Selecting the right measurement microphone requires matching several parameters to your specific application: the acoustic field type, capsule size, calibration class, frequency range, dynamic range, and signal conditioning interface. Making the wrong choice does not always produce dramatically wrong results — but it introduces systematic error that is hard to detect and harder to defend when results are questioned. This guide walks through each decision point.
The single most important selection criterion is matching the microphone design to the acoustic field in which it will be used. Free-field microphones (the most common type) are designed for use in a free progressive wave, pointing toward the sound source. Pressure-field microphones are designed for use in a uniform pressure field — inside a coupler, an IEC 60711 ear canal simulator, or a pistonphone calibrator. Random-incidence microphones are designed for diffuse fields such as reverberation rooms. Using a free-field microphone in a coupler, or a pressure-field microphone in free space, introduces frequency-dependent errors that increase with frequency and can exceed 5 dB above 5 kHz.
Class 1 microphones meet the tighter tolerance requirements of IEC 61094-4 and IEC 61672-1. They are required for precision research, ISO/IEC 17025 accredited testing, and standards that explicitly require Class 1 instruments. Class 2 microphones are appropriate for environmental monitoring, site surveys, occupational noise, and applications with less demanding uncertainty requirements. The class choice should be driven by the standard you are working to — not by a general desire for the best instrument available.
Match the self-noise specification to the quietest sound you expect to measure. A microphone with 25 dB(A) self-noise cannot reliably measure ambient noise below about 30 dB(A). For quiet environments, select a low-noise variant with self-noise below 15 dB(A). Match the upper SPL limit to the loudest source in your environment — leave at least 6 dB headroom above the maximum expected level to avoid clipping on transients.