Microphone array.

Goal is to build a reasonable cheap and simple microphone array, suitable for beam forming. 

Span was originally intended to be meters, but on reflection, that's silly. The span should ideally be something like 1/2 a wavelength. Audio wavelength at 4KHz is about 343/4000 meters or about 8 cm. Note that having only two microphones on each axis will cause aliasing reducing the beam resolution.

The desire to be able to do beam forming requires that the samples from the ADCs be taken in a syncronized fashion.

I'm planning to use the Atmel NWG100 as a driver board, running power and data over cat-5 to it. (it's cheap, supports ethernet, and doesn't require cooling). This board also has 3 x I2S interfaces which most audio ADCs will output.

I'd like to have at least 4 microphones (there's some advantages in having more).

Microphone array background

Design

Board is based on the AD1974 ADC. This is a 4 channel, 24-bit, 192KHz part. It has an on-board PLL, avoiding the need to generate a high frequency clock.

This use the Frame sync signal from the AP7000 CPU to drive the PLL, given a TDM data stream back to the SSC on the CPU. The inputs are intended to be connected to the pre-amp boards.

This is a pretty standard electret mic pre-amp. It should give about 60 - 80 dB of gain. (Which is pretty drastic, but fitting with the distance speaking purpose that it's intended to be used for). The L1/L2 are to filter the power as much as practical. Signal inputs from the WM-61A mic are in the 0.75mV range, so low noise is critical.

The preamp board lays out to around 30 x 15 mm.

The ADC board is around 48x48mm

Board is split into analog and digital ground planes, with inductors connecting power and ground between digital and analog side. Lots of decoupling capacitors everywhere, and the 0.1" header matches the J1 connector on the NGW100.

Boards have been ordered from Olimex.

Parts list from DigiKey (also includes some 1-wire parts).