorded as active.

Clear a latched alarm. Removes the latched flag if the condition has gone. If the condition is still present, most controllers will re-latch immediately.

Clear the fault history. Erases the stored list of past alarms, their counts and sometimes the data stored with them. This is the one that destroys evidence.

Reset counters or learned values. Some controllers keep learned values, such as a soot-load estimate or a baseline pressure. Resetting these returns them to defaults. Workshop guidance for diesel aftertreatment warns against forcing a reset of learned soot values unless the filter has actually been cleaned or replaced and the rest of the system is healthy, since the controller’s picture of the filter will then be wrong (Snap-on, via Vehicle Service Pros).

Power-cycle the controller. Restarts the device. Depending on design, this may clear temporary values, volatile logs or timers.

Reboot after a software or settings change. Applies new settings, which can change what the alarm means.

Several of these look the same from the front panel. Before the next press of the button, check in the manual (or ask the supplier) exactly which of the above it does, and what is stored where.

Three things a careless reset erases

  1. The alarm’s context. Many controllers store the values at the time of a trigger (pressure, temperature, load, heater state). That snapshot is the most useful single record you will ever get, and it can disappear when the history is cleared.
  2. The count and the interval. “Third time in a week” is a very different message from “once a year”. If the counter is cleared, the story is lost.
  3. The alarm’s state. Whether it was still active, had cleared on its own or had been acknowledged tells you a great deal. A reset flattens all three states into “gone”.

Vehicle diagnostics offers a useful parallel. A training text on scan-tool use says to capture trouble codes, their status (pending, confirmed, history), and any stored snapshot before clearing anything, because clearing is a step in a procedure and not proof of a repair (ASE-aligned study guide). The principle carries over to stationary equipment.

Why the alarm returns

An alarm returns for one of four reasons. Which one applies decides everything you do next.

The cause is still there. The most common reason. The condition that set the alarm was never removed, so the controller sets it again as soon as it checks. This is typical of a genuine rise in differential pressure, a failing heater or a leak.

The cause is cyclical. The condition comes and goes with load, temperature, time of day or weather, and the alarm returns whenever the cycle comes round. See our article on RECD faults that appear only in wet weather for the weather case.

The detection is wrong. A sensor, a line, a scaling or a setting is producing a false trigger. The exhaust is fine and the alarm is not. Our guide on separating a sensor fault from an RECD performance problem shows how to test this.

The reset itself changes the state. A reset of learned values can make the controller start from defaults, and some controllers then raise a different alarm or re-learn badly. A fleet-maintenance article on a heavy truck gives a related example: a derate fault kept returning after the code was cleared and a forced regeneration was run, until live data showed that the dosing system was acting when it should not be, which traced to a wiring fault (Fleet Maintenance). The visible fault was the symptom, and the cause lay in a different system.

Each of these four leaves a different trace in the data. That is what the ledger is for.

The evidence ledger

The ledger is a one-page sheet (or a spreadsheet tab) you fill in before any reset beyond a simple acknowledgement. It captures five kinds of fact. Do not worry about being exact. A rough, honest record is far better than none.

SectionWhat to captureWhy it matters
1. The alarm itselfExact wording and code, time first seen, active or inactive at the time you looked, how many times it has occurred, which controller raised itIdentifies the alarm type and its history
2. The state of the setLoad (kW or %), run hours since start, time since last stop, engine speed, ambient conditions, any transfer or load step in the last ten minutesShows whether the condition is tied to duty
3. The numbers at the triggerDifferential pressure, back-pressure if available, inlet and outlet temperatures, heater current, controller supply voltage, any other value the screen showsThe values the alarm was based on
4. The lead-up and the aftermathWhat the same values were an hour before, and an hour afterSeparates a spike from a trend
5. What changed recentlyService visits, cleaning, fuel delivery, engine work, relocation, firmware or setting changes, weatherPoints to a cause or at least to a suspect

Add two more lines at the bottom: who pressed which button, and when, and what the display showed immediately after. Many sites find that the display after the reset is the most informative piece of all. An alarm that returns within a minute is a different problem from one that returns after six days.

Photograph, then record

Photograph the controller screen before and after. A phone photo with a timestamp costs nothing and cannot be edited by a later memory. Also photograph the sensing hardware, the canister and the area around the exhaust for any visible change. Our guide to what to record in an RECD event log suggests the format of a standing log, and the RECD shift handover log shows how to pass the ledger from one shift to the next so that evidence does not depend on one person’s memory.

