The Yokogawa EJA110E specifies ambient-temperature effects per 28°C, including zero shifts as high as ±0.08% of span plus 0.18% of upper range limit [S1].
This specification is not a universal transmitter limit: temperature coefficients, capsule construction, calibration method and applied pressure all affect the residual error at zero and span [S1][S2][S3].
What a 28°C temperature-effect rating means
A 28°C change is an interval, not a temperature limit. The EJA110E table reports the output change over that interval for different capsule types, and its effects include separate contributions from span and upper range limit [S1].
For example, one listed EJA110E condition is ±(0.08% Span + 0.18% URL), while another is ±(0.07% Span + 0.02% URL) [S1]. URL means upper range limit, so the allowable change is not simply a percentage of the selected measurement span. A 2 kPa URL, for example, contributes only 0.0036% of span under the first expression, whereas a 500 kPa URL contributes 0.09% of span [S1].
Do not convert the 28°C rating directly into a universal per-degree coefficient. A datasheet expression may represent a compensated, nonlinear or worst-case envelope, and it should be applied as published. Temperature effects are normally described by TC zero and TC span, with zero-point error referring to the zero signal and span error referring to the full-scale value [S3].
Zero and span shifts behave differently
Zero error changes the lower reference point, while span error changes the distance between the lower and upper reference points [S3][S5]. A transmitter can therefore meet its zero requirement and still fail the span requirement, or the reverse.
WIKA gives 0.2% per 10 K as a typical average zero-point temperature coefficient for a pressure sensor, with the error increasing linearly as temperature departs from the reference point [S3]. If zero and span errors have the same sign, their contributions can add in the worst case [S3]. If their signs differ, the measurement error at an intermediate pressure may partly cancel, but that cancellation is not a valid basis for omitting either specification.
Factory calibration normally applies temperature-correction factors to reduce temperature-induced changes in the sensor signal, including offset shift and span effects [S2]. The pressure transmitter output still includes the residual terms in its published total error band, so compensation does not make the temperature effect disappear.
Comparing the main specification formats
Three formats appear in instrument documentation: a change per stated temperature interval, a coefficient per °C or per kelvin, and separate zero and span coefficients over an operating range [S3][S5][S8]. The interval format is convenient for direct checking, while coefficient format supports calculation at an actual operating temperature.
For a fixed 28°C excursion, a coefficient-based estimate is the coefficient magnitude multiplied by 28 and converted to the datasheet units. That calculation is only valid if the coefficient applies over the complete interval, uses compatible reference conditions, and has not already been included in a 28°C table. The EJA110E interval values should be used directly for that model because they already represent the manufacturer-defined 28°C effect [S1].
Where zero drift dominates the calculation
Honeywell identifies offset error as a major component of the residual pressure-sensor error after calibration and lists temperature variation, mounting stress, pneumatic-system connection stress, adhesives, aging and repeated environmental exposure among its causes [S2]. That makes a pressure sensor mounted directly to a hot pipe, vessel or pump housing a different accuracy case from a remotely mounted instrument with a shorter thermal path.
Mechanical installation can change the zero independently of the electrical calibration. Soldering to a circuit board, adhesive attachment and connection to the process piping can induce stress, while temperature variation can amplify an offset-sensitive sensor [S2]. Factory correction cannot remove every installation-specific change, and auto-zero at a known reference condition addresses offset-related errors but not an independent span coefficient [S2].
Absolute-pressure devices are less straightforward to auto-zero because an accurate implementation needs a vacuum source or an independent barometer; gage and differential-pressure devices are the primary focus of Honeywell’s technique [S2]. For an absolute pressure transmitter, zero behavior should therefore be checked against its specified reference rather than assumed identical to a vented gauge instrument.
Limits that change the real error budget
Ambient temperature is only one constraint. Differential-pressure transmitters can also acquire a static-pressure effect when both sides of the sensing element experience pressure, and this term must be included separately from the 28°C ambient-temperature effect [S8].
The EJA110E covers pressure spans from 0.5 to 5 kPa in the F range, 1 to 100 kPa in the M range, 5 to 500 kPa in the H range and 0.14 to 14 MPa in the V range [S1]. Its reference accuracy for spans above the minimum is ±0.055% of span for the listed F, M, H and V configurations [S1].
For a 5 kPa measurement span, the listed reference-accuracy value is ±0.00275% of span. The same absolute magnitude is ±0.05 kPa. This is the basis for judging whether a specified 28°C effect is acceptable for the measurement, not a separate claim that the complete installed accuracy equals that value.
For a differential-pressure transmitter, thermal performance also depends on capsule and wetted-parts construction. The EJA110E specification distinguishes F, L, M, H and V capsules, and its performance tables include different ambient-temperature effects for those configurations [S1]. A differential pressure transmitter should therefore be checked by exact model, capsule, span, URL and material configuration, not by product family alone.
Specifying and verifying the effect
Start with the manufacturer’s exact test condition for 28°C, then calculate the error at both endpoints. Compare the combined zero-and-span effect with the application’s allowable error, including reference accuracy, static-pressure effect, hysteresis, repeatability and installation stress [S1][S2][S8].
Use a pressure calibrator with uncertainty low enough to resolve the transmitter’s temperature-induced change. If the allowed 28°C effect is only a few hundredths of percent, a calibrator whose uncertainty is comparable with that number cannot provide a meaningful verification.
Verify zero at the transmitter’s true lower reference. A 0 psi display does not necessarily mean a vacuum reference for compound-ranged devices, because the calibrated lower point may be another value [S4]. Check span at the upper calibrated pressure, and do not use a source above the transmitter’s range merely to compensate visually for offset.
Zero and span adjustment can correct calibration errors, but adjustment does not remove a specified temperature coefficient [S6][S7]. For a transmitter exposed to sustained temperature change, evaluate automatic zero correction as an offset-control measure and retain the published TC span contribution in the uncertainty calculation [S2][S3].
Record the transmitter model, capsule, calibrated span, URL, wetted material code, reference temperature, temperature interval and correction method. The first traceable verification should reproduce the datasheet’s 28°C test interval and report zero, span and total effect separately, rather than reducing all three to one percentage.