Signal Conditioners

What a Signal Conditioner Does

Signal conditioning is the practice of preparing a raw input signal for use by instrumentation in a data acquisition or control system. A signal conditioner sits between a sensor and the next device in the chain (a PLC analog input card, a DCS, a recorder, or a display) and converts the sensor output into a clean, standardized format the next stage can read with high fidelity. Without conditioning, a thermocouple's microvolt-level output, a load cell's millivolt-per-volt bridge signal, or a 4-20 mA loop riding on a different ground would be unusable, noisy, or actively damaging to the receiving hardware.

The conditioner is best understood as a toolbox of operations rather than a single fixed function. The most common operations are: amplification, which raises a low-level signal and improves the signal-to-noise ratio so an analog-to-digital converter sees usable resolution; attenuation, which scales a large signal down into the input window; filtering, which removes noise outside the band of interest, including anti-aliasing low-pass filtering ahead of digitizers; isolation, which galvanically separates input from output to block ground loops and surges; excitation, which supplies the voltage or current a passive sensor such as a strain-gauge bridge or RTD needs to produce a signal; and linearization, which corrects the non-linear relationship between a sensor's electrical output and the physical quantity it measures.

On top of these, many conditioners perform range conversion (for example 0-10 V to 4-20 mA, or 0-20 mA to 4-20 mA), splitting (one input driving two electrically isolated outputs so a signal can feed both a controller and a recorder), and computation such as totalization, integration, square-root extraction for differential-pressure flow, and limit switching. A configurable or universal conditioner exposes these through DIP switches, rotary settings, or PC and NFC configuration software so one part number covers many input and output combinations.

Why a separate device rather than building conditioning into the sensor or the PLC card? Three reasons recur. First, ground-loop and surge protection belongs at a defined boundary, and a DIN-rail conditioner gives a serviceable, replaceable point of isolation. Second, hazardous-area energy limiting must happen in a certified interface between the safe and hazardous zones, which is exactly where isolated barriers live. Third, standardizing every field signal to 4-20 mA or a common voltage range lets the control system use uniform, lower-cost input cards instead of a mix of specialized cards. The conditioner is therefore as much an architectural decision as an electrical one.

It helps to place the conditioner in the full acquisition signal chain. A measurement starts at the transducer, which converts a physical quantity into an electrical quantity, then passes through the conditioner, then to the digitizer (the analog-to-digital converter inside a PLC card, a recorder, or a data acquisition module), and finally to processing, storage, or display. The conditioner is the only stage whose explicit job is to match the messy real-world signal to the clean, bounded window the digitizer expects. A 16-bit converter with a 0-10 V input window resolves about 0.15 mV per count, so a 10 mV thermocouple signal fed in raw would use less than one percent of that range and bury the measurement in quantization noise; amplifying it to fill the window recovers the resolution the converter was designed to deliver. This is why conditioning, not converter bit depth alone, often sets the real measurement resolution of a system.

Filtering plays a parallel role on the time axis. Sampled systems must respect the Nyquist criterion: any signal content above half the sample rate folds back, or aliases, into the measurement band as false low-frequency error that no amount of later processing can remove. An anti-aliasing low-pass filter in the conditioner, placed before the converter, suppresses that out-of-band content so what is digitized is genuinely what was measured. Modern conditioners increasingly use digital filters, which are easy to adjust and need no extra hardware, but the anti-aliasing stage in front of the converter must still be analog because it has to act before sampling occurs.

Four engineering metrics dominate conditioner quality: transfer accuracy (how faithfully output tracks input), isolation rating (how much potential difference it withstands), temperature stability (drift over the operating range), and response time (how fast output follows a step change). The remaining chapters decode each of these and map them to real selection choices.

Types and Topologies

Signal conditioners are classified along three axes that matter at the purchase order: how they are powered, what they do to the signal, and whether they are rated for hazardous areas. The power axis splits passive (loop-powered) from active (externally powered). The function axis splits simple isolators from converters, splitters, and universal input modules. The hazardous-area axis splits general-purpose conditioners from intrinsically safe isolated barriers and Zener barriers. The table below compares the main families on the parameters that drive selection.

