[sdiy] DSP cards

[Ethan Duni]

> But OTOH consider this : first I mention my attempt to emulate
> a DSP with a uC, a multiplier and some SRAM, and someone tells
> me there's some easy-to-implement & cheap DSP.
> Then I suggest to put such a DSP on a PCI card, and someone else
> tells me there's plenty of such cards with dev. tools and DSP code.
> Perhaps tomorrow someone will tell me to forget about DIY and buy
> some Korg or Roland gear...

-I think it's a question of the scope of the project you're looking to take
on. For most people that do this stuff at the hobby level, projects must be
limited in terms of size and complexity. So, my guess would be that if you
spend a bunch of time putting together your own dsp on a PCI card, you won't
have much time left to concentrate on the algorithms and software.
Conversely, if you use a Chameleon, you'll be able to spend all of your time
on the software. It could be that you'll want to work on this long and hard
enough to do both, in which case the experience of doing both the hardware
and software will probably be very rewarding. Also, you'll probably end up
with hardware that better suits your software and software that better uses
your hardware if you design both in tandem. However, going down that road,
you may spend years and years in the development phase before you have
anything fun and musical to show for it. Meanwhile, people on Chameleon type
systems will have long since perfected a bunch of really sweet software to
play with.

Using pre-fab systems like Chameleon and development kits also has the
advantage of a user base. That is, you can ask people for help, swap code
and eventually distribute (or even sell) what you invent to the other users.
This generally saves you from having to re-invent the wheel again and again
(provided you'd like to avoid doing that).

If you're serious about spending a lot of time designing a system from the
ground up, perhaps you should consider getting a job for a company like e-mu
that actually does music systems from the hardware level on up. That way
you'd get the opportunity to use expensive state-of-the-art developement and
manufacturing resources while being PAID to do this.. :]

Anyway, give some thought to what you ultimately want to get out of the
process (an actual instrument to play with or the experience of doing the
design?) and how much time you can realistically dedicate to it.
Fortunately, there seems to be a pretty good solution for any niche that you
may find yourself in.

Ethan