Instrumentation & Calibration

Every reading a bridge or engine control room displays has travelled through a chain: a sensing element responds to a physical change, a transmitter converts that response into a standard signal, wiring carries the signal to a marshalling cabinet, and a display or logic system turns it back into a number or an action. A fault, a drift, or a disputed value can sit at any link in that chain, and the instrument itself is only one of them.

When a reading looks wrong — a temperature that will not settle, a level that jumps, a pressure that disagrees with a local gauge — the first useful question is not "is the sensor broken?" but "where in the chain would this symptom appear?" A sensor fault, a wiring fault, an installation fault and a genuine process change can all produce a similar-looking trace on a screen, and a competent engineer works back through the chain rather than reaching straight for a replacement part.

Pressure only means something once its reference is stated. Gauge pressure is measured relative to local atmosphere and is what most engine-room gauges show; absolute pressure is measured relative to a perfect vacuum and is what a condenser or a refrigerant chart needs; differential pressure (DP) is the difference between two points and is what a level transmitter or an orifice-plate flow meter actually measures. Confusing gauge and absolute is a common source of "impossible" readings, particularly around vacuum systems.

Level is very often not measured directly at all — it is inferred from the hydrostatic pressure a liquid column exerts at the bottom of a tank, using a DP cell referenced to atmosphere (or to the vapour space, for a closed tank).

Because the formula divides by density, any error in ρ — the wrong liquid, a temperature-dependent density change, or a calibration table built for fresh water applied to a denser cargo or ballast — becomes a level error, even though the transmitter itself is functioning exactly as designed and calibrated. Flow measurement follows related logic: DP flow meters (orifice plates, venturis) infer flow from the pressure drop across a restriction, so the same reference and density sensitivities apply, while other technologies (turbine, electromagnetic, ultrasonic) sense flow more directly but bring their own installation requirements — usually a minimum straight run of pipe upstream and downstream to let the flow profile settle before it reaches the sensor.

Two temperature-sensing technologies dominate marine instrumentation, and they work on genuinely different physics. A thermocouple is two dissimilar metal wires joined at a measuring junction; the junction generates a small voltage — microvolts per degree — and that voltage is a function of the temperature difference between the measuring junction and the reference ("cold") junction, not of the measuring junction's absolute temperature alone. That is why a thermocouple circuit needs cold junction compensation: the instrument measures or assumes the temperature at its own terminals and adds the correction electronically, so the displayed value represents the true temperature at the measuring end. Without it, the reading silently carries the error of whatever temperature the terminal box happens to be at.

A resistance temperature detector (RTD), typically a Pt100 element, works differently: it exploits the fact that a pure metal's electrical resistance rises predictably with temperature. A Pt100 is defined as 100 Ω at 0 °C, rising by roughly 0.385 Ω per °C near that point, and because resistance — not a microvolt signal — is being measured, lead wire resistance sits directly in series with the element and adds straight onto the reading unless it is removed. A three-wire connection uses a bridge arrangement to cancel the lead resistance in the two current-carrying legs; a four-wire connection removes it completely by separating the current-carrying and voltage-sensing leads. A simple two-wire hookup is only acceptable for a short run where the lead resistance is negligible next to the element's own resistance change.

Almost every transmitter on board, whatever it senses, ends up speaking the same language: a 4–20 mA current loop. The choice of 4 mA as the zero point rather than 0 mA is deliberate. Current, unlike voltage, does not fall off along a long run of cable — the same current flows at every point in a series loop regardless of cable resistance, so the signal is immune to the volt drop that would otherwise distort a voltage-based reading over a long run between the engine room and a remote sensor.

The "live zero" is the other half of the design. Because the bottom of the range is 4 mA and not 0 mA, a genuine zero reading (empty tank, atmospheric pressure, zero flow) still produces a small, positive, alive current. A current of exactly 0 mA can only mean the loop itself has failed — a broken wire, a disconnected transmitter, a blown fuse — and the control system can distinguish that instantly from a real zero process value and raise a fault rather than quietly displaying a false "empty" or "zero" reading.

LRV and URV are the lower and upper range values the transmitter was configured for — not necessarily the physical limits of the sensor, but the span chosen for this particular installation.

Calibrating a transmitter means adjusting two separate parameters, and treating them as one is the single most common calibration mistake. Zero sets where the output sits at the bottom of the range — it shifts the whole output line up or down without changing its slope. Span sets how much the output changes for a given change in the input — it rotates the line about the zero point, changing the slope, or gain, of the relationship between the process value and the signal.

Because span is defined relative to zero, the correct order is always zero first, then span, then re-check zero — adjusting the span control on a transmitter that is not first sitting at the correct zero moves the zero point too, and a technician can chase their tail between the two adjustments if they lose track of which was set last. A proper calibration check does not stop at one point either: the instrument is checked at several points across its range — commonly 0 %, 50 % and 100 %, or five points for a more rigorous check — because a transmitter can be correct at zero and wrong at span, correct in the middle and wrong at the ends, or drifting non-linearly in a way a single-point check would never reveal.

Every calibration check is recorded twice: the as-found value, read before any adjustment is made, and the as-left value, read after adjustment. The as-found value is the evidence — it shows how far the instrument had actually drifted since the last check, which feeds decisions about calibration intervals and instrument reliability, and it is what a surveyor or auditor will ask to see. Recording only the as-left value throws that evidence away.

A transmitter can be perfectly calibrated on the bench and still give a wrong reading once it is installed, because installation introduces error sources that have nothing to do with the instrument's own accuracy. Impulse lines — the small-bore tubing connecting a process tap to a DP transmitter — can trap air in a liquid-filled line or trap condensate in a gas-filled line, and either changes the effective pressure the transmitter actually sees at its diaphragm. A thermowell that has worked loose, or was never fully seated in good thermal contact with the process, insulates the sensor from the temperature it is supposed to be measuring and produces a slow, sluggish, understated reading. A level transmitter referenced to the wrong liquid density will calculate a level that is systematically wrong by a fixed ratio across the whole range, even though every milliamp it produces is exactly what its own internal calibration says it should be.

The practical consequence is procedural: before condemning a transmitter that is "reading wrong", check the installation — vent the impulse lines, verify the thermowell is seated and the sensor is fully inserted, confirm the density or reference used in the level calculation matches the actual product — because replacing a transmitter that was never at fault fixes nothing and leaves the real cause live.

Calibration itself needs to be traceable: the standard used to check a transmitter (a deadweight tester, a precision loop calibrator, a certified thermometer) must itself carry a current calibration certificate that, ultimately, traces back to a national or international standard. A calibration record without a traceable standard behind it is just a number — it cannot demonstrate that the checking instrument was itself trustworthy on the day the check was made.

Three worked checks: catching a zero error and separating it from a span error, correcting a Pt100 reading for lead resistance on a long two-wire run, and tracing a DP level display back to the true liquid height when the product density does not match the calibration.