However, my position is thta on should endeavor to become expert before going to PCB. This is achievable only by some form of hand-wiring, and by carefully examining the effects of certain construction techniques. Wire-wrapping works fine for digital hardware. I've not had problems with it in analog cases because I've never built precise analog circuits with wire-wrap.

I've built them with careful, painstaking, point-to-point wiring directly over a fixed ground plane and, in some cases, with GND wires on either side, in intimate contact (insulation to insulation) and glued together with, the signal wire. Longer leads, 10 CM or more were done with coaxial cable or shielded twinax if frequency or sensitivity was high.

When a newbie learns the benefits of such techniques, he figures out how to implement them in PCB as well. If his only contact with the problems one can unintentionally build into a PCB is on a PCB, he's never going to learn to avoid them because they're nearly impossible to fix on a multilayer board.

I've seen men work for hours on a PCB, using spectrum analyzer and metal dividers to help find improvements in grounding and bypass. I've seen them soldering thick wires on the supply lines to help with supply distribution issues. If you've got to hold a 10 mV power-gnd noise ceiling, there's no end to what you'll have to do.

Of course, most newbies, just consider Mike for example, probably don't even have a DMM. They need an oscilloscope, and, in fact, one of substantial bandwidth and sensitivity in order to see the benefit of these techniques.

You're probably right about the amount of radiation from wire-wrapped circuits. However, generally, I'm more concerned about what gets in than I am about what gets out. I'm sure you understand that, particularly since most of my stuff's in an industrial environment.

One thing anyone should construct is a continuity tester. By that, I mean a circuit that squeeks or buzzes or rings when there's a voltage offset of somewhat less than 0.3 volts between the two probes ... something that won't forward bias a silicon diode or turn on a transistor. The ones built into many DMM's present too high a voltage to the circuit under test. That's a discussion for another day, though.

How can people be expected to design PCB's when they don't have experience, don't have test equipment, don't have proper materials? Having good software won't do any more than make them generate the same rubbish they'd generate without it. Building the rubbish on 4-layer boards won't make them better than on 2-layers, though. It just makes them more costly.

On the other hand, if they get good prototyping boards, they can wire-wrap or hand-solder their circuits and apply proper techniques. I've already described how one can build a multilayer sandwich if it's really critical.

Multilayer boards can be made to work VERY well, but, for the inexperienced, they're very hard to modifiy for the better.

The following might provide some interesting information regarding circuit board layout.

http://focus.ti.com/lit/an/slva232/slva232.pdf
http://focus.ti.com/lit/an/slua392/slua392.pdf
http://focus.ti.com/lit/an/sdya011/sdya011.pdf
http://focus.ti.com/lit/ml/sloa089/sloa089.pdf

RE