The reset ladder

Not every reset is wrong. A reset is a tool, and like any tool it has a place. The ladder below goes from the safest step to the riskiest. Climb only as far as the evidence supports, and write down each rung you use.

Rung 0: Observe and record. Do nothing to the controller. Fill in the ledger. Check whether the alarm is still active. If it is active and the set is safe to run, you can often finish the whole ledger without pressing anything.

Rung 1: Acknowledge. Silence the horn. The record remains. This is almost always safe and should be the default.

Rung 2: Read, do not clear. Page through the alarm history, snapshot data and counters. Photograph them. Download them if the controller allows a USB or network export. See what a regeneration history report should contain for an example of the data worth keeping.

Rung 3: Clear the latched alarm only. If the supplier’s manual says this does not erase history, and the condition is verified to have cleared, a single clear can be reasonable. The key word is “verified”. You clear because you know the cause has gone, not in the hope that it has.

Rung 4: Clear the history. Only after the history has been exported or photographed in full, and only if the supplier’s procedure calls for it, for example after a repair.

Rung 5: Reset learned values or counters. Only after the filter has been cleaned or replaced or a sensor has been changed, and only with the supplier’s procedure. This step changes the controller’s view of the system.

Rung 6: Power-cycle or reboot. A last resort for a frozen controller, taken after the ledger is complete. Check beforehand what is stored in volatile memory.

A simple rule of thumb follows from the ladder: the further up the ladder you go, the more evidence you must already have in hand.

When not to reset at all

Some situations call for leaving the alarm standing until a qualified person has seen it.

  • The alarm relates to temperature during a regeneration, or to a suspected thermal event.
  • The alarm comes with smoke, smell, noise or a pressure that has fallen suddenly. See our guides on rattling inside an RECD and black smoke after RECD installation.
  • The alarm has returned three or more times without a diagnosed cause.
  • A protective action (derate, trip, bypass) has operated.
  • The alarm appeared right after maintenance work or a fuel delivery.

An aircraft maker’s maintenance article states the stance well in its own field: resets can be an authorised first step, but troubleshooting should not end with them, and after three occurrences of a failure, even an unconfirmed one, the manual’s fault-isolation steps should be followed through to a resolution (Airbus). A “three strikes” rule is simple to adopt on any site, and it changes behaviour.

Reading the interval between returns

Once you have a few returns, the time between them becomes data. The table below describes common patterns. Treat them as clues, not verdicts.

Pattern of returnWhat it suggestsWhere to look
Within seconds or minutes of the resetThe condition is still present, or the reset left a bad stateLive values; the ledger’s section 3; a sensor or line fault
At the same point in every run (for example 40 minutes after start)A condition tied to warm-up, temperature or a timed eventTemperature sensors, heater, regeneration timing
At every start or first load transferA start-up or step effectSee RECD pressure spikes at startup and alarms during load steps
After a set number of run hours, shrinking each timeA cumulative cause such as loadingSee why an RECD keeps clogging
On particular days (wet, cold, after outages)An environmental or duty triggerWeather and run-hour ledger
At random, with no pattern at allPossibly a loose connection, supply dip or sensor faultWiring, supply voltage logging
Only after someone pressed resetA reset side effectWhich reset, and what changed in the controller afterwards

Plot the return times on a simple line, one dot per return, with run hours on the horizontal axis. A shrinking gap is a worrying sign. A steady gap suggests a fixed trigger. Scatter suggests an unstable input. Even a sketch on paper tells you more than a list of dates.

What is probably missing from your evidence right now

If your RECD alarm has returned after a reset, the following are the evidence gaps we see most often. Check your own situation against the list.

  1. No record of the first occurrence. The alarm was reset before anyone noted its wording, time or values.
  2. No snapshot. The controller stored one, but the history was cleared, or nobody knows where to look.
  3. No independent reading. Only the controller’s own sensors report the condition, so nobody can tell whether the exhaust or the sensor is at fault. See pressure sensors versus pressure switches.
  4. No load record. Nobody knows what the set was doing when the alarm came.
  5. No change log. Recent maintenance, fuel deliveries or setting changes are not written down anywhere.
  6. No synchronised clocks. The RECD controller, the DG controller and the building system disagree about the time, so events cannot be lined up. See aligning RECD sensor timestamps with generator runtime logs.
  7. No record of who did what. Resets, setting changes and visits are not tied to names and times. Our article on RECD controller settings and access control explains why this matters.
  8. No baseline. There is nothing to compare today’s values against.

