Almost every pressure, level, flow and temperature reading in a plant travels to the control system as a current between 4 and 20 milliamps on one pair of wires. The convention is fifty years old, brand-neutral, and works over a kilometre of ordinary cable, which is why it has outlived several generations of digital networks that were meant to replace it. It is also the source of a large share of commissioning faults, nearly all of them in the scaling. This guide explains what the loop is, why it is built the way it is, and how the number on the PLC screen is derived from the current in the wire.
Why current, not voltage
Send a measurement as a voltage and every ohm between the sender and the receiver takes a share of it. The wire has resistance, the connections have resistance, and the two ends of a long cable rarely sit at the same ground potential, so a 0–10 V signal arrives as 9.8 V plus whatever the nearest motor cable has induced along the way. The receiver cannot tell the loss from a real change in the process.
A current has no such problem. The same current flows through every element of a series circuit, so the receiver at the far end sees exactly what the transmitter regulates at the near end, regardless of the wire in between. The transmitter behaves as a current regulator: if the cable resistance rises, it raises the voltage at its terminals to keep the current the same, up to the limit of the supply. Induced noise has to drive a current through a low-impedance loop rather than a voltage into a high-impedance input, so its effect is small, and a ground-potential difference between the two ends is absorbed by the transmitter's regulation rather than added to the signal. The result is a signal that can run several hundred metres, and with a decent supply over a kilometre, on a single twisted pair, with an accuracy set by the transmitter and not by the installation.
Why 4 mA is zero: the live zero
The range could have been 0–20 mA, and that standard exists in IEC 60381-1. The industry settled on 4–20 mA because of two things a zero-based range cannot do.
The first is power. A two-wire transmitter has only the loop: the same pair brings its supply in and takes its signal out. It draws its own operating current from the loop and regulates the total to represent the measurement, so the total can never fall below what the electronics need. At 4 mA and the 12 V or so left across its terminals, the transmitter has under 50 mW to run a sensor, a microcontroller and often a display, which is enough. A 0–20 mA transmitter must be powered separately, on four wires.
The second is diagnosis. With a live zero, a reading of 4 mA means the process is at the bottom of the range, and a reading of 0 mA means the loop is open: a broken wire, a loose terminal, a failed supply or a dead transmitter. The receiver can tell the two apart and alarm on the second. On a 0–20 mA or 0–10 V signal, an empty tank and a cut cable look identical. The live zero also leaves room below 4 mA for the transmitter to signal that the process has gone under its range, and for the failure levels that NAMUR NE 43 defines.
Anatomy of a loop: supply, transmitter, receiver
A basic two-wire loop is one series circuit with three elements. A 24 V DC supply, usually in the control panel, provides the voltage. The transmitter in the field sits in the loop and regulates the current between 4 and 20 mA in proportion to what it measures. The receiver, which is the PLC's analogue input or a 250 Ω resistor in front of it, turns that current back into a voltage by Ohm's law: 4–20 mA across 250 Ω is 1–5 V, and that is the voltage the input card actually digitises. Many current inputs carry the resistor internally; its value is in the module datasheet, and it matters for the next paragraph.
The supply has to push 20 mA through everything in the loop and still leave the transmitter enough voltage to work. Each element drops voltage in proportion to its resistance: the 250 Ω receiver takes 5 V at 20 mA, a kilometre of 1.5 mm² cable (500 m out and 500 m back, about 12 Ω) takes 0.24 V, and a typical transmitter needs about 12 V at its own terminals, more if it has a display or HART. On a 24 V supply that leaves 24 − 5 − 0.24 = 18.8 V for the transmitter, comfortably above its minimum. Add a local indicator, a signal isolator or an intrinsic-safety barrier, and each takes its own share; add a second 250 Ω receiver and the transmitter is down to 13.8 V. A third does not fit. This voltage budget is the loop's "compliance", and it is the first thing to check when a transmitter reads correctly on the bench and saturates in the field.
Where the PLC fits depends on its input. Some analogue input modules supply the 24 V to the loop themselves and expect a two-wire transmitter; others are passive and expect an external supply, or an active four-wire transmitter that sources the current. Wiring an active output into an input that also powers the loop puts two sources in series and produces readings that make no sense. The module manual has a wiring diagram for each case; follow it.
From milliamps to bar: scaling in the PLC
The analogue input card produces an integer. On a Siemens S7 controller the nominal range 4–20 mA (or 1–5 V) becomes 0 to 27 648; a 12-bit card gives 0 to 4 095; a 16-bit card 0 to 65 535 or a signed value; some cards deliver 4 000 to 20 000 so that the count reads as microamps. The PLC program then maps that integer to engineering units with a scaling instruction that takes two points: the raw count and the engineering value at range low, and the same pair at range high. Everything in between is a straight line.
value = EUlo + percent × (EUhi − EUlo)
The engineering values at the two points must be the transmitter's lower and upper range values, the numbers it was ranged to, and not what somebody remembers from the datasheet. This is the classic mismatch bug. A transmitter ranged 0–16 bar feeding a PLC that scales 0–10 bar reads 62.5 % of the truth at every point and looks entirely healthy on the screen. A PLC scaling block told the signal is 0–20 mA when it is 4–20 mA reads 6.4 bar where it should read 5.5 and, worse, shows 2 bar with the transmitter at zero. Neither error is caught by looking at the trend; both are caught in ten minutes with a loop calibrator injecting 4, 12 and 20 mA at the field end and checking for 0, 50 and 100 % on the screen. HART makes it easier still, because the range values can be read out of the transmitter and compared with the PLC's in the office.
