A five octave, 6 voice synthesizer

After about 2 years of cogitation, I've finally decided to do it - build a synthesizer. And finish it by the end of the year.

I'd originally intended to build an analogue synth, in the mould of the mini-moog. On further investigation, it transpired that the price/feature ratio of such a design was not particularly good. So, being a poor (and hungry) uni student, I'll be going with the cheaper (digital) option.

Basically, I want the synthesizer I design to have the following features

1. polyphony - must be able to play at least three notes simultaneously
2. the ability to split the keyboard, so that each half can be mapped to a different voice
3. pitch bending
4. analogue controls for setting the attack, levels of each harmonic, etc - I find digital controls clunky
5. 16 bit, 44kHz sound output, or better
6. Additive synthesis
   
7. Total cost to manufacture of less than $200

And now, finally, I have a (concept) design which meets all these requirements.

The design will be based around Atmel's atmega8 microcontroller. Actually, in order to meet the performance requirements, it will be based around seven of them. On the surface, using seven low performance, 8 bit micros, seems a little silly, particularly when there are plenty of excellent ARM based products out there, of similar price, but with vastly superior performance. However, there were several reasons for choosing the AVR product.

1. The atmega8 is really cheap (about $3)
   
2. Programmers for AVRs are really easy (and cheap) to build, driving down the total cost of the project (I don't own a JTAG based programmer, which is what most of the ARM based micros seem to require, and such programmers are quite expensive)
3. The documentation for the AVR family is excellent
4. AVR studio, supplied free by Atmel, is an excellent embedded IDE, and I am familiar with it
5. using multiple CPUs allows for a more modular design

So there. I've rationalized away my choice. Probably another reason would be that I'm much more familiar with AVR than ARM. On the downside, using this AVR based design means having to deal with communication and synchronization between multiple CPUs, which tends to be a process fraught with (enormous) difficulty.

But anyway, it's too late to change now (otherwise I'll never get this project finished). I'll keep udating this blog with my progress on this project, mainly as a way of keeping myself motivated. Hopefully, by this time next year there'll be plenty of photos of the finished synth up here, as well as a full set of schematics, firmware and pcb-layout.