Skip to main content

Accelerometer Integration

Image Source: bestforbest/Stock.adobe.com

By Tenner Lee for Mouser Electronics

Published March 14, 2022

Accelerometers are electromechanical devices that can measure constant forces such as gravity and/or dynamic forces on the device due to movement or shocks. Accelerometers are ubiquitous in both commercial and industrial applications, including navigation and motion detection. This article introduces accelerometers and discusses four design considerations for their integration and implementation: Physical layout; operational requirements; noise, temperature, bias, and sensitivity; and interfaces.

Accelerometer Categories

It is important to consider what an accelerometer does and whether it is a right fit for the device or application that is being designed. Accelerometers can generally be described by three broad categories based on how the sensor measures acceleration through forces being applied on the device and how the sensor physically operates. These three categories are briefly described in Table 1.

Table 1: Accelerometer categories: Compression mode, shear mode, and capacitive. (Source: Author)

Type

Explanation

Notes (Broad Categorization)

Compression Mode

Measures acceleration through a piezoelectric material where acceleration is measured through compression when a force is applied.

 

Higher resonant frequency

Large dynamic range

Large bandwidth

High accuracy

Shear Mode

Measures acceleration through a piezoelectric material where acceleration is measured through shear stress when a force is applied.

 

Large dynamic range

Large bandwidth

Higher thermal isolation

Higher accuracy

Capacitive

Measures acceleration through the deflection of a mechanical arm or substrate that results in capacitance changes of the base circuit.

 

Low cost

Relatively poor SNR

Limited dynamic range

Small bandwidth

Good Accuracy

Beyond these three broad categories, accelerometers can be further defined by their configuration and capability:

  • Number of measurable axes (spatial dimensions that need measuring): Accelerometers can come in 1, 2, and 3 axis configurations (in some cases >3 when a gyroscope is integrated within the sensor).
  • Output (digital or analog): Accelerometers are commonly defined by their output or their interfaces with a host device. Accelerometers that are digital typically include the required signal conditioning, control logic, and an A/D converter, while an analog accelerometer output typically omits these additional components and outputs raw voltages that are proportional to the acceleration measured.
  • Ability to measure a rest frame (the frame of reference in which the device is at rest): Accelerometers that can measure the rest frame (gravity) are usually referred to as DC response accelerometers. Those accelerometers that cannot measure low frequencies or a rest frame are called AC response accelerometers.

Accelerometer Physical Layout

Layout includes the mounting, orientation and alignment, and proximity of the accelerometer to other components. The layout and placement of the accelerometer affect how the accelerometer performs over the lifetime of the device.

Mounting

In general, the accelerometer should be mounted onto a rigid board/substrate with direct coupling to the body of the device or the object under test. Improper mounting of the accelerometer will result in degraded readings such as loss of sensitivity or, at worst, erroneous/improper readings. If the mounting substrate or board is not rigid, measurements can be dampened and harmonics/resonances can be introduced to the sensor, causing improper readings.

Mounting Mitigation

Directly mount the sensor to the device body or object under test. This removes concerns of structural resonance and coupling.

Orientation and Alignment

Orientation of the accelerometer to the principal axes of the host device is important. Misalignment in terms of inclination and rotation of the sensors’ axes to the host device will result in improper readings proportional to the misalignment (rotation/inclination) angle.

Accounting for misalignment as part of the systematic (bias) noise of the sensor is important to consider as part of the design process. The misalignment error will be a function of the device/sensor integration tolerances rather than a specification on the device.

Orientation & Alignment Mitigation

Calibration of the accelerometer after the sensor has been integrated can remove any misalignment errors. Otherwise, better integration to the host device is recommended.

Location and Proximity

Finally, location and proximity of the accelerometer to other components is important to consider as temperature and user/component vibrations will affect the performance of the sensor. In every application, measurements from the accelerometer should only come from the host device or desired object under test. Isolation of the accelerometer from other device components is generally good practice. Isolation of the accelerometers from temperature fluctuations can reduce sensitivity drift and noise up to the internal specifications of the accelerometer. The size and packaging of the accelerometer will dictate how and where the accelerometer should be placed.

Location Mitigation

Calibration of the accelerometer after the sensor has been integrated can remove any misalignment errors. Otherwise, better integration to the host device is recommended

Accelerometer Operational Specifications

Operational Range

Operational range is a key consideration when deciding on an accelerometer to meet application specifications. Accelerometers can operate over different measurement ranges depending on configurations sets defined by the manufacturer. For wider measurement ranges (e.g. ±2000g), increased noise or decreased sensitivity is typically observed and can be referenced in the sensors’ respective data sheets for clarification. For higher sensitivity applications, a reduction in the overall measurement range of the accelerometer can be used and suggested. On the other hand, for industrial applications that require large measurement ranges, sensitivity is typically reduced to account for the larger measurement band.

