Accelerometer

What is an Accelerometer

An accelerometer is an electromechanical sensor that converts mechanical acceleration into a proportional electrical signal. The acceleration it measures spans three regimes: vibration (the oscillating acceleration of a structure), shock (a transient peak), and the static gravity vector (a constant 1 g pull used for tilt and orientation sensing). In an industrial test-and-measurement context the accelerometer sits under Vibration & Condition Monitoring, where it is the front-end transducer for measuring casing and bearing-housing vibration on rotating machinery — motors, pumps, fans, gearboxes, compressors, and turbines. It is a single product type, not a category; the broader category contains sibling instruments such as vibration sensors, vibration meters, tachometers, and condition-monitoring systems that the accelerometer feeds.

The signal an accelerometer produces is most commonly an acceleration value expressed in g, where 1 g equals 9.80665 m/s², or directly in m/s². For condition monitoring, raw acceleration is frequently integrated once to produce velocity in mm/s, because the international standards for evaluating machine vibration are written in velocity terms. This integration step is fundamental to how the sensor is used: the accelerometer measures acceleration with the best high-frequency fidelity of any vibration transducer, then downstream electronics derive velocity (and, with a second integration, displacement) as needed for the relevant evaluation zone.

The reason the accelerometer dominates rotating-machine monitoring is physical. Acceleration is the derivative of velocity, so for a given displacement amplitude the acceleration signal grows with the square of frequency. This makes accelerometers exceptionally sensitive to the high-frequency content — bearing defect frequencies, gear-mesh harmonics, blade-pass — where the earliest signs of mechanical degradation appear. A displacement probe is better for low-frequency shaft motion, and a velocity transducer occupies the middle band, but the accelerometer plus integration covers the widest practical band from a single, rugged, externally mounted device.

Photo: Jianming Ding et al., CC BY 4.0, via Wikimedia Commons

Four working facts frame everything that follows. First, the workhorse principle is piezoelectric — a seismic mass strains a crystal, generating charge proportional to acceleration. Second, the workhorse output is IEPE/ICP, a 2-wire low-impedance voltage signal that an estimated 90 percent or more of industrial applications use. Third, piezoelectric devices are AC-coupled and cannot read steady-state 0 Hz acceleration, which is why MEMS variants exist for DC and tilt work. Fourth, the usable frequency band is bounded at the low end by AC coupling and at the high end by the mounted resonance, so mounting method directly determines measurement bandwidth. The remaining chapters expand each of these into a selection-ready framework.

Transduction Types

From the operating principle that converts motion into a signal, three dominant transduction types divide the accelerometer market: piezoelectric, capacitive MEMS, and piezoresistive MEMS. Each has a distinct working mechanism, a distinct ability or inability to read DC, and a distinct application home. Choosing the wrong transduction type is the most consequential early mistake, because it determines whether you can even measure the quantity of interest — for example, no amount of tuning will let an AC-coupled piezoelectric unit read a static tilt angle. The table below summarizes the three.

Piezoelectric is the workhorse of industrial vibration monitoring. A seismic (proof) mass presses on a piezoelectric crystal; acceleration creates a force (F = m·a) that strains the crystal and generates an electric charge proportional to the acceleration. Because the charge leaks away under a constant force, the device is AC-coupled and cannot measure steady-state (0 Hz / DC) acceleration. Piezoelectric units come in two sub-modes that differ in where the charge-to-voltage conversion happens, covered in detail below: IEPE/ICP voltage mode and charge mode.

The first sub-mode, IEPE / ICP (voltage mode), places a built-in microelectronic charge amplifier inside the sensor that converts the high-impedance charge into a low-impedance voltage output over a 2-wire line. It is powered by a 2-20 mA constant-current source on an 18-30 VDC supply. IEPE dominates general-purpose testing and condition monitoring — an estimated 90 percent or more of industrial accelerometer applications. The terms "IEPE" and "ICP" are equivalent; ICP is PCB Piezotronics' trademark for the same architecture.