Passive loop-powered isolators are the simplest member of the family. They draw operating energy from the 4-20 mA loop itself, so they need no separate supply and add no power-budget complexity. The price is voltage drop, also called burden or insertion loss: representative passive isolators consume roughly 1.6 to 4 V (for example a Phoenix Contact MINI MCR-SL passive isolator at about 1.7 V, a Weidmuller passive isolator near 4 V), which the loop's supply voltage must cover in addition to the field device and any sense resistor. They isolate and filter but generally do not convert ranges.

Active isolating amplifiers run on 24 V DC and add value: they convert between current and voltage ranges, present a very low input burden, drive longer loops, and often configure several input and output ranges on one part via DIP switches. The Phoenix Contact MINI MCR-SL-UI-UI, for instance, isolates and converts among 0-20 mA, 4-20 mA, 0-10 V, 2-10 V, 0-5 V, and 1-5 V using side-accessible switches. Splitters extend this to one input feeding two galvanically isolated outputs, which removes the classic problem of a shared loop where one open instrument kills the whole circuit.

Universal temperature conditioners accept thermocouple millivolts, RTD resistance, raw millivolts, ohms, and potentiometer inputs, performing cold-junction compensation and linearization internally (covered in Chapter 4). Isolated barriers and Zener barriers are the hazardous-area members: both limit energy into the field circuit so it cannot ignite an explosive atmosphere, but the isolated barrier adds galvanic isolation, tolerates field-side common-mode voltage, and needs no dedicated intrinsically-safe ground, while the Zener barrier shunts fault energy to a high-integrity IS earth. Chapter 3 details the isolation mechanisms behind these.

Isolation Methods

Galvanic isolation is the defining feature of most conditioners, so the coupling method deserves its own chapter. Isolation breaks the conductive path between input, output, and power so that a difference in ground potential cannot drive current through the signal path. This stops ground loops (the leading cause of drifting or noisy 4-20 mA readings in large plants), protects the receiving hardware from surges and lightning-induced transients, and, in barriers, prevents hazardous-area energy from crossing into the field. Three coupling technologies dominate: transformer (magnetic), optical (optocoupler), and capacitive.

Isolation voltage is rated two ways and the distinction trips up buyers. The test (dielectric) voltage is a one-minute high-potential withstand figure used in type testing, commonly quoted around 1.5 kV, 2.5 kV, or higher between ports. The working (continuous) isolation voltage is the steady-state potential difference the device is rated to tolerate indefinitely, governed by IEC 61010 insulation coordination and the overvoltage category, and is often a few hundred volts even when the test figure is several kilovolts. For a three-port device, confirm the rating applies separately input-to-output, input-to-power, and output-to-power, because some datasheets quote only the highest pair.

Transformer isolation modulates the signal onto a carrier, passes it across a magnetic gap, and demodulates it on the far side. It achieves high isolation, can also transfer the power needed by the far side (enabling some single-supply or loop-powered designs), and is robust against long-term drift, at the cost of limited bandwidth and a more complex circuit. It is the traditional choice in process barriers where high common-mode rejection matters.

Optical isolation drives an LED whose light is detected on the isolated side. It offers wide bandwidth and a simple signal path, making it common in faster or digital-signal conditioners, but the LED's light output ages, so high-accuracy analog optical designs add feedback to correct for drift. Capacitive isolation couples through small series capacitors and is compact and fast, favored in modern integrated isolators, though it generally tolerates less common-mode transient than transformer designs. Many real products combine methods, for example optical or capacitive signal coupling with a transformer for isolated power. When the field and control sides can sit at genuinely different mains potentials, prioritize the working voltage and overvoltage category over the headline test-kilovolt figure.

Sensor-Specific Conditioning and Standards

Beyond generic current and voltage isolation, the largest specialization in conditioners is by sensor type, because each sensor presents a different electrical problem. Thermocouples need cold-junction compensation and linearization. RTDs need excitation and lead-wire compensation. Strain gauges and load cells need bridge excitation and high-gain amplification. Each of these maps to a specific public standard, which is the language an RFQ should use.

Thermocouples produce a small, non-linear voltage (microvolts to tens of millivolts) proportional to the temperature difference between the measuring junction and the reference (cold) junction. A thermocouple conditioner performs cold-junction compensation (CJC): an RTD or thermistor at the terminal block continuously measures the terminal temperature, and the device adds the corresponding correction so the reading reflects the absolute hot-junction temperature rather than a difference. It then linearizes the output using the standard thermoelectric reference functions of IEC 60584-1, which defines EMF-versus-temperature tables and polynomials for letter types including J, K, T, E, N, R, S, and B. Choosing the wrong type code in the conditioner, or omitting CJC, is a frequent field error worth tens of degrees.

