Issue link: https://iconnect007.uberflip.com/i/275965
28 The PCB Design Magazine • March 2014 Those of us in the industry for long enough can recall the move from analog voltmeters to digital and from vernier scale to digitally en- coded micrometers and calipers. Digressing slightly, I recall a very analog moment when, as a naïve 11-year-old, I and a school friend decid- ed to measure the current available from mains electricity—with an analog AVO meter (safety note: not a wise move!). The result was a large blue flash, melted me- ter cables and a visit from the local electricity authority as we had vaporised the main fuse in the supply box. Fortunately, we are both still here, and I still use a trusty old analog AVO me- ter for electrical jobs around the house. Perhaps that's why I ended up with a career in the elec- tronics industry, but I have to say it's not as ex- citing as that first experience, and maybe that's not a bad thing. Back to tolerance. In a supply chain, prob- lems set in when tolerances are set incorrectly or when the tolerance specified is unachievable. The story in the PCB industry dates back to the introduction of CAM and the ability to set hole tolerances far tighter than those achievable with the available drills or drilling machines. It is important that everyone in the supply chain understands what is achievable and what is not. The same is true for electrical measurement. It also comes as no surprise that a long, complex supply chain also has a lot to take on board as each particular measurement can present indi- vidual challenges. Take electrical measurements: I have de- scribed how the nature of electrical measure- ment changes as frequency increases. The same is true for tolerance. When measuring at low frequencies, current, voltage and resistance are all capable of being measured with a high de- gree of precision with many decimal places of accuracy and tolerance on measurements can be tight indeed. But as the frequency starts to creep up it becomes harder to hold to such tight tol- erances. At DC, tenths or hundreds of an ohm present little problem; but at mid frequencies, say, from 100MHz to 2GHz, then measurement of impedance presents more of a challenge. Per- haps a tolerance of 0.2 or 0.5 ohm is the best you can expect, and that depends on the im- pedance measured. Even the precision imped- ance air lines made traceable via NIST or NPL can present the user with a surprising amount of impedance uncertainty, limiting the achiev- able tolerance and absolute accuracy of mea- surement systems. At even higher frequencies, resistance, voltage and current become even more tricky to measure, and RF engineers resort to power measurement, where the heating effect of the RF power is really the only way of measuring the gain or attenuation of a system, the volt- ages and currents being too elusive to measure. TOLERANT OF TOLERANCE? continues feature Figure 1: Misunderstanding tolerance can lead to disappointment.