The second sub-mode, charge mode (charge output), has no internal electronics and instead outputs raw charge (pC/g) to an external charge amplifier. It is required above roughly 150 °C — up to 250 °C and beyond, with charge-mode units rated to about 482 °C — precisely because there are no internal semiconductors to fail from heat. This is the choice for gas-turbine and engine high-temperature monitoring.

Capacitive MEMS takes a different path. A micromachined silicon proof mass moves between fixed electrodes; the displacement changes capacitance, which on-chip circuitry converts to a voltage. The decisive advantage is that it is DC-responsive — it can measure 0 Hz static tilt and very low frequencies — and it is low-power, low-cost, and small. It is used in mobile and automotive devices and increasingly in low-frequency and wireless condition-monitoring sensors, but it generally offers narrower bandwidth and a higher noise floor than piezoelectric at high frequency, which limits it for high-frequency bearing diagnostics.

Piezoresistive (MEMS) mounts a proof mass on flexible beams that carry piezoresistive (strain-gauge) elements; beam deflection changes their resistance, read out through a Wheatstone bridge. Like capacitive MEMS it is DC-responsive, and it is particularly strong for high-g shock testing — crash, ballistics, and drop testing — and for low-amplitude and gravity measurements where a true DC reference is essential.

Sensing Technologies and Crystals

Within the dominant piezoelectric family, two further engineering choices shape performance: the piezoelectric crystal material and the internal mechanical design that couples the seismic mass to that crystal. These are not marketing distinctions — they directly set charge sensitivity, thermal stability, and immunity to the false signals that base strain and thermal transients inject into a poorly designed sensor. A purchasing engineer who understands these two axes can read why one 100 mV/g unit costs more than another with the same nominal sensitivity.

Crystal materials fall into two camps. Natural quartz is very stable and has low charge sensitivity, which makes it a natural fit for IEPE voltage-mode units where an internal amplifier supplies the gain and long-term stability is paramount. PZT (lead-zirconate-titanate) ceramic offers much higher charge sensitivity, so it is preferred for charge-mode units and for sensitive, high-output devices where raw charge per g must be large. The trade-off is the classic one: quartz favors stability, PZT favors output level, and the right pick depends on whether the application is long-term monitoring or high-sensitivity measurement.

Internal mechanical design describes how the seismic mass loads the crystal. Three common configurations exist: shear, compression, and flexural (bending) modes. In a shear-mode design the crystal is loaded in shear by a mass clamped around a central post; in a compression-mode design the mass presses axially onto the crystal; in a flexural-mode design a beam bends the crystal element. For industrial units, shear mode is favored because it has low sensitivity to base strain and to thermal transients — two error sources that plague compression designs mounted on machines that flex and change temperature.

The reason this matters in the field is reliability of the reading, not just the sensor's survival. A compression-mode sensor bolted to a hot, flexing motor frame can report acceleration that is partly an artifact of the frame straining the sensor base, or of a sudden temperature step. A shear-mode industrial accelerometer largely rejects both, so the velocity value derived for ISO 20816 zone classification reflects real machine vibration rather than mounting and thermal noise. This is why nearly all general-purpose industrial IEPE accelerometers are built as shear-mode ceramic or quartz devices.

Pulling the technology choices together: an IEPE shear-mode quartz or ceramic accelerometer is the default industrial transducer, a PZT charge-mode shear unit handles high temperature, and MEMS capacitive or piezoresistive devices handle the DC and very-low-frequency or high-g cases that piezoelectric physics cannot serve. The remaining chapters turn these principles into mounting practice, the numbers on the spec sheet, and a selection sequence.

Materials, Media, and Mounting

Unlike a process sensor that contacts a corrosive fluid, an accelerometer's "media" interaction is mechanical and environmental: it bolts to a vibrating surface and must survive the surrounding temperature, moisture, and wash-down conditions while faithfully transmitting motion. Two material and mechanical decisions dominate: the housing material and seal, and the mounting method. The mounting method in particular is not an installation afterthought — it is a measurement parameter, because every gram of added mass and every reduction in coupling stiffness lowers the mounted resonance and therefore the usable high-frequency band.

