I want to build a (SMALL & NARROW FOOTPRINT) board that does audio DSP, so far, I tried :
- using a teensy 3.6, but it has a terrible ADC (50% SNR) and the I2S interface is glitchy at best
- I also tried on a RPI Zero, interfaced with an MPC3002, but, it's really hard to make the cpu access an ADC in real time (although going through many kernel and package optimizations)
then I wondered, is it that hard to make a board myself ?
- what CPU should I use (let's consider using JLPCB / PBWAY services to build the board) ?
- I would add fast/close to cpu ADC & DAC chips ?
- how would I implement the usb programming of it (or anything similar) ?
- is it possible to have it, arduino (or ATMEL studio) compatible ? you know, for c/c++ arm libraries (math, FFT...)
- should I instead use an FPGA ? I heard they were not optimized for DSP audio ?
just out of curiosity, I wanted to check that direction instead of ready made boards, that never do what I need
thanks for the time you'll spend on this question
As for RPI Zero: Do not use Linux!
not using linux is a huge gap, it means no broadcom binaries; so no GPU access, so no GPU math
one also have rewrite it's own USB code
so I'd rather make my own simple board
=> so not GPU as well ;-) And GPU for audio? Sounds like overkill.Anyway, if it is for fun, no problem. Else, there are soooo many boards out there. And adding am ADC/DAC via SPI should be possible with most of them.Or you go for a board with Xilinx ZYNQ.
GPU can do FFT/iFFT on raspbian (heavily used when processing audio), so if you wanna do DSP it's something the CPU would not have to deal with, so more room to do audio effects
I found Nucleo boards, still checking on these
Xilinx ZYNQ are not the same format of RPI nor teensy :-)
You mean "mechanical" format?
View all questions in Cortex-A / A-Profile forum