Resolution is rarely the problem. One count on a 0–27 648 scale is 0.0036 % of span; on a 12-bit card, 0.024 %. The transmitter's accuracy and the process noise are both larger. The signal scaling calculator does the arithmetic in both directions for any signal type and any raw range.
Faults and NAMUR NE 43
A live zero gives the loop somewhere to put a failure signal, and NAMUR NE 43 (2003) says where. The measuring information occupies 3.8 to 20.5 mA: a transmitter that runs off the bottom or top of its range saturates at 3.8 or 20.5 mA and stays there, so the receiver knows the process has left the range but not by how much. Below 3.6 mA and above 21.0 mA are failure signals. A transmitter that detects an internal fault, a sensor break or a loss of its measurement drives the loop to one of them, and which one is a configuration choice on the transmitter. 0 mA remains what it always was: an open loop.
The PLC must be programmed to use this. Most input cards set a status or flag the value as under- or overflow outside the nominal range; the program should alarm on it, hold the last good value or a safe default rather than scaling the fault current into a plausible number, and choose the alarm direction so that the control action fails safe. A level transmitter set to fail high on a tank whose pump stops at high level will, when it fails, stop the pump; the same transmitter set to fail low will run the pump into an empty tank.
Noise, grounding and shields
The loop is robust, not immune. Use shielded twisted pair, and connect the shield to the instrument earth at one end only, normally the panel end, so that the shield carries no circulating current between two grounds. Give each loop one ground point, usually through the supply or the input card, and if the transmitter body is grounded elsewhere in the plant, put an isolator in the loop or use an isolated input. Route signal cables in their own tray or conduit away from power cables, and especially away from the output cables of variable-frequency drives, crossing them at right angles where you must. On the input card, set the interference-frequency suppression to 50 Hz, the mains frequency in Pakistan, so that the converter integrates out the hum that a long cable picks up.
When a reading jumps or drifts, the order of suspicion is: a loose terminal, a shield grounded at both ends, a cable sharing a tray with a drive output, and only then the transmitter.
The example in numbers
A pressure transmitter ranged 0–10 bar, 4–20 mA, into a Siemens S7 input (0–27 648) through 250 Ω, with a meter in the loop reading 12.8 mA:
- Percent of span: (12.8 − 4) ÷ 16 = 55.0 %.
- Engineering value: 0 + 0.55 × 10 = 5.50 bar. Raw count: 0.55 × 27 648 = 15 206.
- Across the 250 Ω resistor: 12.8 mA × 250 Ω = 3.20 V. One count is 0.000362 bar.
- Reverse: to show 7.5 bar the loop must carry 4 + 0.75 × 16 = 16.0 mA, a raw count of 20 736.
If the screen says 6.4 bar, the scaling block thinks the signal is 0–20 mA; if it says 8.8 bar, the transmitter has been ranged 0–16 bar and nobody told the PLC. The calculator reproduces every step and also flags a reading in the NE 43 under-range, over-range or failure bands.
When 4–20 mA is not the answer
A loop carries one variable, one way, per pair of wires. A modern transmitter measures several things and knows a great deal about its own health, and a modern plant wants to see all of it. Digital fieldbuses (PROFIBUS PA and FOUNDATION Fieldbus for process instruments, PROFINET and EtherNet/IP for everything faster) put many devices and all their variables on one cable; IO-Link (IEC 61131-9) does the same at the machine level for sensors and actuators, over a plain three-wire cable to a master that speaks the plant network; HART gives most of the diagnostic benefit over the loop you already have. Where a plant has dozens of instruments per area, or needs the diagnostics, the digital options win on cabling and on information.
Where a plain loop still wins is everywhere else: one or two instruments on a pumping station, a remote tank, a boiler house, a water scheme, any site where the people who maintain it own a multimeter and not a configuration tool. It works across every brand made in the last fifty years, it needs no addressing, no network design and no software, it can be checked with a meter and a loop calibrator, and it passes through intrinsic-safety barriers that are understood by everyone. Ahmedonics designs and integrates both, and the choice is a matter of the site and the people who will keep it running, not of fashion.
References
- IEC 60381-1:1982, Analogue signals for process control systems — Part 1: Direct current signals
- IEC 60381-2:1978, Analogue signals for process control systems — Part 2: Direct voltage signals
- NAMUR NE 43 (2003), Standardization of the signal level for the failure information of digital transmitters — measuring band, saturation and failure levels
- ANSI/ISA-50.00.01-1975 (R2012), Compatibility of Analog Signals for Electronic Industrial Process Instruments — formerly ISA-50.1; two-wire and four-wire transmitters, load and compliance
- Siemens S7-1200 System Manual, analogue value representation — raw counts, over-range and overflow bands, interference-frequency suppression
- IEC 61131-9, Programmable controllers — Part 9: Single-drop digital communication interface for small sensors and actuators (SDCI) — IO-Link
- FieldComm Group (formerly the HART Communication Foundation), HART protocol specifications