Measuring vibration in three directions with one compact sensor requires more than adding sensing elements to a chip. The mechanical structure must direct acceleration-induced stress to the appropriate sensing elements while limiting responses to movement on other axes. A reported MEMS three-axis accelerometer design addresses this challenge with double L-shaped beams, two connected mass blocks and three piezoresistive Wheatstone bridges. Its results illustrate how a structural design choice—the width of the intermediate beams—can affect sensitivity and cross-axis response.
How the three-axis accelerometer converts motion into voltage
The device combines an elastic structure with a piezoresistive sensing circuit. Four double L-shaped beams connect the supporting body to two mass blocks, while a pair of intermediate beams connects the blocks to each other. Twelve piezoresistors are positioned at the roots of the beams, where deformation produces stress.
Each group of four resistors forms a Wheatstone bridge for one measurement axis. The bridge associated with the x-axis responds to stress in the corresponding L-shaped beams; a second bridge measures the y-axis. Piezoresistors on the intermediate beams make up the third bridge, which measures z-axis acceleration.
When the chip accelerates, the masses and beams deform. Some resistors experience compressive stress and others tensile stress, changing their resistance through the piezoresistive effect. The resulting imbalance in a bridge produces an output voltage. At a constant supply voltage, that output reflects the relative resistance changes and, in turn, the stress produced by acceleration.
This arrangement gives each axis a primary sensing channel. In practice, however, mechanical coupling can also cause a response in a channel that is not aligned with the applied acceleration. These unintended responses are described as cross-axis sensitivity or cross-interference.
Why intermediate beam width matters
The researchers examined how the width of the intermediate double beams affected stress at the piezoresistor locations. Simulation compared six widths, from 250 to 500 micrometers, under acceleration applied along each axis. The simulated stress for the z-axis sensing elements decreased as beam width increased, while stress for the x- and y-axis elements changed comparatively little.
Because the bridge output depends on stress-induced resistance changes, this variation influences the sensitivity of each axis. A design that produces very different sensitivities from one direction to another may be less consistent across its three measurement channels. The simulation indicated that a 500-micrometer width brought the average stresses across the axes closer together, prompting fabrication and testing of chips at the selected dimensions.
The comparison is useful for sensor designers because it shows that optimizing one mechanical feature can change the balance among measurement axes. Beam width is not simply a packaging dimension: it affects the stress delivered to sensing elements and therefore the electrical response.
Fabrication and measurement approach
The accelerometers were fabricated on silicon-on-insulator wafers using MEMS processes. The process formed the piezoresistors and electrical connections, then etched the silicon to release the beam and mass structures. A grooved glass plate was electrostatically bonded to the chip to provide movement space for the masses. The chips were attached to printed circuit boards and connected using wire bonding.
Testing was conducted at room temperature with a 5.0 V supply. Sensitivity measurements applied acceleration along the respective sensing axes from 0 to 15 g in 3 g steps. Frequency response was evaluated over a 100–10,000 Hz sweep, with a 3 g excitation. These conditions matter when interpreting the reported values: they describe the tested devices and setup, not a universal specification for accelerometers of this design.
Reported sensitivity and cross-axis response
For the device with 500-micrometer intermediate beams, the measured sensitivities were 0.302 mV/g on the x-axis, 0.235 mV/g on the y-axis and 0.347 mV/g on the z-axis. The reported linearity figures were 4.1%, 3.4% and 3.4%, respectively. Compared with the other tested widths, this configuration showed more closely matched sensitivities across the three axes, especially because z-axis sensitivity decreased as intermediate beam width increased.
The test results also show that outputs on the non-sensitive axes changed only slightly when acceleration was applied along a given axis. The reported cross-axis sensitivities for the 500-micrometer device were:
- With x-axis acceleration: 0.011 mV/g on y and 0.004 mV/g on z.
- With y-axis acceleration: 0.019 mV/g on x and 0.014 mV/g on z.
- With z-axis acceleration: 0.001 mV/g on x and 0.004 mV/g on y.
Resonance frequencies for the same device were 8,898 Hz on x, 8,395 Hz on y and 3,270 Hz on z. The associated reported bandwidths were 100–8,000 Hz, 100–7,500 Hz and 100–3,000 Hz, respectively. These axis-specific results underline the importance of considering frequency response alongside sensitivity when selecting or designing a vibration sensor.
Practical implications for automation teams
For engineers and system integrators, the central lesson is that multi-axis performance depends on the interaction between mechanical geometry and signal conversion. A three-bridge electrical architecture can separate the intended axis outputs, but the beam-and-mass structure determines how acceleration becomes stress—and how much of that stress appears in other channels.
When evaluating a MEMS accelerometer for equipment monitoring or another measurement task, teams should consider the required axes, operating frequency range, sensitivity balance and acceptable cross-axis response together. The reported design provides a fabrication example, not evidence of suitability for every industrial environment or application. Installation conditions and system-level requirements still need to guide sensor selection.
Conclusion
A double L-shaped beam structure paired with piezoresistive Wheatstone bridges offers a method for measuring acceleration along three axes on a MEMS chip. In the tested design, adjusting intermediate beam width improved the consistency of axis sensitivities and supported low measured cross-axis responses. For automation decision-makers, the practical takeaway is to treat mechanical geometry as a key part of sensor performance—and to assess sensitivity, linearity, cross-axis behavior and frequency response as a combined set of design criteria.