Housing and sealing. Industrial accelerometers are built in stainless steel or titanium and are hermetically sealed for industrial and wash-down duty. Sealing is rated by ingress protection — for example IP67 or IP68 — or specified as hermetic, and the connector style (a 2-pin MIL connector, an M8 connector, or an integral cable) is chosen to match the environment and routing. The housing protects the crystal and any internal IEPE electronics from moisture, dust, and the electromagnetic environment of a motor or drive.

Mounting directly sets the usable frequency band, and the options form a clear hierarchy from highest to lowest fidelity. The table below ranks them.

Stud mount gives the stiffest coupling and the highest resonant frequency, so it delivers the best high-frequency accuracy and is the reference condition under which spec sheets quote resonance and frequency response. It is the correct choice for permanent fixed monitoring where the sensor stays put for years. Adhesive mounting with cyanoacrylate, epoxy, or adhesive pads lowers the resonance slightly; cyanoacrylate gives the best high-frequency performance of the adhesives, but it needs a clean surface and a thin film to preserve transmissibility.

A magnetic base is quick and portable, which makes it ideal for route-based data collection, but the added mass and softer coupling lower the resonance and limit the response to roughly a few kHz — adequate for many overall-vibration trends but not for the highest bearing-defect frequencies. A probe or handheld tip has the lowest fidelity and should be reserved for spot checks. Across all methods, the surface must be clean and flat, the stud properly torqued, and a thin couplant film applied to improve transmissibility. As a working rule, take the usable upper frequency at about one-third to one-fifth of the mounted resonant frequency.

Key Specification Parameters

Reading an accelerometer spec sheet is a core skill for purchasing engineers. Manufacturers list a dozen or more parameters, but a handful drive the selection decision: sensitivity, measurement range, frequency response, mounted resonant frequency, transverse sensitivity, temperature range, and the IEPE power requirement. The table below cross-references typical industrial values and real example models so you can calibrate what "normal" looks like before comparing quotes.

Sensitivity is the output per unit of acceleration. The industrial de-facto standard is 100 mV/g for general machinery, with 10 mV/g for high-g and wide-range work, 50 mV/g and 250 mV/g for intermediate needs, and 500 mV/g for low-amplitude, low-frequency monitoring. Charge mode is rated in pC/g rather than mV/g. Sensitivity tolerance is typically ±5% to ±10% at a 25 °C reference. The key engineering relationship is that sensitivity and measurement range trade off inversely: a 500 mV/g unit gives better low-amplitude resolution but a smaller usable g-range, while a 10 mV/g unit gives a wider g-range with coarser resolution.

Measurement (amplitude) range for industrial general-purpose units runs ±50 g to ±80 g — for example ±50 g on a 100 mV/g unit. High-vibration and impact units extend to ±500 g, and dedicated shock units reach thousands of g. Because of the inverse trade-off above, the range you can buy is tied to the sensitivity you choose, so the two parameters must be selected together rather than independently.

Frequency response defines the usable band, bounded at the low end by the AC coupling and at the high end by the mounted resonance. A typical IEPE industrial specification reads ±5% from about 3 Hz to 5 kHz, ±10% from about 1 Hz to 9 kHz, and ±3 dB from about 0.5 Hz to 14 kHz (Wilcoxon 786A); broad general-purpose units quote 0.5 Hz to 10 kHz (PCB 603C01). The low-frequency limit is set by the internal AC coupling and the discharge time constant, which is why a piezoelectric unit cannot reach 0 Hz.

Mounted resonant frequency for a stud-mounted unit is typically about 25-35 kHz — 30 kHz on the 786A, for instance — and the usable upper limit is generally taken at about one-third to one-fifth of this value. Transverse (cross-axis) sensitivity is the response to acceleration perpendicular to the main axis, and quality units specify it at ≤ 5%. Temperature range for standard IEPE is −50 °C to +120 °C, high-temperature IEPE reaches +175 °C, and charge-mode units go well beyond, from 250 °C to about 482 °C.

