Sensors Converge 2024 offered the opportunity to see 200 exhibitors showing off sensors and related wares, some of them recently announced projects, or development projects being shared with the public for the first time. A big focus was on tiny electronics like new microphones that require lower power while still improving performance over previous generations. Many products incorporated artificial intelligence and Gen AI capabilities.
In just one example, TDK Invensense launched to global distribution and put on display its low power Invensense SmartSound T5848 MEMS microphones with an I2S interface. Along with the previous SmartSound T5838, the new T5848 microphone supports edge and Gen AI systems with what the company calls Acoustic Activity Detect (AAD) to offer high sound fidelity at ultra-low power.
The product is designed to enable IoT and edge AI wearables, TWS earbuds, AR glasses, smart speakers, home security, action cameras, voice-enabled TV remotes and other AI systems, managers at the TDK Invensense showfloor booth said.
In high quality mode, the two microphones deliver 68 dBA SNR (Signal-to-Noise-Ratio) and 133 AOP for what the company described as “excellent” sound fidelity which would help an application ensure accurate keyword detection in noisy environments while only consuming 330 microamps at 1.8 volts. When a voice keyword is detected, they operate at this higher power value, but drop to just 130 microamps in always-on low-power mode, prolonging battery life for systems that are always on. An ultra-low power edge processing feature consumes as low as 20 microamps, according to an online product description.
Battery life for remote IoT devices like smart speakers and home security—even voice-activated TV remotes-- is on the minds of designers everywhere. A device able to move from low-power to high-power scheme for devices is what edge device makers are asking for, TDK Invensense said.
AAD means the two microphones can be programmed to listen for acoustic activity such as keywords or voice commands so the application processor can efficiently manage power consumption at lower levels during idle periods. It is necessary so IoT and edge AI gear can remain in an always-on state and ready to respond to user commands, without unnecessarily draining the battery.
The T5848 also supports the I2S interface which saves components and reduces processing requirements such as filtering the mic output in the system hardware and software. (I2S is also known as Inter-IC Sound, a serial bus interface standard that provides digital audio data communication between integrated circuits in audio and other devices.) This interface allows direct connection to a wide variety of SoCs and microcontroller units. The high SNR is designed to help AI system receive high quality input by reducing noise where a user wants voice commands distinguishable from unimportant background noise. The microphone package is 3.5 x 2.65 x 0.98 mm.
“Voice interface is now ubitiquitous in IoT devices. Progress in AI technology makes voice interactions more natural and user friendly,” said Uday Mudoi, vice president and general manager of the microphone business unit at InvenSense, a TDK Group company. With the T5848 with its I2S interface, designers can enable simpler designs for always-on and Gen AI systems, he said. Mudoi spoke in a video interview with Fierce Electronics about the growth in microphone applications that are further enabled with Gen AI.
Separately, at the same booth, TDK showed off its range of sensors products for use in controlling electric motors in battery-power electric vehicles. One of several temperature sensors shown in different versions provides Negative Temperature Coefficient sensing, which a thermistor that is a resistor with a negative temperature coefficient. The resistance in an NTC thermistor decreases as the temperature goes up.
The company also makes integrated pressure sensors, high voltage controllers and valve position sensors as well as a TMR angle sensor. A TMR (tunnelling magnetoresistance) angle sensor can be used to detect the direction of a magnet’s magnetic field and then detect the angle between its magnetization directions. This insight is useful in detecting if a component in an engine is operating at the correct position.