This video provides a comprehensive review, teardown, and experimental analysis of the FOTRIC TD2 Acoustic Camera. The core of the device is an acoustic beamforming system that uses a sparse, spiral-patterned array of digital MEMS microphones to localize sound sources across a wide frequency range (2kHz - 65kHz). A detailed teardown reveals the internal architecture, which relies on an FPGA for real-time processing of the microphone data, which is then overlaid onto an image from a standard visual camera.
Through a series of controlled experiments, the video demonstrates the camera's performance characteristics, confirming that localization accuracy improves significantly with higher frequencies and with sounds that have broader spectral content (like noise or square waves). Practical applications are showcased, including the successful identification of ultrasonic coil whine from electronics and pinpointing a tiny, inaudible vacuum leak in a noisy mechanical environment. The camera proves to be a capable and effective tool for industrial diagnostics, leak detection, and identifying sources of electrical or mechanical noise.
Analyzing the FOTRIC TD2 Acoustic Camera: From Teardown to Practical Application
0:07 Introduction to Acoustic Cameras: The FOTRIC TD2 is an instrument that localizes sound pressure waves using a beamforming microphone array and overlays the location of the sound source onto a visual image, allowing the user to "see" sound.
1:21 Teardown and System Architecture: The camera's front-end is a self-contained, modular unit. This design allows the manufacturer to use the same main body with different sensor heads (e.g., thermal, acoustic).
4:25 Spiral Microphone Array Theory: Unlike the square grids used in RF applications, the camera employs a spiral pattern of microphones. This configuration, with microphones spaced closer in the center and farther at the edges, allows for effective spatial sampling across the very wide range of wavelengths found in acoustic signals.
5:57 Sparse Array for Cost Efficiency: Not all possible microphone positions on the spiral are populated. This "sparse array" design reduces the cost and computational power required for processing, as fewer microphone signals need to be handled.
8:08 Digital MEMS Microphones: The system uses digital MEMS microphones that output a Pulse Density Modulated (PDM) signal. This simplifies the design by eliminating the need for complex analog routing and simultaneous A/D conversion, with all microphones synchronized by a single clock.
9:26 FPGA for Real-Time Processing: A dedicated FPGA is used to process the digital signals from all microphones in real time. This allows the complex beamforming calculations to be performed on the sensor head before the final data is sent to the main computer for display.
13:42 Powerful Filtering Capabilities: The user interface allows for both spatial filtering (isolating a specific area of the scene) and frequency filtering (selecting specific frequency bands, such as "audible" or "ultrasonic"). This is crucial for isolating sounds in noisy environments.
16:40 Identifying Hidden Electronic Noise: The camera accidentally discovered a persistent ~42 kHz sound, which was traced to the DC-DC converters in the overhead LED lights. This demonstrates its ability to detect high-frequency coil whine from electronics that is inaudible to the human ear.
19:03 Performance Across Frequencies: Controlled tests show the camera struggles to pinpoint low-frequency sounds (2 kHz) but becomes increasingly accurate as the frequency rises (10-25 kHz). This is an expected characteristic of beamforming arrays.
22:42 Multi-Source Detection: In a multi-tracking mode, the camera successfully identified and tracked two separate speakers emitting different ultrasonic frequencies simultaneously, proving its ability to handle complex soundscapes.
26:17 Practical Application: Finding a Vacuum Leak: The camera was used to find a tiny, inaudible leak in a vacuum hose, even while the loud vacuum pump was running nearby. By using its spatial filtering feature to ignore the pump, it easily pinpointed the hissing sound from the leak.
28:24 Post-Analysis with PC Software: The included PC software allows for offline analysis of saved images and recordings. All data, including frequency content and the sound itself, is embedded in the file, enabling a thorough review after the measurement is taken.