Power (IEPE) is supplied as 18-30 VDC with a 2-20 mA constant current, and the sensor's bias or standoff voltage is typically 8-14 VDC. Output axes are single-axis (uniaxial) or tri-axial (X/Y/Z in one housing). Beyond these headline numbers, precision work also examines dynamic range and broadband resolution (the noise floor, in µg or g rms), settling time, base-strain sensitivity, electrical isolation, sealing (such as IP67/IP68 or hermetic), and connector type (2-pin MIL, M8, or integral cable).

Selection Decision Factors

To turn the preceding five chapters into a specific model, follow the decision sequence below. As with most instrument selection, the costly mistakes come not from a single wrong number but from deciding parameters in the wrong order — for example fixing a sensitivity before confirming the frequency band of interest. These nine steps work as a fixed RFQ template for accelerometers.

1. Sensitivity vs. range: Use 100 mV/g for general machinery, 500 mV/g for low-speed, low-frequency, low-amplitude monitoring, and 10 mV/g for high-g or high-vibration duty. Pick the sensitivity that keeps the expected signal within the unit's measurement range without clipping.
2. Frequency band of interest: Match the flat (±5%) band to the fault frequencies you must detect, from 1× running speed up to bearing and gear-mesh harmonics. High-frequency bearing defects demand stud mounting and a higher resonant frequency.
3. Temperature: Choose IEPE for service up to about 120-175 °C; switch to charge mode plus an external charge amplifier above roughly 150 °C, where charge-mode units run from 250 °C to about 482 °C.
4. DC vs. AC measurement: If you need static, very-low-frequency, or tilt measurement, use MEMS (capacitive or piezoresistive). For general machine vibration, use piezoelectric IEPE.
5. Number of axes: Choose tri-axial where you need multi-direction data from a single mounting point; otherwise a single-axis unit is simpler and cheaper.
6. Environment: Specify sealing (IP rating or hermetic), wash-down capability, and EMI tolerance; for a hazardous area, require an intrinsically-safe ATEX or IECEx-rated unit (or the equivalent North American Class/Division/Zone rating).
7. Mounting access and permanence: Use a permanent stud for fixed monitoring and a magnetic or adhesive mount for route-based or portable work, remembering that each non-stud method lowers the usable high-frequency limit.
8. Cross-axis, base-strain, and noise floor: For precision or low-amplitude work, scrutinize transverse sensitivity (≤ 5%), base-strain sensitivity, and the broadband resolution (noise floor).
9. Cable, connector, and signal conditioning: Confirm compatibility between the output and the signal chain — an IEPE current source for voltage-mode units, or an external charge amplifier for charge-mode units — and choose the connector (2-pin MIL, M8, or integral cable) to suit routing.

One dimension that buyers often overlook is calibration and certification provenance. Quality industrial units ship with ISO/IEC 17025 accredited single-point or swept calibration, and machine-vibration evaluation should reference the ISO 20816 series (with Part 3 covering bearing-housing measurement and A/B/C/D zones), while transducer calibration traces to the ISO 16063 series. On the supply side, established makers include PCB Piezotronics / IMI Sensors and Wilcoxon Sensing Technologies (both Amphenol), Brüel & Kjær (HBK), Kistler, Endevco (a PCB brand since 2019), Dytran (by HBK), Hansford Sensors, Metrix Instrument, CTC (Connection Technology Center), and the condition-monitoring lines from SKF, Baker Hughes (Bently Nevada), and Emerson; MEMS component makers include Analog Devices, Bosch Sensortec, TE Connectivity, STMicroelectronics, NXP, and Honeywell. Representative models include the PCB 603C01 (ceramic shear ICP, 100 mV/g, 0.5-10 kHz, ±50 g) and the Wilcoxon 786A (100 mV/g ±5%, ±5% 3-5,000 Hz, 80 g range, 30 kHz resonance), with the 787A-M8-IS as an intrinsically-safe low-profile option and CTC AC916 / AC982 for ATEX Zone 0/1 duty.