An off-normal alarm is a discrete alarm that annunciates whenever a device is in a state other than the one it is supposed to be in. Unlike an analog limit alarm, which watches a measured number against a threshold, an off-normal alarm watches a boolean status, running or stopped, open or closed, energized or de-energized, and it alarms when that status is not what normal operation calls for. A pump that should be running but has stopped, a breaker that should be closed but has tripped open, or a door that should be shut but is standing open are all off-normal conditions. This page defines the off-normal alarm class, explains how the normal state is declared for each point, and distinguishes it from a plain change-of-state event that merely records that something switched.
Off-normal alarm in one line: An off-normal alarm is a discrete, boolean alarm that activates whenever a device leaves its defined normal state, such as a pump that should be running having stopped or a breaker that should be closed having opened. Rather than comparing a measured value to a limit, it compares a device's status to the state that is considered normal for that point and annunciates any deviation. It differs from a change-of-state event because it judges the state against an expected normal, not merely reports that a transition occurred.
The clearest way to understand an off-normal alarm is to contrast it with the analog limit alarm most people picture when they hear the word alarm. A limit alarm watches a continuous measured value, a pressure, a temperature, a level, and compares it to a numeric threshold, annunciating when the number crosses the line. An off-normal alarm watches a discrete status that has only a small set of possible states, typically two, and it has no threshold to cross because there is no continuous scale. Its input is a boolean or a status: a motor run feedback, a valve limit switch, a breaker auxiliary contact, a door switch. The alarm's job is not to compare a number to a limit but to compare a state to the state that should be true.
This makes the off-normal alarm the natural way to annunciate the many field conditions that are inherently on-or-off rather than more-or-less. Equipment running status is one big category: a pump, fan, compressor, or agitator that is commanded to run should be running, and a loss of its run feedback is an off-normal condition worth an operator's attention. Protection and isolation devices are another: a breaker, a contactor, or a safety valve that should be in one position and is found in the other. And environmental or security conditions round it out: a door or hatch that should be closed, a sump that should be dry, an enclosure that should be sealed. In every case the underlying signal is discrete, and the abnormality is a wrong state rather than an out-of-range number.
Because the signal is discrete, the off-normal alarm avoids some of the complications of analog alarming, such as deadbands sized against noise and hysteresis bands, but it introduces its own considerations. A discrete status can bounce during transitions, a contact can be noisy, and a device that is legitimately in transit, a valve mid-stroke, a motor spinning down, may momentarily read a state that would look off-normal if judged instantly. So while there is no amplitude threshold to tune, an off-normal alarm often still needs a short time filter to ignore the brief, expected excursions that occur while a device is legitimately changing state, so that only a genuine, settled off-normal condition annunciates.
The heart of an off-normal alarm is the declaration of what normal means for that specific point, because the same physical state can be normal for one device and abnormal for another. A valve that is normally open and a valve that is normally closed produce off-normal alarms on opposite states, even though the raw signal, open or closed, is identical in kind. So configuring an off-normal alarm always includes specifying the expected normal state for the point: this pump is normally running, so stopped is off-normal; this isolation valve is normally closed, so open is off-normal; this door is normally shut, so open is off-normal. Without that per-point declaration, the system cannot tell which state should raise the alarm.
In some cases the normal state is fixed and static, set once at configuration and never changed, which suits devices with a permanent expected condition such as an emergency isolation valve that is always meant to be open in normal operation. In other cases the normal state is conditional, because what counts as normal depends on the plant's mode or on a command. A pump's run feedback is only off-normal when the pump is commanded to run; if the operator has deliberately stopped it, stopped is the correct state and no alarm should fire. So a well-built off-normal alarm often compares the actual status not to a fixed constant but to the commanded or expected status, alarming only when the device disagrees with what it was told to do or with the state the current operating mode requires.
Getting this expected-state logic right is what separates a useful off-normal alarm from a noisy one. If the normal state is defined too rigidly, the alarm nuisances every time the process legitimately changes the device's state, training the operator to ignore it. If it is defined too loosely, or if the comparison is against the command rather than a true expectation when it should be the reverse, a genuine failure can slip through because the alarm thinks the current state is acceptable. The design question for each point is therefore not just what state is normal, but under what conditions, and the best off-normal alarms encode that context so they fire on real disagreements between what the device should be doing and what it is actually doing.
An off-normal alarm is easy to confuse with a simple change-of-state event, but they are not the same thing, and the difference matters for how the operator experiences them. A change-of-state event merely records that a discrete signal transitioned, from off to on or on to off, without any judgment about whether the new state is good or bad. It is a factual log entry: this contact changed at this time. An off-normal alarm, by contrast, embeds a judgment: it only annunciates when the resulting state is the wrong one, the one that departs from normal. A pump stopping because the operator commanded it is a change of state but not an off-normal alarm; a pump stopping while it was commanded to run is both a change of state and an off-normal alarm.
This distinction keeps the alarm list meaningful. If every discrete transition raised an alarm, the operator would be buried in annunciations for perfectly normal operations, starting and stopping equipment, opening and closing valves, that are not problems at all. The off-normal alarm filters that flood down to the transitions that actually indicate something is wrong, by measuring each resulting state against the expected normal and staying silent for the changes that land in a normal state. Change-of-state records still have their place, in the event history and sequence-of-events logs used for after-the-fact analysis, but they belong in the historical record rather than in the operator's live alarm list where only genuine abnormalities should demand attention.
In a SCADA and remote-monitoring context, off-normal alarms carry an outsized share of the useful information, because so many remote-site problems are discrete rather than analog. At an unmanned station the questions that matter most are often binary: is the pump that should be pumping actually running, is the gate that should be closed still closed, has the enclosure door been opened, has a breaker tripped. A cloud SCADA platform such as Merobix that ingests these discrete statuses and evaluates each against its declared normal state can push a single clear notification, the pump has stopped unexpectedly, rather than a stream of raw state changes. Because the platform knows the expected state, it can also suppress the transitions that come from deliberate remote commands, so the operator is alerted only when a device has drifted off-normal on its own, which is exactly the condition that warrants a look at a site nobody is standing next to.
An analog limit alarm watches a continuous measured value against a numeric threshold and fires when the number crosses it. An off-normal alarm watches a discrete status with only a few possible states and fires when the device is in a state other than its declared normal one. There is no threshold to tune on an off-normal alarm because the input is boolean, so the concern is a wrong state rather than an out-of-range value, though a short time filter is often still used to ignore transitions.
The normal state is declared per point during configuration, because the same physical state can be normal for one device and abnormal for another, such as a normally open valve versus a normally closed valve. In simple cases the normal state is a fixed constant, but in many cases it is conditional on a command or operating mode, so the alarm compares the actual status to the commanded or expected status and only fires when the device disagrees with what it should be doing.
No. A change-of-state event simply records that a discrete signal transitioned, with no judgment about whether the new state is good or bad. An off-normal alarm adds a judgment and only annunciates when the resulting state departs from the expected normal. A pump stopping on operator command is a change of state but not an off-normal alarm, while a pump stopping while commanded to run is both, which is why off-normal alarms keep the live list meaningful instead of flooding it with routine transitions.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.