Each gap has a cheap fix: a notebook, a phone, a USB export, a clock check, a one-page change log. None of them needs new equipment.

A worked scenario: the third reset

The following is an illustrative scenario, not a real customer. A packaging plant has a 500 kVA DG set with an RECD. Over three weeks, a high differential pressure alarm appears on the controller. Each time, the night operator presses reset and the alarm clears. On the third occurrence, the maintenance lead decides to stop and build a ledger first.

Section 1. The alarm wording and code are written down. It is the third occurrence in nineteen days. The history shows two earlier entries, and the controller stored a snapshot for each.

Section 2. The set was at about 70% load on the third occurrence, 52 minutes after start-up, about five minutes after a load step when a compressor started.

Section 3. The controller reports ΔP at 9% above its limit at the moment of the trigger. A portable gauge connected in parallel reads 6% below the limit. The two disagree.

Section 4. An hour before, both were within 2% of each other. An hour after, the gap returns to 2%.

Section 5. The change log shows that the sensing tubes were disturbed during a service visit five weeks earlier, and that the compressor was added to the plant two months ago.

Reading the ledger. The independent gauge says the exhaust did not cross the limit. The disagreement appears only around the load step. Two causes are plausible: a transient that the controller’s fast sampling catches and the gauge’s damping misses, or a sensing-line issue created at the last service. The returns also line up with compressor starts. The team compares snapshots from the first two occurrences and finds a similar step in load each time.

The action. The supplier reviews the sensing-tube routing and finds a sag where condensate collects. The tube is re-routed. Because the load-step transient is real but brief, the team also asks the supplier to review the alarm’s persistence settings as described in our article on alarms during generator load steps, and keeps the hard limit unchanged. The alarm has not returned in the following two months. Had the operator reset the history on the first night, the snapshots that pointed at the load step would have been gone.

A short routine for operators

Print this and keep it at the controller.

  1. Read the alarm. Do not press anything yet.
  2. Photograph the screen.
  3. Acknowledge the horn only.
  4. Fill in the ledger (sections 1 to 5).
  5. Check whether the alarm is still active. If yes, and it is safe, leave it and call the maintenance lead.
  6. If it has cleared on its own, note the time.
  7. If a reset is needed, use the lowest rung that works and write down which one.
  8. Note what the display showed afterwards.
  9. If this is the third occurrence, stop and escalate.

Our RECD alarm response matrix for operators explains what to do for each alarm type, and the article on an RECD fault business continuity plan covers what to do if the alarm means the set cannot run compliantly.

Questions people ask

Is it wrong to reset an alarm?

No. Resetting is a normal part of operating equipment. The error is to reset without capturing evidence, or to treat a reset as a repair.

The alarm cleared by itself. Does that mean it was a glitch?

Not necessarily. A self-clearing alarm usually means the condition came and went. It could be a transient, a weather effect or a loose connection. Record it. If it recurs, it joins the pattern.

Can I export the alarm history?

That depends on the controller. Many offer USB, serial or network export, and a photograph is a fine fallback. Ask the supplier how to export before the next visit, and see preserving RECD data during a controller replacement if a controller change is planned.

Who should be allowed to clear history?

Ideally a small, named group, with every clear logged. Operators can acknowledge and record, and a supervisor or the supplier can clear history after the ledger is complete.

Should the alarm limit be raised if it keeps coming back?

Only after the cause is understood, and only with the supplier’s agreement. Raising a limit to stop an alarm without knowing why it fires removes protection when you most need it. See reducing nuisance RECD alerts without hiding faults.

Where to go from here

A returning alarm is a message that the cause is still out there. The reset did not answer it, and a better answer comes from the record. Know what each reset does, fill in the evidence ledger before you press anything beyond an acknowledgement, climb the reset ladder only as far as the evidence supports, treat the third return as a stop sign and read the intervals as data. If you would like help setting up a ledger for your controller or reviewing a snapshot, you can explore our retrofit emission control devices, see the FAQ or talk to the Aceget team.

Related Reading

Sources



Leave a Reply