RTDs (most commonly Pt100 and Pt1000 platinum elements) change resistance with temperature. A conditioner sources a small, stable excitation current and reads the resulting voltage, then linearizes per the Callendar-Van Dusen equation defined in IEC 60751, which also fixes the tolerance classes (for example Class AA, A, B). Lead-wire resistance is a major error source, so conditioners support 2-, 3-, and 4-wire connection: 2-wire makes no compensation and adds lead resistance directly to the reading, 3-wire cancels most lead resistance assuming equal-length leads, and 4-wire eliminates lead error entirely by sensing voltage on separate wires from the excitation. Current excitation is generally preferred over voltage excitation in noisy industrial environments for better noise immunity.

Strain gauges and load cells output a tiny ratiometric signal, typically 1 to 3 mV per volt of bridge excitation at full load, so the conditioner must supply a clean, stable bridge excitation voltage and apply high gain with low noise and low drift. Because the output is ratiometric, excitation stability directly limits accuracy, and metrological weighing follows OIML R60. Process signals (4-20 mA, 0-10 V) need isolation, conversion, and splitting rather than excitation. Hazardous-area field circuits add the energy-limiting requirements of the IEC 60079 series, specifically IEC 60079-11 for intrinsic safety and IEC 60079-25 for intrinsically safe systems. Where the conditioner sits in a safety-instrumented function, its random-failure data must be assessed against IEC 61508, and many isolated barriers carry a documented SIL 2 capability for exactly this reason. Always treat the sensor tolerance and the conditioner error as separate budgets that add together.

Key Specification Parameters

A conditioner datasheet may list 20 or more lines, but a manageable set drives selection: transfer accuracy and temperature coefficient, isolation rating, response time and bandwidth, input burden and output load, power and consumption, configurability, environmental and EMC ratings, and certifications. Each is decoded below.

Transfer accuracy (also transfer error or transmission error) is the core fidelity number, usually quoted as a percentage of full scale (% FS). General current and voltage isolating conditioners commonly specify 0.05 to 0.2 % FS at a 23 degrees C reference. Pair it with the temperature coefficient (TC), typically 50 to 200 ppm of FS per degree C: for example an Omega strain conditioner specifies a 200 ppm/degree C gain TC. Multiply TC by the deviation from reference to get added drift, so a 0.1 % FS module with 100 ppm/K TC operating 40 K above reference loses an extra 0.4 % FS. Always confirm whether the headline accuracy already folds in linearization and, for temperature units, cold-junction error.

Isolation rating separates test (dielectric, one-minute) from working (continuous) voltage, as Chapter 3 explained. Confirm the figure for all three port pairs. Industrial ground-loop breaking is satisfied by about 1.5 kV test isolation; choose 2.5 kV or higher and check the working voltage and overvoltage category where input and output can sit at different mains potentials. Some load-cell conditioners reach 3500 V isolation for demanding separation.

Response time and bandwidth matter for control and dynamic measurement. DC-coupled process isolators often quote step response from tens to hundreds of milliseconds (for instance an Omega load-cell module at 75 ms or less), while dynamic strain and vibration conditioners are specified by bandwidth (for example DC to 3 Hz for a slow strain module versus DC to 2 kHz for a wideband version). A slow conditioner in a fast loop introduces phase lag and degrades PID control; a needlessly wide one passes more noise.

Input burden and output load set wiring feasibility. For a passive isolator, burden (voltage drop, often 1.6 to 4 V) eats into the loop supply budget. For active modules, check the maximum output load the device can drive at 20 mA (commonly 500 to 800 ohm) and the minimum input the source must provide. Power and consumption determine the 24 V DC budget across a packed DIN rail; modules often accept a wide 19.2 to 30 V DC supply range.

• Configurability: fixed-range, DIP-switch selectable, or fully software/NFC configurable universal input. Universal modules cut spares inventory but cost more per unit.
• Width and mounting: ultra-narrow 6.2 mm housings (for example Phoenix Contact MINI MCR) maximize channels per meter of rail; wider modules add displays or relays.
• Environmental: operating range often -20 to +60 degrees C, with extended ranges to -40 to +85 degrees C; verify the temperature at which accuracy is still guaranteed, not just survival.
• EMC and surge: immunity per the IEC 61326 series for measurement equipment; surge protection on field terminals where lightning exposure exists.
• Certifications: hazardous-area ATEX, IECEx, NEPSI, FM; functional safety SIL per IEC 61508; plus regional UL and CE marks.

