An accelerometer that has drifted 4% in sensitivity will still produce a clean, plausible vibration spectrum. Nothing on the analyzer flags it. The bearing that was trending toward failure just looks a little quieter than it is, or the shaker test over-drives a product because the control accelerometer reads low. Accelerometer calibration is what keeps that from happening — it determines the sensor’s true sensitivity relative to a traceable reference and indicates how much that figure can be trusted.

This guide covers how accelerometers are calibrated under governing standards, what a calibration measures, the mounting and setup errors that quietly undermine results, and what makes a certificate defensible in an audit.

Accelerometer Calibration: A Complete, Standards-Based Guide

What an Accelerometer Measures — and Why It Drifts

An accelerometer converts mechanical acceleration into an electrical signal. Most measurement-grade units are piezoelectric: a mass presses on a crystal under acceleration, generating a charge proportional to the force. That charge is read out either as a charge signal (pC/g) or, in an IEPE accelerometer, converted to a low-impedance voltage (mV/g) by built-in electronics. MEMS and piezoresistive types work differently and can measure down to steady (DC) acceleration, which piezoelectric sensors cannot.

Whatever the technology, an accelerometer’s sensitivity shifts over time. Temperature cycling, mechanical shock, cable wear, and simple aging all move it. Because the sensor reports a number regardless of whether it is correct, only periodic accelerometer calibration reveals the drift.

Calibration Is Not Adjustment

This distinction matters more for accelerometers than almost any other instrument, because a piezoelectric accelerometer has nothing to physically adjust.

• Calibration determines the accelerometer’s actual sensitivity against a reference, with an associated uncertainty. It produces a number, not a modified sensor.
• Adjustment, for an accelerometer, means updating the sensitivity value the measurement system uses — writing it onto the certificate, into the analyzer, or into the sensor’s TEDS chip. The crystal itself is untouched.

So the job is to measure sensitivity accurately and record it, keeping the as-found value on file. Overwrite the stored sensitivity without recording what it was, and nobody can answer the question every auditor asks: were the vibration measurements this accelerometer produced last quarter valid?

The Two Ways an Accelerometer Is Calibrated

All traceable methods live in the ISO 16063 series, “Methods for the calibration of vibration and shock transducers.” They divide into two families.

The comparison method (ISO 16063-21) covers roughly 0.4 Hz to 10 kHz and is how the vast majority of accelerometers get calibrated. A modern back-to-back comparison system can reach an uncertainty near 0.6–1%, provided its reference accelerometer was itself calibrated by a laser primary method — which typically contributes about 0.3% of that budget. That chain, reference-to-primary-to-national-standard, is what traceability means here.

What a Calibration Actually Measures

“Calibrating an accelerometer” usually means measuring its sensitivity — the electrical output per unit of acceleration, in mV/g for IEPE sensors or pC/g for charge-mode. But sensitivity alone is only part of the picture:

• Reference sensitivity is quoted at a single reference frequency — conventionally 159.2 Hz (1000 rad/s) or 100 Hz — chosen because it lies in the flat part of the response for most sensors. Common nominal values are 100 mV/g and 10 mV/g.
• Frequency response sweeps the shaker across the working range and records how sensitivity deviates from the reference value. A sensor that is flat at 159 Hz but rising at 2 kHz will over-report high-frequency content, so broadband work needs the full curve, not a single point.
• Transverse (cross-axis) sensitivity is the unwanted response to acceleration at 90° to the measurement axis. Ideally, zero; it can reach 5% with a normal accelerometer; ISO 16063-31 covers its measurement.
• Amplitude linearity and phase matter for high-level and multi-channel work, where channels must stay in phase alignment.

Mounting: The Error That Hides in Plain Sight

More accelerometer calibrations are ruined by poor mounting than by any electronic cause. The sensor only reports the motion of the surface it is bonded to, so the coupling has to be rigid across the whole frequency range.

• Stud mounting into a clean, flat, correctly torqued surface gives the widest usable bandwidth and is the reference method.
• Adhesive and magnetic mounts are convenient but roll off the high-frequency response; the softer the coupling, the lower the frequency at which readings go wrong.
• Surface flatness, thread engagement, torque, and a thin film of coupling fluid all shift the mounted resonance, which for a typical accelerometer sits above 20 kHz.

A calibration performed with a mount different from the one used in service can be accurate on the bench and misleading in the field.

Sensor Types and TEDS

The calibration approach shifts with the sensor:

• IEPE (voltage-output) accelerometers are the default for vibration testing. Many carry a TEDS chip (IEEE 1451.4) that stores the calibrated sensitivity, date, and serial number on the sensor itself, so the controller reads the correct value directly and transcription errors disappear.
• Charge-mode accelerometers suit high-temperature and high-frequency work but need a charge amplifier in the measurement chain, which forms part of what is calibrated.
• MEMS and piezoresistive accelerometers respond down to DC. These are the sensors where zero-offset and bias genuinely matter, and they can be calibrated statically.

Low-Frequency and MEMS: The Gravity Method

Because MEMS and piezoresistive sensors respond to steady acceleration, they can be calibrated against the most traceable reference there is: gravity. Rotating the accelerometer through known orientations swings its input between +1 g and −1 g, giving a static two-point (or multi-position) calibration of sensitivity and zero offset. A rotating table extends this into the low-frequency range, typically 0–8 Hz, where shakers struggle. Piezoelectric accelerometers cannot be calibrated this way, since they do not respond to constant acceleration.

