Multichannel Acoustic Measurements: When to Verify Synchronization and Phase Matching

When a multichannel acquisition system is used to compare sound pressure levels at individual positions under steady operating conditions, interchannel phase may not be the leading specification. If the data will feed array beamforming, time-difference-of-arrival analysis, or transfer-function calculations that include phase, checking amplitude calibration alone is insufficient. Before selecting equipment, ask whether the final decision depends on the timing relationship between channels.

Separate three checks

The first concerns sampling clocks and acquisition starts. Setting two devices to the same sampling rate, or sending them the same start command, does not by itself establish synchronization throughout a recording. NI distinguishes shared triggering from clock synchronization: devices with a common trigger but unlocked clocks can drift relative to one another during acquisition. The implementation must follow the supported hardware and driver configuration. (Source 1)

The second concerns relative measurement-path delay. Even with synchronized clocks, differences in digital-filter group delay require attention. Establish whether compensation is applied and which module combinations and configurations it covers. Automatic compensation on one device should not be assumed to cover every channel combination. (Source 1)

The third concerns phase response across sensors. Feeding the same electrical signal to acquisition inputs bypasses the microphones, so that test cannot establish phase matching of the complete acoustic measurement chain. GRAS identifies phase matching as an important characteristic of array microphones used for source location. Amplitude calibration records and evidence of phase matching should therefore be reviewed separately. (Source 4)

Why delay matters

Delay-and-sum beamforming compensates for propagation-time differences based on source direction and array geometry before combining the channels. Additional channel delay enters that timing relationship. Transfer-function analysis includes both magnitude and phase, so similar magnitudes do not establish correct phase results. (Sources 2 and 3)

For a pure time delay, the absolute unwrapped phase difference is 360 × frequency × delay, with frequency in hertz and delay in seconds. For example, a delay of 10 microseconds corresponds to 3.6° at 1 kHz and 36° at 10 kHz. This calculation illustrates frequency dependence; it is not an equipment acceptance limit. A chain with a frequency-dependent phase response also cannot automatically be corrected using one fixed delay.

Verify the configuration being purchased

The following checks are engineering recommendations based on these principles, not universal acceptance requirements.

First define the deliverable: independent sound pressure levels, array localization results, or a complex transfer function. Independent level comparisons under steady conditions should not automatically require full phase calibration simply because many channels are involved. Comparisons of transient events still require attention to time alignment.

Request synchronization documentation for the intended sampling rate, channel count, and intermodule connections. Distinguish within-module specifications from cross-device specifications. Also check filter settings, whether compensation is enabled, and the frequency range covered by the sensor phase information.

Verification can be organized in two stages. First, where the electrical interfaces are compatible and the connections have been checked, apply a common-source electrical signal to test the acquisition paths. Then use an acoustic method with a defined reference and controlled propagation paths to check the chain including the microphones. Do not treat genuine acoustic propagation differences between spatial positions as instrument errors, or remove the test object's real delay merely to make curves coincide.

For acceptance, agree on the frequency range, recording duration, and allowable residual differences, and examine phase versus both frequency and elapsed recording time. Tolerances should follow the target analysis's sensitivity to errors rather than values borrowed from another model or application. The purchasing evidence should establish whether the intended configuration supports the analysis, not simply how many waveforms the equipment can display at once.

References

1. NI: NI-DAQmx Synchronization of PXI Express Modules https://www.ni.com/en/support/documentation/supplemental/10/synchronization-explained.html
2. MathWorks: Beamforming Overview https://www.mathworks.com/help/phased/ug/beamforming-concepts.html
3. NI: Modal Analysis in LabVIEW https://www.ni.com/en/shop/labview/modal-analysis-in-labview.html
4. GRAS: Array microphones https://www.grasacoustics.com/products/array-microphones