FPGA board choice

I have already described the project goals of XFM in the main article. However, many builders are curious on what was the selection criteria behind the Mojo v3 (S6) and the CMOD (A7).

XFM is an exercise on finding limits. It’s part of a series of sequentially growing synthesizers, drum machines and noise-making devices, which attempt to squeeze every bit and cycle of a silicon device. I find the process of exploring the limitations of every device very enlightening and inspiring.

This synthesizer is the very first hardware project that goes public. This is because I’m under the impression that a) it’s good enough as to be interesting to many, and b) it has something to offer to many different enthusiasts.

Musicians in a budget can find a builder who makes them an XFM, and have a very inexpensive and powerful module to expand their performance musical set. Builders could build table-top, rack-mount or Eurorack casings for XFM and sell them for a small profit.

Experienced builders and programmers could create interesting user interfaces for XFM, with lights and buttons and knobs and displays and what not. Even a keyboard synthesizer based on the module would be an interesting project. While the HDL for XFM isn’t open, it can essentially be considered a “32-voice, 6-op synthesizer IC commanded by MIDI an fast UART”.

Using multiple modules, a bigger synthesizer can be easily created through aggregation (with a structure similar to the TX-816). I myself have an already running 320-voice monster prototype based on the A7.

And finally, expert builders could design a dedicated synth board, including one or many XFM instances, and build a next-gen version of it.

So to support all those possibilities, here’s the criteria list. I’ll of course leave FPGA capacity aside, as that was already covered in the main article.

Price
Of course. $50 is magnetic enough as to try it. Aggregating 8 x S6 boards can result on an 8-part, 256-voice, 6-operator FM synthesizer with individual outputs for less than $400 (only one photocoupler is needed). And for the expert board designers willing to do some EAGLE, much less.

Power consumption
This is probably the most distinctive advantage of having a FPGA-centric design: it’s extremely battery-friendly. The S6 version can run 12 continuous hours on a single power pack load.

Size
Fitting the design in the smallest possible space is always important, yet more for aggregation projects.

The Mojo and the CMOD are of two different capacities (theoretically, the 35T version of the CMOD is 3.6x the Mojo, yet that doesn’t translate exactly because of structural differences). Yet, it is the most competitive option in the market I found when going for the above list.

Did I miss a board? Let me know!