A few spec-sheet conventions cause repeated confusion and deserve naming. Accuracy quoted as percent of full scale (% FS) and accuracy quoted as percent of reading (% of reading, sometimes % SP) are not interchangeable: a 0.1 % FS figure on a 0-20 mA module is a fixed plus-or-minus 20 microamp band across the whole range, so at the low end of the scale it is a large fraction of the reading, whereas a percent-of-reading figure shrinks with the signal. For temperature inputs the datasheet often separates a fixed term and a proportional term, for example plus-or-minus 0.1 degrees C plus 0.05 % of reading, and you must combine both at your operating point. Long-term stability, usually expressed as drift per year, is yet another independent line that recalibration intervals depend on.

Two further numbers matter when the conditioner sits in a noisy plant. Common-mode rejection ratio (CMRR) describes how well the device ignores voltage common to both input terminals, which is exactly the interference a ground-potential difference injects; a high CMRR is what lets an isolating conditioner deliver a clean reading where a non-isolated path would not. Ripple and residual noise on the output, often specified as a peak-to-peak figure, set the practical floor below which small signal changes cannot be trusted. For loop-powered devices, also check the minimum loop voltage the module needs to operate, because falling below it during a supply sag silently corrupts the reading rather than producing an obvious fault.

Treat these as independent budgets. A module can have excellent reference accuracy yet poor temperature stability, or generous isolation yet a slow response that ruins a control loop. The spec sheet is only meaningful when each line is checked against the actual application conditions rather than the headline at the top of the page.

Selection Decision Factors

To turn the preceding chapters into a specific model, follow the sequence below. As with other instruments, most selection mistakes come not from a single wrong line but from deciding range or function before the safety and isolation context is settled. These steps double as an RFQ template.

1. Hazardous area first: establish whether the field circuit is in a classified zone. If yes, you need an intrinsically safe barrier (isolated or Zener) certified to IEC 60079-11/-25 with ATEX/IECEx/NEPSI/FM as required, and this constraint overrides later choices. If no, a general-purpose conditioner suffices.
2. Function: decide what the device must do, isolate only, convert range, split into two outputs, or condition a specific sensor (thermocouple, RTD, strain gauge). The function sets whether a passive isolator or an active conditioner is appropriate.
3. Input and output: specify exact ranges (4-20 mA, 0-10 V, Pt100 3-wire, type K thermocouple) and confirm bidirectional compatibility with both the field device and the receiving card.
4. Power and burden: if no spare 24 V exists and the loop has voltage headroom, a passive loop-powered isolator wins; if you need conversion, splitting, or excitation, budget the 24 V DC supply and check output drive capability.
5. Accuracy and stability: match transfer accuracy (for example 0.1 % FS) and temperature coefficient to the loop's accuracy budget, remembering to add the sensor tolerance separately for temperature and strain inputs.
6. Isolation level: 1.5 kV test isolation for routine ground-loop breaking; 2.5 kV or higher, with working-voltage and overvoltage-category verification, where input and output can sit at different mains potentials.
7. Response and bandwidth: fast enough for the control loop or dynamic signal, but no wider than needed so noise is not passed; confirm step response or -3 dB bandwidth against the process time constant.
8. Functional safety and environment: if the loop is part of a safety-instrumented function, require the IEC 61508 SIL assessment for the exact part; confirm the operating temperature at which accuracy holds, the EMC class (IEC 61326), and mounting width per channel.

One last commonly overlooked dimension is manufacturer serviceability and ecosystem: configuration software and NFC tools, DIN-rail power-feed and bus-connector systems that simplify packed cabinets, spare-part availability, and the exact certificate coverage per part number rather than per series. Pepperl+Fuchs (K-System and H-System), MTL, R. STAHL, Phoenix Contact (MINI MCR, MINI Analog Pro, MACX Analog), Weidmuller (ACT20M), Knick, PR electronics, and Acromag all maintain broad ranges, and the right choice usually balances certificate fit, accuracy, channel density, and local support rather than headline price alone.