SPICE for audio?

[Tibbon]

I'm working on prototyping a new product. I have the ability to build projects (kits) and I have a block-diagram chart of what I want to build. I have built similar sounding/acting things in Logic/Reaktor. I could probably whip it up in csound in a few minutes.

However, before I try to get PCBs made, or do anything, I'd like to try to do it electronically through simulation. Apparently SPICE does this, but I'm a little confused as to how it works. Is spice useful for modeling analogue time-based (bbd) audio units?

Thanks in advance. Spice looks a bit confusing, and rather like learning a new programming language.

[the scum]

An easy way to see what it'll do would be to play with it a little. There's a free version here:
Linear Technology - Design Simulation and Device Models

You can whip up simple opamp/transistor/logic circuits, and tinker with them pretty quickly. This version uses a GUI for putting the circuit together, and assessing the results, so there's not too much programming involved.

The version linked above is somewhat notable in that you can use WAV files for input/output.

The simulation is constrained by the component models used. Some models are better than others...for instance, a generic opamp might have infinite CMRR and PSRR, output current capability, GBW, etc. Anything I spice, I also build on breadboard to make sure that theory and practice actually overlap.

You might be hard pressed to find a model of a BBD, though you could always write your own.

[structuredloud]

National instruments multi-sim has a gazillion different models (including opamps that you might actually use) and a decent virtual oscilloscope (it's an HP/agilent clone) as well as the traditional spice/nodal analysis tools. You can easily view the effects of filters and stuff. I can't really see how this would relate to csound or reaktor though.

[tINY]

Spice is a bit weird to get used to - the GUIs help a lot.

There are other simulators out there too, like Micro-CAP.

As already mentioned, breadboarding is a good idea. But, I have left room for rework on PCBAs (sometimes with spare op-amps too) in case things didn't work so great. A field of through holes (or pads for mounting SMT caps and resistors with through holes for wires) is a good way to go if you are only planning on making a dozen or so units.

If you are planning on making hundreds of units, just plan on spinning the board at least once - If it doesn't happen, you have a bit of budget for new tools or a few beers.....



-tINY

[Tibbon]

I'm going to breadboard a good bit first, but I guess I was hoping to be able to try out ideas from my laptop and then try them again back home (on a breadboard).

I should mention that i'm on a Mac, but that doesn't really matter since I have VMWare Fusion running WinXP too.

[Wavebourn]

I am an old school engineer, so I've learned to listen to what I do; I've played with oscilloscopes, synthesizers, stuff I develop, and always listened to results watching them on an oscilloscope, so now I can see, touch schematics, I hear sounds they will produce depending on character of nonlinearities of devices and schematical solutions I am familiar with. So, my advice is, use all sensors you can to learn to touch, hear, see, smell what are you developing, to get better end results instead of [using fashionable resources].

Don't rely on computer programs developed by somebody who in turn rely on data from manufacturers of details, you need your end results sounding as you want, right?

...but if you like to repeat what was done before you it is the best solutions to use computer models/analyzers/blind test results/etc...

I.e., the question is, are you going to be The Master, or an engineer with a good salary who make fashion designs...

BREADBOARD, CHANGE, LISTEN, BREADBOARD, ETC...