[sdiy] Taking a Step towards - - --((FUTURE-PREDICTIONS))-- - -

[Magnus Danielson]

From: Brock Russell <brockr0 at shaw.ca>
Subject: Re: [sdiy] Taking a Step towards - - --((FUTURE-PREDICTIONS))-- - -
Date: Sat, 10 Jan 2004 13:33:00 -0800
Message-ID: <5.1.0.14.0.20040109110827.00a5ae40 at Pop3.norton.antivirus>

> Magnus wrote:
> 
> >From: James Patchell <patchell at cox.net>
> >Subject: Re: [sdiy] Taking a Step towards - - --((FUTURE-PREDICTIONS))-- - -
> >Date: Thu, 08 Jan 2004 18:27:32 -0800
> >Message-ID: <5.2.1.1.2.20040108182329.00b12308 at pop.west.cox.net>
> >
> > > At 12:05 PM 1/8/2004 -0500, RevTor at aol.com wrote:
> > >
> > > >so guys, all this talk of programable logic and "brute force" stuff
> > > >getting so cheap...
> > > >
> > > >how can it (or anything new) be applied to make new analog
> > > >toys???  besides 5.1 surround (which would be a cool module....)
> > > >Paul, you once dreamed of a programmable modular with a bus style
> > > >backplane or something TITAN you were going to call (?) it but it was $$
> > > >back before MOTM (??97-98?)  would this be more feasible today?
> > > >what about today?    the state of the art in opamps and IC's and
> > > >microprocessors fueled the programmable polysynths of the 80's, now the
> > > >computing power per dollar and DSP is fueling all the VA's and 
> > softsynths,
> > > >I guess today all the $$ is in tiny chips for celphones and flat tv's and
> > > >market driven make me vomit me too technology..  booring stuff....   how
> > > >can we apply whats new and hot to analog synths?? is there just too much
> > > >of a gap?
> > > >I really have no clue as to whats out there today on the periphery, Im
> > > >still trying to catch up with 70's tech.....
> > > >anyone care to comment?
> > > >are us through-hole dinosaurs destined for extinction? Say it aint so!!!!
> > > >Anyone?
> > > >FPGA's programmable as CEM style synths on a chip?
> > > >just a fun topic Id like to hear more about....
> > >
> > >          Well, I can easily see the XC3S400 being able to "emulate" at
> > > least 256 oscillators and 256 multi mode filters and 512 envelope
> > > generators (VCAs too), all at one time...in fact, if anything, I am
> > > probably underestimating.  The part seems to be capable of clocking at
> > > 200MHz,  the part should be able to do all that in about 1.3 micro
> > > seconds....kinda gives you an idea of the power we are talking about.
> >
> >Naturally it also has MIDI, key-assigner, CPU etc. all embedded.
> >
> >I've been toying witht he idea to answer this thread with a
> >MIDI-keyassigner-DCO design for an FPGA. I guess I am just being a bit lazy,
> >since it isn't very difficult to do. The DCOs are trivial really. I think I
> >would spend more time on the MIDI and key-assigner side. Maybe things like
> >pitch-bend would be something to spend time on how to solve neatly. Oh well.
> >I guess the project would die from creeping featurism.
> >
> >Cheers,
> >Magnus
> 
> I think a MIDI key assigner is an excellent idea. I'd use it. I've been 
> putting off doing one myself for over a year now.

Ah, see... ;O)

> Why would it die from creeping featurism? VHDL is inherently modular,
> the key assigner would simply be a component.

Don't preach for the choir! You have to consider that I looked at it from the
"lets do a quick project". A quick project ceases to be a quick project when
you are trying to make it more complex by adding more features. Now, that quick
project is dead, but the overall project might very well live. See the
difference?

> Isn't this the some idea that Paul M. had some time ago when he started
> the FPGA synth mailing list, a bunch of people contributing components
> that could be used as desired? Perhaps it would work better on this list.

Well, Paul got a whole bunch of inspiration from me among others.

> This seems like an interesting project but in the little bit of poking 
> around at this I've done, there is more involved than just cranking out VHDL.
> It is not a trivial design exercise to get a clean analog signal from a 
> circuit that includes a digital section clocking away at 100 or 200 MHz.

Well, to some degree this is true, but then again, those are problem to solve.
I think the main thing is to isolate functionality which may be done as
separate and reusable designs, specify that functionality, do the design
properly (i.e. not cheat on the MIDI-implementation for instance, with all the
known problems that brings us) and then test the hell out of that block.

A few design-blocks that I see:

MIDI SERDES
MIDI Decoder
Key assigner (think multitimbrality etc. and you get the problem)
"CV"-processor (do all the processing of various control values, including
expo-function)

and only then we are set for hacking out stuff like NCOs etc.

While the core of a NCO is trivial, it is getting all the other stuff in place
which is anoying.

Some thought has to be spent on how to best use up the FPGA since many things
operate at an insanely low speed compared to what the core-clock can be, so
multiplexing the functionality to reuse the logic comes as a natural thing to
do. It was done in the PPG Wave 2.3 Proz-board, so why not in an FPGA? ;O)

Cheers,
Magnus