Calibration is a comparison against a known reference. You apply pressures from a stable source, read the true pressure on a reference gauge and the transducer's output on a meter, and calculate the error at each point. If every point is within tolerance, you record the "as-found" data and stop. If not, you adjust zero and span (where the model allows it), repeat the test and record the "as-left" data.
What equipment do you need to calibrate a pressure transducer?
You need four things: a pressure source, a reference, a way to read the output and a way to power the device. The reference must be clearly better than the transducer you are checking.
| Item | Typical choice | What to check |
|---|---|---|
| Pressure source | Hand pump (pneumatic up to a few tens of bar, hydraulic for higher ranges) or a dead-weight tester | Covers 100% of the transducer range; fine adjustment (vernier) so each point can be set and held |
| Reference | Calibrated digital reference gauge or dead-weight tester | Valid calibration certificate; uncertainty ≤ ¼ of the transducer tolerance |
| Output meter | mA meter or process calibrator in series with the loop; voltmeter for 0–10 V outputs | Resolution of at least 0.001 mA for a 4–20 mA device |
| Power | 24 V DC loop supply (SEGMENsensor 4–20 mA transmitters accept 12–36 V DC) | Total loop resistance within limits: max. 600 Ω at 24 V |
| Fittings | Tee or manifold, adapters for G1/4, G1/2 or NPT threads | Leak-free at full pressure; same medium (gas or oil) as the range requires |
How good must the reference be? A widely used rule of thumb is a test uncertainty ratio (TUR) of at least 4:1. For a ±0.25% FS transmitter on 0–10 bar (±0.025 bar), the reference should be within about ±0.006 bar, and the mA meter within about ±0.01 mA (calculated values). For a ±0.1% FS high-accuracy version the reference must be 2.5 times better again. Reference instruments should themselves be calibrated by a laboratory working to ISO/IEC 17025 ↗, with traceability to a national standard such as those maintained by NIST pressure and vacuum calibrations ↗.
How do you calibrate a pressure transducer, step by step?
- Prepare. Note the range, output, accuracy class and reference type (gauge or absolute) from the label or ordering code. Let the transducer and reference reach room temperature and power the loop for at least several minutes before reading.
- Connect and leak-test. Mount the transducer and the reference on the tee in their normal orientation. Pressurise to 100%, close the source and watch for a falling reading; fix any leak before going on.
- Exercise. Cycle the pressure from 0 to 100% two or three times, then vent. This settles the diaphragm and seals, and is common practice before a performance test.
- Read zero. With a gauge transducer vented to atmosphere, record the output at 0. An absolute transducer needs a vacuum reference for its zero instead.
- Rise in steps. Set 25, 50, 75 and 100% of span on the reference, approaching each point from below without overshooting, wait for the reading to settle and record it.
- Fall in steps. Come back down through 75, 50, 25 and 0%, approaching each point from above. The difference between the rising and falling readings is the hysteresis.
- Evaluate as-found. Calculate the error at each point (next section). If all errors are within tolerance, record the data and finish.
- Adjust and repeat. If not, correct zero first, then span, using the adjustment the model provides. Repeat steps 4–7 and record the as-left data.
The general test methods for process instruments, including test points and up-and-down cycles under reference conditions, are described in IEC 61298-2 ↗.
How do you calculate the error and decide pass or fail?
Work in percent of span, because that is how a "% FS" accuracy is specified. For a 4–20 mA output the span is 16 mA:
Iideal = 4 mA + 16 mA × P ÷ PFS
Error (% of span) = (Imeasured − Iideal) ÷ 16 mA × 100
Pass if |error| ≤ the accuracy class (for example 0.25% FS) at every point.
Worked example: 0–10 bar, 4–20 mA, ±0.25% FS
The readings below are example values chosen to show the method, not test data from a specific unit. Errors are calculated values.
