Digital designs?

[Tony Clark]

> Let's take a look at your typical PC soundcard. Most of them are
> internal. Which means that you need some pretty good circuit design to
> get rid of all of that noise that comes from the good old PC. Now, the
> really good cards - your Card Ds, your Event Electronics - stuff
> designed for H/D recording - they have a S/N noise and frequency
> response up to spec. Even a Turtle Beach Tropez has a relatively good
> S/N and freq. response., though probably not better then a well designed
> standalone audio output section. 

   This is very true.  The computer power supply is serious noise.  
Granted, you can get rid of it fairly easily with careful design, but 
SHIELDING is where the biggest problem lies.  That's why it's better to 
remove the analog stuff to someplace outside the computer.

> Agreed, but a proprietary driver is not exactly trivial. It's not
> difficult, it's just not trivial. It mainly becomes difficult when
> manufacturers make SDKs that are not free of charge (Creative was guilty
> of this for a while... not a strategy I agree with personally). But the
> indication I get is that DSP-based coding by itself is even more
> non-trivial. I wouldn't trust Win95 for anything more than a simple
> monosynth (real-time at least) at this point. 

   Actually, I'd rather program a DSP than a microprocessor, especially 
of the Intel type.  With DSP chips, the instruction set is very small and 
therefore much more open to hacking than a micro.
   Granted, programming certain functions in DSP do take time to flesh 
out.  But they'll take the same amount of time to do it on a micro, and 
probably longer.

> The best solution seems to be a computer designed with soundcards or
> DSPs that have hack ability... which might be a rather interesting
> way to go, and the easiest way to achieve a powerful system with little
> hardware effort. That seems to be the indication of you and others as
> well. (:

   The best solution is to have a micro mastering a set of dedicated DSP 
chips.  Micros excell in I/O communication, where DSP chips excell in 
crunching numbers.  So you use the Micro to run the slow interfacing 
(i.e. keyboard, graphics) and to tell the DSP chips what programs to run, 
and let the DSPs crank out the sound.
   You can actually do this by buying a DSP evaluation board.  These 
usually run around $100 for a parallel port version (usually $400 for a 
computer compatible card).  You can then program the DSP chip and run all 
sorts of cool stuff off of it independent of how powerful your CPU is.  
The downside is that they typically don't have a lot of onboard RAM, so 
small programs can only be used.  :(

   Tony

-----------------------------------,----------------------------------
I can't drive (my Moog) 55!        |     
-----------------------------------|     The E-Music DIY Archive
Tony Clark -- clark at andrews.edu    |          COMING SOON!
http://aupe.phys.andrews.edu/~clark| 
-----------------------------------'----------------------------------