Gather together the data sheets for the parts you wish to use. Sketch up a diagram of your system on a few sheets of A3 paper. NOTE The Microprocessor may have pins that are dedicated to functions that you need, or have pins that are easier to use from the point of view of the compiler, select your pins with care. Join up the parts with the logic lines you are going to use, adding notes as well. Remember the parts are not likely to be in PADS, although the footprints may be. Keep a lab book, note your design philosophy and subsequent design, parts, any quirks to watch out for, PADS names for parts entered, etc. Put your A3 sheets into your lab book.
Add the parts to the PADS data base, be careful to get the pins assigned and numbered correctly in LOGIC. The footprints may be in LAYOUT, check that you get the correct ones by going to the footprint for the selected parts in the data sheets. If not there, add them, this can sometimes be done by modifying an existing part and renaming it.
I cannot emphasis strongly enough that the key to getting you board out the first time (neglecting logic design errors) is to be 100% sure that the entries into LOGIC and LAYOUT of your parts is correct, double and triple check!
After that, get your design into LOGIC, do your routing. Place components on your board carefully, connectors on the edge normally, and high speed lines short. Decouple power at IC pins. If you auto route you will not get the best layout, for a board of this nature the push and shove router is good. Power and ground planes in the middle layers, signal routing top and bottom. This allows for repair of the board if there are mistakes. Put a ground plane top and bottom as well. You have to know manufacturing guide lines in deciding line widths, via sizes etc, an art in itself.
There is so much more that could be said, but not enough space, and you have to learn yourself anyway. Other people will have a different approach, this just happens to be a brief overview of how I do it in a commercial environment,
cheers,
Richard
