[sdiy] dsPIC fun board

[Eric Brombaugh]

Tim Ressel wrote:
> I believe there is a certain set of features a DSP board needs to have. Here are things I think are mandatory:
> 
> Codec (at least 2x2 channels)
>  -- How else to get audio in/out?

Although this board supports an external codec, that's not the main 
focus. The primary motivator for this design is to explore algorithmic 
CV processing. The fact that the ADCs & DACs are capable of audio-rate 
sampling is incidental.

> Midi
>  -- This ubiquitous standard cannot be ignored

Supported with external optocoupler. I didn't want to commit to any 
particular serial interface on this design, so I provide a generic 4-pin 
  0.1" header with TX/RX/GND/+3.3V. Do whatever you want with it.

> Swimming in RAM
>  -- for time delay and table-driven effects

The dsPIC has 16kB of on-chip RAM which should be sufficient for DSP and 
short delay effects, but since audio processing isn't the prime focus of 
this project no additional RAM is provided. If you're interested in 
longer delays and/or larger buffers, please check out Seb Francis' 
DigiMod project which was discussed at length on this list a few months ago.

> CV I/O
>  -- again, a modular standard

4in / 4out, 12-bit resolution, ~50kHz sample rate and +/-5V range.

> USB
>  -- for fast comms to a PC

There are to my knowledge no dsPICs which support on-chip USB. Kind of a 
bummer from my perspective since that would be a nice feature to have. 
MCHP does have a similar processor family - the PIC24 series - which has 
several parts with USB OTG capability. They share the same processor 
architecture as the dsPIC, excluding the DSP MAC function.

An alternative is to use an inexpensive USB/Serial converter chip, such 
as the FT232R to connect to the UART port on this board.

> Now there are implementation details not covered. For example I would have an AVR coprocessor handle Midi, USB, and possibly CVs. In other words, off load all non-dsp stuff onto a processor better suited for those kinds of tasks.
> 
> I think a board like that would be pretty universal. Now I know some are a bit intimidated by the thought of coding a DSP, and adding a coprocessor just ups the ante. Perhaps a clever implementation could have the AVR as an option and run without it if needed.
> 
> Man, this is exciting! 

Sounds like you've got some interesting ideas. Why not do your own 
design and add to the fun?

Eric