Bandwidth/Frequency Response

The bandwidth/frequency response or sampling of the accelerometer is also important and is fundamental to the operational range of the accelerometer. Typically, the bandwidth of an accelerometer can range from kHz to MHz and depends on the design of the manufacturer and type of accelerometer used. For digital accelerometers, the bandwidth defines the rate in which measurements are sampled without aliasing. For analog accelerometers, bandwidth defines where the mechanical resonances occur or when the response falls to -3dB of the nominal range of the circuit. Accelerometers will always have a defined frequency operating range where measurements are valid (as tested by the manufacturer)

Low-Frequency Cutoff

For AC response in accelerometers, a low-frequency cutoff is also defined. The low cutoff frequency is often defined in the data sheet when applicable. It should be noted that impact events, such as car crashes or dropping a phone, are inherently high frequency and high magnitude response changes in acceleration. These events or applications to sense these events typically require large operational ranges to capture variations of these events (high bandwidth and operating measurement ranges).

Linearity

Nonlinearity of the accelerometer must be considered to make sure outputs remain consistent/accurate and do not clip. If an accelerometer is incorrectly configured such that the accelerometer is operating around the nonlinear region of the device, the accelerometer in most cases will not function as desired with clipping and incorrect/saturated measurements as a result. In very specific cases (and if the designer knows what he is doing) the nonlinearity of an accelerometer can be exploited for increased sensitivity.

Operational range Mitigation

Careful accelerometer selection and design. Mitigations are limited. Expertise and good system design.

 

Nonlinearity Mitigation

Design of mechanical dampening can be used to extend the operating range by moving resonances to higher frequencies outside the desired range. Signal processing to exploit nonlinearity.

Accelerometer Noise, Temperature, Bias, and Sensitivity

Spectral and Total Noise

Noise within an accelerometer is often defined in two ways: Spectral noise/noise density and total noise. In most cases, spectral noise/noise density is the more applicable figure to be utilized. Noise density is generally defined in μg / √Hz. As defined and stated in most spec sheets, a given noise density is only valid for a given frequency range and at an assumed temperature. As frequency increases, the noise density typically decreases. As a result, accelerometers operating at higher frequencies exhibit a lower noise density.

In most applications, low frequency is of interest, and an accelerometer with low noise density at low frequencies (DC) is ideal. The RMS acceleration noise of the sensor (total noise) is calculated by multiplying the noise density by the square root of the measurement bandwidth.

Noise Mitigation

Temperature control. Good design placement. Calibration and/or averaging measurements across time or multiple accelerometers.

Thermal Noise and Sensitivity

Temperature is a key component that needs consideration when talking about noise. Increased temperature results in increased noise and sensitivity shifts of the accelerometer. Likewise, low temperatures result in decreased sensitivity. Specific accelerometer configurations need to be selected in order to operate at high temperatures and adverse environments.

Depending on the accelerometer type and design, accelerometers need compensation in order to adjust shifts and noise levels. In many accelerometers, this step is already done and thus no additional compensation is needed through temperature feedback compensation (must verify when selecting a device). If operating temperatures are expected to deviate from the nominal 20 degree Celsius, it is worthwhile to measure and see deviations, as some spec sheets do not have corresponding data.

Thermal Mitigation

Temperature control. Additional feedback loops to compensate for temperature drifts.

Bias

All accelerometers contain some bias or offset in measurements. For DC response accelerometers, this is typically seen in the output value added in the accelerometer output in order to measure/provide a reading the rest frame. Bias can be confirmed for any given accelerometer by referencing the specifications with measurements from the accelerometer in hand.

Bias Mitigation

Calibration.

Cross-Axis Sensitivity

For multi-axis accelerometers, cross axis sensitivity is another source of noise that must be accounted for. Ideally, since measured axes are orthogonal to each other, coupling of measurements is zero (0), and this is not a problem. Unfortunately, measurements can and do leak from other axes into the axis of interest due to manufacturing tolerances or design. Cross-axis sensitivity is inherent to the accelerometer and, unlike orientation error, is not part of the integration process. Cross-axis sensitivity is measured as a percentage.

Cross-Axis Sensitivity Mitigation

Calibration and additional processing if cross-axis sensitivity is an issue. Consideration within design specifications to account for errors.

Interfaces

Unlike analog accelerometers, digital accelerometers have defined interfaces requirements in order to read measurements and control the sensor. Typical interfaces include I2C and SPI; insight and how these interfaces can be used to extract and configure the accelerometer is beyond the scope of this article. It will be noted, however, that digital accelerometers due to added control logic allows for more capabilities and ability to access measurements at specific instances when needed (or not needed for sleep mode). Selecting the right accelerometer and having the right system design will determine the best interface whether it is analog or digital.

Conclusion

Accelerometers are electromechanical devices that measure acceleration forces and are used in a wide range of applications, ranging from cell phones to cars and navigation devices. Categories of accelerometers include compression mode, shear mode, and capacitive, and they can be further categorized according to configuration and capability. When integrating and implementing accelerometers, accounting for multiple design choices is important, including physical layout; operational range and linearity; noise, temperature, and sensitivity, and interfaces. It is recommended you refer to application notes and technical journals for insights and specific details, as this article is only a first order description.

About the Author

Project and program technical lead for Machine Learning/Artificial Intelligence research and development. 15 years of experience leading, developing, managing projects, and advising/consulting on algorithm development/design, system optimization, and algorithm testing/validation. Graduate degree in electrical engineering with foundation in signal processing and EM.

Profile Photo of Tenner Lee