Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!apple!netcom!mcmahan From: mcmahan@netcom.UUCP (Dave Mc Mahan) Newsgroups: comp.dsp Subject: Re: DSP project Message-ID: <17482@netcom.UUCP> Date: 26 Nov 90 07:00:21 GMT References: <17526@hydra.gatech.EDU> Organization: Dave McMahan @ NetCom Services Lines: 48 In a previous article, gt7865a@prism.gatech.EDU (COMER,MATTHEW BRIAN) writes: >Can anyone give me any practical advice on building my first DSP >project? I would like to build something that would sample at a rate >of at least 41KHz, 16 bits. I have some project building experience >and a pretty good electronics background. Directions to appropriate >literature would be apreciated, as well as any direct advice anyone >might have. I'd suggest doing a meticulous layout on a PC board if you want to use all 16 bits of accuracy. Crummy layout will always kill you with that type of resolution, as noise will creep in from EVERYWHERE! You might even want to try and put the A-to-D on a seperate board that is totally isolated from the rest of the board. By isolated, I mean the whole analog section, starting with a seperate transformer tap for the analog stuff, copper shield box that is soldered to the ground plane, and a liberal sprinkling of .1 ufd ceramic caps on every chip in the system. Noise will kill you if you really need all 16 bits. Just think, if you are sampling a 0 to 1 volt signal, the lsb represents only 15 microvolts. Also, stay away from switching power supplies if you can. Use analog regulators like the 7805, 7812, etc, with both large electrolytic caps and (once again) .1 ufd ceramic caps. Keep your leads to these caps as short as possible by soldering those puppies snug onto the PC board. The digital stuff you can wire-wrap, but do yourself a favor and make the PC-board for the A-to-D converter. If you don't have the resources, use copper clad vector board for the A-to-D section and get somebody with lots of low signal level analog layout experience to show you how to design the vector board and cut pads to maintain as much isolation as possible. Using vector board and then hacking up the groundplane is just a bad idea. In case you haven't figured it out yet, you should get paranoid about noise if you really want to use the full range of the A-to-D converter you have chosen. If you plan to have a D-to-A converter to create an output signal, use the previous tips for that section as well. As for the rest of your project, we might be able to help you more if we knew what you wanted to produce and the types of design tools (compilers, assemblers, and debuggers) you have available. >Comer, Matthew Brian >Georgia Institute of Technology, Atlanta Georgia, 30332 >uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt7865a >Internet: gt7865a@prism.gatech.edu -dave -- Dave McMahan mcmahan@netcom.uucp {apple,amdahl,claris}!netcom!mcmahan