Excellent article. The first thing I thought of when reading your post was the care needed when programming. I learned to program using PASCAL--a teaching language. My instructor was absolutely paranoid regarding students applying proper and copious notes to code so transparency could be obtained. In his way of thinking, documentation was critical to understanding and remembering the "why" of lines and lines of code. Your design for testability certainly falls along the very same lines. I am completing a project right now in which sensor ports are located strictly for design confirmation and pre-pilot testing. I have no idea if we will use all of the ports but they are there. For production, we will lessen the number but still provide ability for trouble- shooting and field service. Again--excellent post and a good reminder that considerable time can be saved if proper planning, including DFT, is considered.
Poor quality solder can certainly bring about the failure of an electronic system, there is no question about that at all. And not just the Xbox. I repaired a Kenwood stereo amplifier that died completely because of poor solder in the power supply area. It is not just the fact of a bad alloy mix, sometimes the wave solder machine is just a bit off, or the solder had excessive scum, or the pre-fluxing was inadequate, or possibly the bare circuit boards had excesive oxidation on the surface. Or maybe the temperature was a bit too high. Lots of different things can lead to poor board soldering, and inadequate inspections won't see the failures, and if they pass initial testing they won't be fixed, they will fail in a few months, and be out of warranty, so why should they care?
Now about line conditioners? Is that power line conditioners or audio line conditioners. If you have a noisy power line a good inlin filter may be a lot of help, if it is installed in a manner that allows it to prevent the noise from passing through. And a good filter is much cheaper than one of those good line conditioners like the ones that5 SOLA makes.
All of the systems that I have designed have been designed not only to be testable, but also to be repairable. That goes way beyond just putting an adequate number of accessable test points on the PCBs, it also includes designing a package so that it can be opened for servicing after the product is built. Those wonderful snap assemblies are not so very wonderful a few months later when the plastic parts have become less ductile and much more brittle. That is what is never mentioned in those "design for assembly" classes. I often add the claim that "our products are worth repairing, while other products are certaainly worth recycling." That imessage, that our products will have a much longer lifetime because they are repairable does make some sales, especially when it is along side a long warranty period.
The article was good, and the points well made, but don't forget that there is more to many products than just a circuit board. And the rest of the product should be testable as well.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.