MEMS accelerometers

Part 2 — Receivers

Learning objectives

  • Distinguish velocity (geophone) from acceleration (MEMS) measurement
  • Interpret noise-density curves for both sensor types
  • Identify the frequency band in which each sensor is preferred
  • Explain why MEMS has taken over land nodal recorders despite a higher noise floor

A MEMS (microelectromechanical systems) accelerometer is a silicon chip holding a microscopic proof mass on a silicon-spring suspension. Displacement of the proof mass relative to its frame is read out capacitively (two thin electrodes; displacement changes capacitance). Unlike a geophone, the MEMS output is proportional to acceleration, not velocity, and the transfer function is flat from DC up to several hundred Hz — no f₀ roll-off.

Mems Vs GeoInteractive figure — enable JavaScript to interact.

Two very different response shapes

Geophone: rolls off as f² below f₀, flat above. MEMS: flat everywhere within its bandwidth. Plotted together on log-log axes, the two curves cross near the geophone’s corner frequency. Below that, MEMS wins by a lot; above that, geophone (flat with LOWER noise) wins.

The noise-floor story

Geophones convert mechanical velocity to voltage via a massive moving coil in a magnetic field. The electromotive force is large: a good 10 Hz geophone has self-noise around 1–3 nm/s/√Hz in its flat region.

MEMS accelerometers rely on a tiny proof mass (millionths of a gram) read by capacitive sensing, whose thermal floor is fundamentally higher than coil-magnet induction. Modern seismic-grade MEMS such as the Sercel DSU3 or Colibrys SF3000L show self-noise around 20–100 nm/s/√Hz at 10 Hz — roughly 10–30× noisier than a good geophone.

But the MEMS’s flat DC-up response means it does hold its noise floor down to 1 Hz and even below. A 10 Hz geophone down there is rolled off by 100×, so its referred-to-input noise explodes. Below ~5 Hz, the MEMS wins on equivalent-input-noise terms.

Why MEMS has taken over land nodal systems

  • Size: MEMS chip is the size of a grain of rice; a 3C node is pocket-sized.
  • Power: milliwatts, vs the substantial power a geophone+amplifier system needs.
  • Reliability: no moving coil, no wire fatigue, no liquid-damping seal.
  • Tilt-insensitivity: MEMS can be left in any orientation and software can rotate the data.
  • Production: silicon processes allow millions of identical units — essential for 100,000-channel nodal surveys.

Per-channel noise is still worse than a premium geophone; projects needing ultimate low-f performance (marine FWI, passive microseismic) still use geophones or hybrid systems. For a 100,000-node land seismic survey, MEMS is the only economically-feasible choice.

References

  • Mougenot, D. (2013). MEMS-based 3C accelerometers for land seismic acquisition. The Leading Edge, 32(4), 388–396.
  • Sheriff, R. E., Geldart, L. P. (1995). Exploration Seismology (2nd ed.). Cambridge University Press.
  • Pritchett, W. C. (1990). Acquiring Better Seismic Data. Chapman & Hall.

This page is prerendered for SEO and accessibility. The interactive widgets above hydrate on JavaScript load.