Process instruments — temperature, pressure, level, flow

A thermocouple is two dissimilar metals joined at a tip. The junction generates a few tens of microvolts per °C (Seebeck effect), and the card measures that against a known cold-junction temperature. It is cheap, physically tough, and covers enormous ranges — a Type K runs to about 1260 °C, Type N and Type S higher still.

An RTD is a resistor whose value rises predictably with temperature. Pt100 means 100 Ω at 0 °C; Pt1000 is the same curve scaled ten times. RTDs beat thermocouples on accuracy and stability below about 500 °C, which is why almost every food, pharma and HVAC application uses a Pt100 and almost every furnace uses a Type K.

The rule people break is extension wire. A thermocouple must be extended with matching compensating cable of the same type, all the way to the card. Splice copper into a Type K loop and you have created a new, uncontrolled junction — the reading will be plausible and wrong, which is far worse than obviously broken.

A card measures an RTD by pushing a small current through it and reading the voltage. With two wires, the resistance of the cable is added to the resistance of the sensor, and the card cannot tell the difference. Ordinary 0.5 mm² copper is about 0.037 Ω/m per conductor, so a 50 m run adds roughly 3.7 Ω to the loop — about 9.5 °C of error on a Pt100.

A third wire lets the card measure the lead resistance separately and subtract it, on the assumption all three leads are identical. Four wires remove the assumption entirely: two carry the excitation current, two sense voltage at the sensor with no current flowing in them, so lead resistance cancels exactly. That is the laboratory arrangement, and it is worth using on any long or critical Pt100 run.

Pt1000 sidesteps much of this by arithmetic: the same 3.7 Ω of lead sits against a sensor ten times larger, so the same cable costs a tenth of the error. On long runs where 4-wire is awkward, a Pt1000 is often the cheaper right answer.

Gauge pressure is measured against atmosphere — the number on a tyre gauge. Absolute pressure is measured against a sealed vacuum, and it is what you need for anything where atmospheric change matters: vacuum systems, boiling-point control, altitude-sensitive processes. Differential pressure is the difference between two ports, and it is the most useful of the three because so many other measurements are secretly a DP measurement.

A modern transmitter such as a Rosemount 3051 or Siemens SITRANS P is a silicon or capacitance cell plus its own electronics, and its accuracy claim (typically ±0.05 to ±0.5 % of span) assumes you use a sensible fraction of its range. Turn-down matters: a cell ranged 0–10 bar and used over 0–0.5 bar is running at 20:1 turn-down, and the error as a percentage of your actual reading multiplies accordingly. Size the cell to the job.

Hydrostatic / DP level uses the fact that a column of liquid produces pressure proportional to its height and density: p = ρ g h. Cheap, robust, and completely dependent on knowing the density. Change product from water to a 1.2 relative-density syrup and every level reading is 20 % low until someone updates the calculation.

Guided-wave and non-contact radar time a microwave pulse to the surface, which makes them largely indifferent to density, temperature and pressure. Radar is now the default for new installations on anything valuable, with guided-wave (a probe in the liquid) chosen for low-dielectric products, small tanks and turbulent surfaces.

Ultrasonic level does the same trick with sound and costs less, but the speed of sound moves with air temperature and the pulse dies in foam or heavy vapour. Capacitance and vibrating-fork switches remain the sensible answer for simple point detection: full, empty, or alarm.

Electromagnetic flowmeters (magmeters) apply a magnetic field across an insulated tube and measure the voltage a conductive fluid generates as it moves. No obstruction, no moving parts, excellent accuracy — but the fluid must conduct, so they are perfect for water, slurries and most food products and useless on oil or pure hydrocarbons.

Coriolis meters vibrate a tube and measure the twist the flowing mass induces. They read true mass flow and density simultaneously, are accurate to a few hundredths of a percent, and cost accordingly. If you are buying a fluid rather than just controlling it, this is usually the meter in the contract.

Differential-pressure flow (orifice plate plus DP cell) is the oldest method and still very common on gas and steam. The one thing everybody must remember about it: flow is proportional to the square root of the DP. That is why a plain 4–20 mA from a DP cell across an orifice cannot be treated as linear flow, and why so many old loops have a square-root extraction block either in the transmitter or in the PLC — and occasionally, disastrously, in both.

Almost every modern transmitter is HART-capable, which means it superimposes a small FSK digital signal on the existing 4–20 mA loop without disturbing the analog value. The analog reading keeps working exactly as before; the digital layer carries the range settings, the tag, up to three extra variables and — the useful part — the device's own diagnostics.

A handheld communicator or a HART-capable AI card gives you sensor temperature, diaphragm health, plugged-line detection and calibration history on instruments you already own. Most plants pay for this capability and never enable it. If you are specifying new AI cards, HART-capable channels on the critical loops pay for themselves the first time you avoid pulling an instrument that turned out to be fine.