Calibration

What is calibration?

Calibration is the comparison of a measuring instrument’s indication with a reference standard of known measurement uncertainty, under specified conditions, in order to determine and document the instrument’s error. In the process industry it mostly concerns transmitters for pressure, temperature, level and flow that send their measurement as a 4-20 mA signal to a DCS or PLC. Without calibration nobody knows whether 12 mA really means 5 bar, and control loops, quality data and safety functions end up resting on assumptions.

🧠 What is the difference between calibration, adjustment and verification?

The terms are often used interchangeably on site, but the International Vocabulary of Metrology (VIM, JCGM 200:2012) draws a sharp line between them. Calibration measures the error; it does not change the instrument. The VIM explicitly warns against confusing calibration with adjustment, which is often mistakenly called “self-calibration”.

Legal metrology applies to instruments used in trade, such as weighing scales, fuel dispensers and energy meters; in the EU it is shaped by the Measuring Instruments Directive (MID), and in the Netherlands by the Metrology Act (Metrologiewet) of 2006, in force since 2007. A pressure transmitter in a plant is therefore calibrated, not legally verified. If you adjust an instrument after a failed calibration, you calibrate it again to record its new condition.

🔗 How does traceability to the SI work?

A calibration is only worth something if the reference itself is demonstrably correct. That property is called metrological traceability: a documented, unbroken chain of calibrations, each of which contributes to the measurement uncertainty.

1. SI units — since the redefinition of 20 May 2019, based on fixed constants of nature
2. National metrology institute — VSL in Delft for the Netherlands, NPL in the UK, PTB in Germany; they maintain the national measurement standards
3. Accredited calibration laboratory — assessed against ISO/IEC 17025:2017 by the national accreditation body, such as the Dutch RvA or UKAS
4. Working standard — the process calibrator or deadweight tester in your own instrument workshop
5. Process instrument — the transmitter in the field

A common rule for the working standard is a test uncertainty ratio (TUR) of at least 4:1: the reference should be at least four times more accurate than the tolerance of the instrument under test. ANSI/NCSL Z540.3 accepts a 4:1 TUR as an alternative to calculating a false-accept risk of at most 2%.

🔧 How do you perform a 5-point transmitter calibration?

The usual method for a 4-20 mA transmitter is a 5-point calibration at 0, 25, 50, 75 and 100% of range, corresponding to 4, 8, 12, 16 and 20 mA. Ideally you test both upscale and downscale so that hysteresis becomes visible.

1. Prepare — work order from the CMMS, SIS bypass or agreement with the operator, loop released
2. Record as-found — take the five points before you change anything; this is the condition the instrument was operating in
3. Assess — calculate the error per point as a percentage of span and compare it with the tolerance
4. Adjust — if it fails, perform a sensor trim (input side) and, if needed, an analogue output trim (mA output)
5. Record as-left — take the five points again and record them
6. Report — certificate with reference, uncertainty, as-found and as-left back into the CMMS

Worked example

A pressure transmitter has a range of 0–10 bar, so a span of 16 mA. The tolerance is ±0.5% of span, or ±0.08 mA. The error per point is (measured − expected) / 16 mA × 100%.

The error increases with pressure: a small zero offset plus, mainly, a span error. Three points are out of tolerance, so the transmitter is adjusted. Crucially, the as-found failure must also be reported to the process owner, because every reading since the previous calibration may have been up to almost 0.9% too high. To achieve a 4:1 TUR, the calibrator in this example needs an uncertainty of no more than ±0.02 mA.

📅 How do you determine the calibration interval?

There is no statutory interval for process instruments; you have to justify it yourself. The joint guideline ILAC-G24 / OIML D 10 describes methods such as the staircase method (extend the interval after a pass, shorten it after a fail) and the control chart method (plot drift across successive calibrations). Inputs include the criticality of the measurement, the manufacturer’s stability specification, process conditions such as vibration and temperature cycling, and above all the as-found history. In practice intervals range from three months for critical GMP measurements to four years for stable, non-critical transmitters.

🏭 What applies in pharma and for safety functions?

• GMP and the FDA — in pharmaceutical manufacturing every instrument that affects product quality must be calibrated against an approved procedure. Electronic calibration records fall under 21 CFR Part 11 (1997): audit trails, electronic signatures and access control. The data must meet ALCOA+: attributable, legible, contemporaneous, original and accurate, plus complete, consistent, enduring and available.
• IEC 61511 and proof testing — transmitters in a SIS are proof tested periodically to reveal hidden dangerous failures. A calibration covers part of that test, and the test equipment itself must be calibrated traceably. As-found results feed the reliability data that underpins the claimed SIL.

🛠️ Which tools and software are used?

The instrument list and tag numbers come from the P&ID; the calibration history belongs with asset management.

🔐 What does calibration have to do with OT security?

Calibration is about the integrity of measurements, the I in the CIA Triad. Through HART, a transmitter’s range, damping, trim and engineering units can be changed remotely: from a handheld, from an asset management system, or over the network via HART multiplexers and HART-IP. A changed range produces a signal that looks plausible but is wrong, misleading the control loop, the quality data in the historian or a safety function. TRITON showed in 2017 that attackers are prepared to target safety systems.

Measures:

• Write protection — the hardware switch or jumper on the transmitter blocks trims and reconfiguration
• Configuration management — periodically compare the actual instrument configuration with the approved baseline
• Change management — every trim or rerange goes through a work order, with as-found/as-left as evidence
• Access to HART tools — restrict who may use communicators and asset management software, and log their actions

❓ Frequently asked questions

How often should a transmitter be calibrated?

There is no fixed calibration interval; you set it based on criticality, the manufacturer’s specification and the as-found history. A typical calibration interval is three to twelve months for critical GMP measurements and one to four years for stable process transmitters. The methods in ILAC-G24 / OIML D 10 help you base the calibration interval on data.

What do as-found and as-left mean in calibration?

As-found is the reading of the instrument as it was found in service, before any adjustment. As-left is the reading after any adjustment, the condition in which the instrument returns to service. Recording both is essential in calibration, because the as-found result shows whether earlier process data can be trusted.

Is calibration the same as legal verification?

No, calibration determines an instrument’s error against a traceable reference and has no legal status in itself. Legal verification is a statutory check under metrology law for instruments used in trade, such as scales and fuel dispensers. Technically the two are similar, but legally they are not.

How do you calculate the error of a 4-20 mA transmitter?

You calculate the error of a 4-20 mA transmitter as (measured mA − expected mA) divided by the 16 mA span, multiplied by 100%. For example, a transmitter reading 12.10 mA where 12.00 mA is expected has an error of +0.63% of span. You then compare that value with the tolerance, often between ±0.25% and ±1%.

Is reranging a transmitter via HART the same as calibration?

No, reranging only changes which process value corresponds to 4 and 20 mA; it does not correct a measurement error. Calibration requires a comparison with a physical reference, optionally followed by a sensor trim or output trim. After a rerange you should therefore check the transmitter against a reference again and record the new range.

Who can issue calibration certificates?

Anyone can perform a calibration, but only a laboratory accredited to ISO/IEC 17025 can issue an accredited calibration certificate. National accreditation bodies such as the Dutch RvA or UKAS grant that accreditation. In-house instrument workshops usually calibrate themselves, using working standards that are calibrated by an accredited laboratory.

📌 In summary

Calibration establishes how far a process instrument deviates from a traceable reference, making it the foundation of reliable control, product quality and safety functions. Always record as-found and as-left, justify the interval with data, and protect the configuration of field instruments as carefully as that of a PLC.