Compliance and Certificate Content

For accredited laboratories, ISO/IEC 17025 governs competence, traceability, and reporting, with NABL operating the accreditation scheme in India. The reference accelerometer’s traceability and its uncertainty should give an adequate test uncertainty ratio for the tolerance being verified. A defensible accelerometer calibration certificate should carry:

• Sensor identification, type (IEPE, charge, MEMS), and serial number
• The reference standard used and its traceability
• Method (primary or comparison, with the ISO 16063 part)
• Reference sensitivity, with the reference frequency and amplitude
• Frequency response data across the calibrated range, not just one point
• Transverse sensitivity where specified
• Expanded measurement uncertainty and coverage factor
• Ambient conditions, calibration date, and due date where requested

Best Practices Worth Enforcing

A few habits separate a controlled accelerometer calibration program from a reactive one:

• Calibrate across the frequency range the sensor is actually used in, not just at the reference point.
• Mount the sensor as it is in service, or document the difference.
• Record the as-found sensitivity before writing any new value to TEDS or the analyzer.
• Keep the reference accelerometer’s own calibration current; the whole chain depends on it.
• Set intervals from drift history and measurement risk; annual is a common starting point, tightened for critical or heavily shocked sensors.
• Assess the impact of any out-of-tolerance result on measurements taken since the previous calibration.

Digital Transformation, AI, and IoT

Manual accelerometer calibration breaks in two familiar places: the uncertainty arithmetic and the records. Modern platforms fix both. Data flows straight from the calibration system, sensitivity and frequency-response results are captured automatically, and the uncertainty budget is applied the same way every time. IoT-connected condition-monitoring sensors increasingly stream continuous vibration data, blurring the line between periodic calibration and live verification, while drift analytics support risk-based intervals — flagging a sensor trending toward its limit before it starts hiding a fault.

How Zeptac Helps in Accelerometer Calibration

Accelerometer and vibration calibration sits within the mechanical calibration scope of Zeptac’s CalTac calibration management software. CalTac automates Type A and Type B measurement uncertainty calculations, generates ISO/IEC 17025- and NABL-ready certificates instantly, and maintains complete instrument and reference-standard records — including the calibration status of the reference accelerometer the whole chain depends on. Internal audit tools mapped to ISO 17025 clauses and NABL quality-document control keep the laboratory inspection-ready, and InstaCertify turns finished measurements into issued certificates immediately.

In practice:

• A calibration laboratory issues accelerometer and vibration-sensor certificates with automated uncertainty budgets, cutting turnaround from hours to minutes.
• A manufacturer tracks the calibration status of every reference accelerometer and control sensor across test cells and blocks use of anything overdue.
• A quality manager in a testing laboratory pulls a sensor’s complete as-found/as-left sensitivity history during an audit in seconds rather than hours.

Where This Is Heading

Expect condition-monitoring IoT sensors to provide continuous between-calibration confidence, cloud records that assessors review remotely, and TEDS-driven workflows that eliminate manual sensitivity entry. Growth in EV, aerospace, and structural health monitoring is also driving demand for higher-frequency, low-uncertainty accelerometer calibration.

Conclusion

Sound accelerometer calibration comes down to four disciplines: use the right method for the sensor — comparison under ISO 16063-21 for most, primary for references — measure sensitivity and frequency response rather than a single point, respect mounting and the reference chain, and record a complete uncertainty budget with the as-found value. Software that automates the arithmetic and the documentation turns an error-prone manual process into one that holds up under any audit.

Frequently Asked Questions for Accelerometer Calibration

Q1. What is accelerometer calibration?

Answer:  It is the process of determining an accelerometer’s true sensitivity — its electrical output per unit of acceleration — against a traceable reference, together with the uncertainty of that result. It is governed by the ISO 16063 series.

Q2. What standards govern accelerometer calibration?

Answer: The ISO 16063 series: part 11 (primary calibration by laser interferometry), part 12 (reciprocity), part 21 (comparison to a reference transducer), and part 31 (transverse sensitivity), all under the laboratory requirements of ISO/IEC 17025.

Q3. What is the difference between primary and comparison calibration?

Answer: Primary calibration measures acceleration in absolute terms, usually by laser interferometry, with no reference sensor. Comparison (back-to-back) calibration compares the accelerometer against an already-calibrated reference accelerometer on a shaker, and is the common method for most labs.

Q4. What is the reference frequency in accelerometer calibration?

Answer: Sensitivity is quoted at a single reference frequency, conventionally 159.2 Hz (1000 rad/s) or 100 Hz, because that region is flat for most sensors. A full calibration also sweeps the working frequency range.

Q5. Can I calibrate an accelerometer without a reference sensor?

Answer: For AC-response piezoelectric accelerometers, no — you need either a reference accelerometer or a primary system. MEMS and piezoresistive sensors that respond to DC can be calibrated statically against gravity, tilting between +1 g and −1 g.

Q6. How often should an accelerometer be calibrated?

Answer: There is no universal interval. Base it on drift history, usage, and measurement risk. Annual calibration is a common starting point, shortened for critical, heavily shocked, or reference sensors.

Q7. What is transverse sensitivity?

Answer: The unwanted response of an accelerometer to acceleration at 90° to its measurement axis. Ideally zero, it can reach around 5% in a normal sensor; keeping it low is important for accurate multi-axis measurement.