| Point | Reference | Ideal | As-found ↑ | Error ↑ | As-found ↓ | Error ↓ | As-left ↑ / ↓ | Error ↑ / ↓ |
|---|---|---|---|---|---|---|---|---|
| 0% | 0.0 bar | 4.000 mA | 4.048 | +0.30% | 4.050 | +0.31% | 3.998 / 4.000 | −0.01% / 0.00% |
| 25% | 2.5 bar | 8.000 mA | 8.050 | +0.31% | 8.054 | +0.34% | 8.000 / 8.004 | 0.00% / +0.03% |
| 50% | 5.0 bar | 12.000 mA | 12.052 | +0.33% | 12.058 | +0.36% | 12.002 / 12.008 | +0.01% / +0.05% |
| 75% | 7.5 bar | 16.000 mA | 16.051 | +0.32% | 16.054 | +0.34% | 16.001 / 16.004 | +0.01% / +0.03% |
| 100% | 10.0 bar | 20.000 mA | 20.050 | +0.31% | 20.050 | +0.31% | 20.000 / 20.000 | 0.00% / 0.00% |
As found, every point is about +0.05 mA (+0.31%) high, so the unit fails ±0.25% FS. Because the error is the same at 0% and 100%, this is a zero shift, not a span error. After a zero correction of −0.050 mA, the largest as-left error is +0.05% of span, well inside tolerance. The hysteresis at 50% is 12.008 − 12.002 = 0.006 mA, or 0.04% of span, both before and after the adjustment. You can check any single point with the 4–20 mA calculator, and convert psi or kPa readings with the pressure unit converter.
How do you tell zero shift, span error and hysteresis apart?
The pattern of the errors tells you what to adjust, and whether adjusting will help at all.
| Error pattern | Likely cause | What to do |
|---|---|---|
| Same offset at all points | Zero shift (mounting stress, temperature, ageing) | Adjust zero only |
| Error grows with pressure, ≈ 0 at zero | Span (gain) error | Adjust span at 100%, then re-check zero |
| Largest error at mid-scale, small at both ends | Non-linearity | Cannot be trimmed with zero/span; compare with the accuracy class |
| Rising and falling readings differ | Hysteresis, or a leak or friction in the test rig | Re-check the rig for leaks; if it persists and exceeds tolerance, replace the unit |
| Reading drifts while pressure is held | Leak, temperature change or unsettled reference | Leak-test and wait for thermal stability before recording |
| Output stuck at or below 4 mA or above 20 mA | Wiring, supply or overpressure damage | Check loop voltage and polarity; see the 4–20 mA wiring guide |
A transducer with a millivolt output (no built-in amplifier) is checked the same way, but the ideal output is sensitivity × excitation voltage × P ÷ PFS, so hold the excitation constant and record it. The transducer vs transmitter guide explains the output types.
How often should a pressure transducer be calibrated?
There is no single fixed interval: it is set by your quality system, the accuracy you need and the history of as-found results. Many plants start with 12 months and lengthen the interval when as-found data stay well inside tolerance, or shorten it after a failure, overpressure event or repair. For reference, the long-term stability of SEGMENsensor 4–20 mA pressure transmitters is ±0.1% FS per year, as listed in the 4–20 mA pressure transmitter specifications. Keep in mind the difference between % FS and % of reading when comparing devices, explained in accuracy: % FS vs % of reading.
Calibration of SEGMENsensor pressure transmitters
| Item | Value |
|---|---|
| Factory calibration | Calibration report at 5 points: 0%, 25%, 50%, 75%, 100% FS |
| Range | 0–0.1 bar to 0–400 bar, specified at order |
| Accuracy | ±0.25% FS standard; ±0.1% FS high-accuracy version |
| Long-term stability | ±0.1% FS per year |
| Output and supply | 4–20 mA (2-wire, 12–36 V DC, max. 600 Ω at 24 V); 0–10 V or RS485 options |
| Field zero / span adjustment | Depends on model and connector; confirmed with the order |
| Samples | Qualification samples 1–5 pcs in 5–7 working days; MOQ and production lead time confirmed with the quotation |
Ranges, outputs and process connections are listed on the pressure transmitter product page.
Need a transmitter with a range and accuracy that fit your test plan?
Send the medium, pressure range, output, process connection and accuracy you need. Our engineers will propose a model. Qualification samples 1–5 pcs in 5–7 working days · FOB China · 100% T/T in advance · reply within 24 hours.
Frequently Asked Questions
No. Simulating a 4–20 mA signal only checks the PLC input and wiring, not the sensor. To calibrate the transducer itself you must apply real, known pressures and compare the output with a reference gauge.
Calibration is the comparison with a reference and the recording of the errors. Adjustment is changing zero or span to reduce those errors. A unit that is within tolerance as found is calibrated but not adjusted.
Five points at 0, 25, 50, 75 and 100% of span, read rising and falling, is the common choice. It shows zero, span, non-linearity and hysteresis. A quick field check may use 0, 50 and 100% only.
A common rule of thumb is at least 4 times better than the transducer. For a ±0.25% FS device on 0–10 bar, that means a reference uncertainty of about ±0.006 bar or better (calculated value).
Yes. Each transmitter is factory-calibrated and shipped with a calibration report at 5 points (0, 25, 50, 75 and 100% FS). Accuracy is ±0.25% FS as standard or ±0.1% FS in the high-accuracy version.