Posted on 2019-03-06 | Tags: research-area-ears
I've finished testing version 0.1 of the ear controller prototype board and made significant progress on version 0.2. Here is a summary of the results.
The three power supplies worked well. The TI LMZ14203H, using a design generated by the WEBENCH Power Designer, produced a solid 6 volt supply from the 12 volt input. It handled sudden servo movements without major voltage variations. The two linear regulators that provided separate 3.3 volt supplies from the 6 volt supply worked as expected.
The STM32F051 microcontroller was relatively easy to work with. I developed drivers in Rust for the analog-to-digital converter and for PWM output from timers.
The analog parts of the board, which were intended to extract the amplitude of the audio signal from a microphone, were only partially functional. Some of the intermediate signals had inexplicable periodic pulses.
To solve these problems and move closer to a production-ready design, I'm working on version 0.2, which has some major changes.
Version 0.2 works with MEMS microphones that have integrated ADCs and produce digital audio over I2S. This eliminates the large analog section of the board and the separate analog power supply.
Version 0.2 uses an STM32F411 Cortex-M4 microcontroller that supports DSP instructions and should be powerful enough to do whatever signal processing is necessary. It can receive digital audio from up to 8 microphones.
Version 0.2 uses only 26 distinct components, compared to 31 for version 0.1. The board is smaller, measuring 46 x 53 mm compared to 56 x 65 mm.
Version 0.2 can connect to up to 8 microphone boards. Each board is only 14 mm long and 8 mm wide.
After checking the design and reviewing the bill of materials, I'll publish the design. Later, I'll assemble and test a board.