NAMUR NE43 is a recommendation that turns a plain 4-20 mA analog signal into something that can also announce its own failure. It carves the current range into a valid measurement band and two fault bands, so a transmitter that has lost its sensor or its own health can drive the loop below or above the normal range instead of parking at a value the operator would trust. The point is simple but important: a genuine 0 percent reading and a dead measurement should not look identical to the control system. NE43 makes them look different, and that difference is what lets a SCADA system alarm on a broken loop rather than acting on a lie.
NAMUR NE43 in one line: NAMUR NE43 is a NAMUR recommendation that reserves 3.8 to 20.5 mA for the valid measured range of a 4-20 mA transmitter and pushes the output below 3.6 mA (downscale) or above 21 mA (upscale) to signal a sensor or transmitter fault. It gives an analog loop a way to distinguish a real low reading from a failed measurement.
A raw 4-20 mA loop uses only the span from 4 mA at 0 percent to 20 mA at 100 percent. NE43 keeps that span for the process value but adds structure above and below it. The measurement information range is defined as 3.8 to 20.5 mA, giving the signal a little headroom at each end so a slightly under-range or over-range process does not immediately look like a fault. Below 3.6 mA and above 21 mA, the current is reserved to mean failure, and the transmitter is expected to drive its output firmly into one of those bands when it detects a problem it cannot measure through.
The choice of which fault band to use is deliberate. A downscale fault drives the loop below 3.6 mA, which reads as a hard low value the control system can recognize as impossible for a healthy signal. An upscale fault drives the loop above 21 mA, which reads as a hard high value. Whether a device goes upscale or downscale on failure is a configurable, safety-relevant decision, because it determines which way a controlled process moves when the measurement dies. On a heater interlock, for example, an engineer usually wants the fail direction to trip toward the safe state rather than toward more energy.
What NE43 does not do is diagnose the cause. It only guarantees that a fault produces a current outside the plausible process band, so downstream equipment can catch it. The transmitter still has to detect its own trouble - a broken sensor lead, an out-of-range input, an internal self-test failure - and then respond by slamming the loop into the reserved band. NE43 standardizes the language of that response so that any compliant input card interprets it the same way.
The weakness NE43 addresses is that a plain 4-20 mA loop has no spare vocabulary. Every current from 4 to 20 mA already means something valid, so there is no value left over to mean broken. If a temperature transmitter on a raw loop loses its sensor and happens to output 4 mA, the control system sees a perfectly legitimate 0 percent and treats it as truth. The loop is dead, but the number is plausible, and nothing about the signal itself gives the operator a reason to doubt it.
This is exactly the failure mode that gets missed on unmanned sites. A frozen or failed measurement that lands somewhere inside the normal range can sit on a trend for hours looking like a calm, well-behaved process. Control loops keep acting on it, alarms never fire because the value is in range, and the first sign of trouble is often a physical consequence rather than an instrument alert. Reserving current outside the valid band is the cleanest way to make that failure visible without adding wires.
NE43 solves this at almost no cost, which is why it became the de facto behavior for smart transmitters. The same two conductors carry the same 4-20 mA signal, but the device now has a defined place to put a fault, and the input card has a defined rule for recognizing it. A modern instrument that supports NE43 essentially gets basic failure signaling for free on top of its analog output, and operators come to rely on it enough that a loop reading exactly 3.5 mA becomes an instantly recognizable I need attention flag.
For a SCADA system, NE43 is only useful if the input scaling knows about the fault bands. A well-configured analog input treats 3.8 to 20.5 mA as the process value and treats anything below 3.6 mA or above 21 mA as a distinct condition rather than clamping it to 0 or 100 percent. That means the point should raise a bad quality or sensor fault status, not simply report a scaled value at the edge of the range. Getting this wrong is a common mistake: if the input card clamps a 3.4 mA fault to 0 percent, the whole benefit of NE43 is thrown away and the operator once again sees a plausible low reading.
A cloud SCADA platform such as Merobix, which historizes readings from remote wellsites and tank batteries, benefits directly from NE43 because it can flag the fault current as bad data instead of storing it as a real measurement. The trend then shows a break in the signal rather than a false excursion, alarms fire on the loss of a healthy measurement, and an operator gets a notification that a specific transmitter needs attention. On sites that may go hours without anyone present, catching a dead loop the moment its current leaves the valid band is often the difference between a scheduled visit and an emergency callout.
NE43 also pairs naturally with the diagnostic states defined in NAMUR NE107, which sorts device conditions into categories such as failure, function check, out of specification, and maintenance required. NE43 gives the analog signal the ability to announce a Failure state through the current level itself, while NE107 provides the richer status - usually carried digitally over HART - that explains why. In a monitoring stack, the NE43 fault current is the coarse, always-available alarm on the wire, and the NE107 state is the detail that tells maintenance what to bring to the field.
NE43 defines the valid measurement information range as 3.8 to 20.5 mA. A downscale fault is signaled by driving the loop below 3.6 mA, and an upscale fault by driving it above 21 mA. Anything in those reserved bands is meant to be read as a transmitter or sensor failure rather than a process value.
That depends on the process and is usually a configurable choice. The rule of thumb is to pick the fail direction that drives the controlled process toward its safe state when the measurement dies. For example, on a loop where a high reading trips equipment off, failing upscale can force a safe shutdown, while other loops are safer failing downscale.
No. NE43 is purely about the analog current level, giving the 4-20 mA signal a defined out-of-range value to mean failure. HART diagnostics ride digitally on the same loop and carry detailed device status and NE107 states. They complement each other: NE43 is the coarse fault flag on the wire, and HART carries the explanation of what went wrong.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.