Pressure Calibration: Gauges, Transmitters and the Measurement Chain

Pressure calibration can be interpreted only when the pressure type, configuration and value actually compared are stated. A transmitter certificate does not, by itself, demonstrate that the system records its signal correctly. Start from intended use and reconstruct the path from the pressure connection to the data used.

1. Absolute, gauge or differential?

Absolute pressure is referenced to vacuum; gauge pressure is the difference from local atmospheric pressure. Differential pressure is the difference between two specified pressures: declare the positive side and convention, for example Δp = p_H − p_L. Zero differential pressure does not mean zero absolute pressure on both sides.

For a differential transmitter, the same difference may exist at very different line pressures. Performance verified with one side at atmospheric pressure does not automatically demonstrate performance at high line pressure. Specify differential range, relevant common pressure, units and conditions. Negative gauge pressure is possible; it is not negative absolute pressure.

Pa, kPa and bar express the same quantity with defined conversion factors: 1 kPa = 1 000 Pa and 1 bar = 100 000 Pa. Unit conversion does not turn a gauge measurement into an absolute one: that also requires an appropriate atmospheric value with its uncertainty.

2. Measurement chain and safe connections

Identify the sensing element, indicator, transmitter, power supply, output, acquisition channel, scaling and configuration. If local display, analogue output and digital communication exist, state which are included. Electrical simulation checks the part downstream of the injection point: it does not include sensor response, pressure connections or pressure lines.

Select references suitable for the quantity, range, medium and required uncertainty: a pressure balance, electronic standard or differential standard, according to the method. For small differentials derived from two large pressures, the budget must justify the subtraction result; two individually valid instruments do not guarantee a sufficiently well-known difference. Consider correlations where present.

The medium, seals, fittings and tubing must be compatible with both instruments and the assembly’s limits. Agree on isolation and reinstatement with the system owner. Do not disconnect under pressure, exceed differential or line-pressure limits or disable protection. Stopping the source does not prove that every isolated volume is depressurised.

3. A matrix for defining the comparison

Use this original matrix when requesting and reviewing the service. Select the reference through the required performance and actual conditions, not just full scale or an accuracy ratio assumed to be universal.

An elevation difference between reference and instrument introduces a hydrostatic pressure difference. In a simple model of static fluid with constant density, p_DUT = p_ref + ρg(z_ref − z_DUT), with elevation z increasing upwards. A lower point has higher pressure. Assess density, elevation, gravity and configuration; the correction and its uncertainty may be significant even without a leak.

Check tightness, thermal stability, orientation, relevant mounting torque and trapped fluids according to the approved safe method. Apparent drift may arise from a leak or incomplete equilibration. Do not compensate by changing zero without understanding its origin.

4. Points, cycles and original data

Define the operating interval, decision points, sequence and number of series according to the method and uncertainty objective. Guides propose different programmes: do not extract a minimum number and turn it into a rule for every instrument. Include zero and critical points where relevant, respecting component limits.

Ascending and descending series allow hysteresis to be evaluated; repetitions under the same conditions inform repeatability. Record direction, stabilisation, actual reference and indication, without replacing the latter with the setpoint. If zeroing, preconditioning or adjustment occurs, retain the initial state and interventions: the meaning of the result also depends on that history.

Reporting paths separately and using a mean curve are different treatments. Averaging does not eliminate hysteresis; if used, the model must account for its effect. Do not apply an interpolated correction outside the justified range or present a static calibration as evidence of response to rapid transients.

5. Error, correction and uncertainty

Define e = indication − reference and c = −e, after bringing readings to the same pressure type, level and unit. For an electrical output, state the transfer function, offset and range. Percentage of reading and percentage of span are not interchangeable; near zero, a percentage of reading may convey little useful information.

The budget considers the reference in use, drift, resolution, zero, repeatability, hysteresis, environmental conditions, elevation and any electrical chain. Do not automatically add contributions already included, and assess correlations. Distinguish standard uncertainty u from expanded uncertainty U = ku, stating k and the meaning of the interval. Resolution does not replace uncertainty.

A conformity statement requires an agreed specification and decision rule, including treatment of uncertainty and risk. A result at one point does not authorise the entire range. Subsequent use may require further contributions, such as temperature, drift and installation conditions.

6. Simulated case: correct transmitter, incorrect scale

A differential transmitter is configured linearly for 0–100 Pa with a 4–20 mA output, without square-root extraction. At a reference pressure of 50 Pa, 12 mA is measured. The intended conversion is Δp = (I − 4 mA) / 16 mA × 100 Pa, giving 50 Pa. However, the system is configured for 0–250 Pa: it records 125 Pa, with recorded-value error e = +75 Pa.

These simulated numbers identify a scaling inconsistency; they are neither a certificate nor a full demonstration of transmitter conformity, because uncertainty and results at other points are needed. Adjusting the sensor to make the incorrectly configured system read 50 Pa would introduce a second problem.

Retain the configuration and data before correction, assess use of affected records and authorise the scaling change. Then repeat a physical comparison through the chain at relevant points, with defined uncertainty and criteria; current injection alone does not complete this verification. Document before/after results and the fitness-for-use decision, without rewriting historical data.

Connect the report to the calibration strategy, metrological traceability and the uncertainty budget. Other pathways are in the Calibration & Laboratory Metrology hub.

7. Sources and scope

Checked: 1 October 2026. Technical documents were consulted within the stated scopes; their specific programmes are not universal GMP procedures. The matrix and case are original GuideGxP material.

1. EURAMET cg-17, version 4.1, September 2022 — mechanical and electromechanical manometers; full PDF linked from the current catalogue.
2. PTB/DKD-R 6-1, edition 03/2014, revision 3 — chain, conditions and uncertainty; English translation published in 2021.
3. Cofrac LAB GTA 11, revision 03 — pressure metrology, applicable from 16 January 2023; excludes electrical simulation from its scope.
4. CEM ME-020, digital edition 2.2020 — model procedure for differential pressure gauge calibration.