Estimated reading time: 2 minutes

Correlation, Communication, Calibration

It is interesting to stand back every now and then and Look at an engineering task from a greater distance. I am also always interested in the extent to which some technologies change, but how the overall concepts do not. At the recent ElectroTest Expo at Bletchley Park, the UK’s WWII center for code breaking, and arguably the home of computing as we know it (though I am sure that the U.S./Germany/Japan/France, etc., have similar sites where such original research took place), and alongside the WWII Enigma code-breaking ma- chines, stands a collection of computers. Many are in working condition, and they take the computing story from vacuum tubes (valves), punched cards and paper tape, through to tablets and ash drives. Indeed, so much has changed, while at the same time, so little progress has been made!

More bits, more speed, more resolution, but the underlying concept, stored programs (called routines in the early days – this is where we get the word subroutine), some working memory and an arithmetic and logic unit. The paper tapes are now ash drives, the magnetic cores are now static RAM and the ALUs comprised of vacuum tubes are now compressed and implemented in silicon on an unimaginably small scale. Also at Bletchley Park last week, the Polar booth was alongside Tektronix, who are celebrating 65 years in the business. Tek had a 1960’s oscilloscope next to their latest models. Despite the passing of decades, the prospective customers still ask exactly the same questions as they did 50 years ago: “What’s the bandwidth? Will it work in my application? How accurate?” Followed by the predictable, “How much does it cost?”

Correlation is a subject that also leads to similar questions, and as production techniques, circuit speeds and the required measurements have become ever more complex and demanding, it gets harder for a production engineer to track down the source of errors in any situation requiring correlation. The PCB world is no exception, equipped with sophisticated modeling tools, often presenting predicted results to many significant places, it can be easy to get lulled into false illusion that “because I can predict it, I can produce it.” However, it is important to keep recalling that although modeling tools may be able to predict a result with many decimal places, that ideal can only be realized when “ideal” materials are used. Richard Attrill, our engineering director, recalls the early days of CAM (sorry, Rich – I am not saying you are old!) when designers suddenly found the ability to specify hole diameters to many more decimal places than the drills or plating processes of the time could ever dream of achieving.

Sometimes a measurement result is quoted with an emotional statement. “Tell me what’s wrong with your model – I can’t get my measurements to agree.” In these situations, step back and look at the materials in use: What is the mechanical tolerance achievable? What is the electrical tolerance?

Read the full article here.

Editor's Note: This article originally appeared in the June 2013 issue of The PCB Design Magazine.