[sdiy] Question:

[thescum at surfree.com]

On Wed, 13 March 2002, Sowa Roman wrote:
> When something goes wrong in analog, you hear something.
> When the same happens in DSP, you need to spend hours
> with emulator to see where it hungs up.

You may hear something in analog, but you might also still spend a lot of time with your diagnistic tools to find out what's wrong.  Especially if it's a new design that's never made any noise before.

In the end, I think that a lot of debugging (in either hardware or software) comes back to your overall understanding of the design.  You'll understand it's weaknesses better, and have better ideas of where to look when something goes wrong.  Folks like Knuth on the software side and Pease on the hardware side try to make this apparent.  All too often, we just copy some other design without really understanding how it works.

Better debugging tools can help a lot too.  An environment that lets you contidionally execute your source code on the actual hardware (like a JTAG or ICE environment), and a processor that supports breaking & tracing are easier to debug than "closed box" hardware and a standalone simulator.  The same might be said for a high bandwidth multi-channel sampling oscilloscope versus a simple DVM.

Byron Jacquot