[sdiy] dsp box hacking?

[Rainer Buchty]

> I'm not sure if the software has become more complicated (which is 
> possible) or whether lawyers started getting involved.

I'd say neither.

Once the custom chips get complex enough, it's hard to get the register 
interface right. For instance, when I was rev-engineering the Ensoniq 
SQ80, I was rather puzzled by the fact that the sync bit in the 
oscillator control word was set for both, hard-sync and AM.

It just didn't make sense, and I also completely overlooked that also 
the oscillators got shuffled depending on no or hard-sync set or AM. 
Once I got my hands on the DOC programming manual (which was also in the 
Apple IIGS manual) it was all clear, because hard-sync can only go from 
even to odd oscs whereas AM goes only from odd to even oscs.

If you want to have some quality time, grab the Casio FZ-1 OS and try to 
rev-eng

- the display protocol for its custom text/graphics controller
- the register interface for its GAA/GAB ASICs

Even though some kind soul from within Casio leaked the OS-internal data 
structures, memory map, and OS function list to some Norwegian FTP 
server, it's still not obvious what's going on here and would take some 
significant time.

And we're talking about 1986 technology here.

Today one would pack everything into a single ASIC. Custom CPU for 
control, dedicated DSP for computation, maybe custom logic for hardware 
acceleration of some tasks. Just have a try with your DVD player's next 
BIOS update...

Changing the menu texts or the power-on picture is easy. But anything 
else is hardest work.

Rainer