Issue link: https://iconnect007.uberflip.com/i/509724
26 The PCB Design Magazine • May 2015 which involves inserting a precision piston into the bore and measuring the resistance to air passing through the piston into the bore. It's a science in its own right and an interesting sub- ject if you happen to be into mechanical me- trology. Now, back on track with impedance. The availability of air line standards allows measure- ment systems manufacturers to prove accuracy traced back to a national standard. Air lines are commonly available in impedances of 50, 75, and 100 ohms, and, more rarely, 28 ohm air lines. Why 28 ohms? Well, this was the imped- ance of Rambus RAM data lines, and back in the day when that technology was introduced the impedance requirements were demanding and led to 28 ohm air lines being developed as a standard for that application. A side benefit of the availability of 28 ohm lines is the ability to traceably calibrate a TDR over a very broad range of commonly used impedances. At a low- er cost, but also providing an economic alterna- tive to air lines, semi-rigid precision coax can be used as a transfer standard and has the benefit of robustness if used in a production environ- ment. Expectations Newcomers to impedance control are some- times tempted to tie down specifications more tightly than is practical. Perhaps this arises be- cause the units of lossless impedance are ohms, and it is customary to expect that something specified in ohms (resistance) can be specified very tightly indeed. However, you have to re- member that impedance is in fact a high-fre- quency characteristic of a transmission line, and in general the higher the frequency gets the less realistic it is to think you can specify down to fractions of an ohm or less. In fact, the traceable air line standards mentioned earlier in this ar- ticle may have an uncertainty of 0.2–0.5 ohms, and that's for a reference line possibly costing many hundreds of dollars. So you would not expect a line based on the finest geometry pos- sible on an FR-4 board to come even close to those levels. Often impedance is ±10%; tighter specs may demand ±5% and some critical appli- cations may be more tolerant of higher or lower impedances—so you may encounter specs of say 50 ohms +7 ohms – 4 ohms, for example. Test coupons As frequency increases the test coupon (sometimes called the test vehicle) becomes an increasingly important part of the measurement system as, coincidentally, do the probes and the interconnect cables to the measurement sys- tem. Careful design of the coupon ensures that artefacts which could spoil the measurement quality are minimised. Launch pad diameter is a compromise be- tween ease of probing and launch aberrations; having a coupon that is long enough (IPC still specifies 6"/150 mm) gives an adequate sample of the trace to measure over a reasonable length of PCB; after all, the impedance test is meant to measure the consistency as well as the abso- lute value of the fabricated impedance. Also, coupons should be designed to ensure that any nomenclature in the copper is well away from the trace under test and that the transmission lines are back away from the coupon edges to avoid any effects on impedance from the prox- imity of the edge of the ground plane. Like- wise, on the coupon power and ground planes should be interconnected—but not on the PCB itself! Modelling Coaxial standards can be calculated with precise closed loop equations. However, PCB transmission lines are far from uniform cylin- drical coaxial structures. At their simplest they could be regarded as rectangular traces over in- finite planes, and whilst approximations do ex- ist for calculating the basic variations of such structures, they are limited when pushed to modern trace geometries, which leaves you re- quiring a field solver if you wish to make ac- IMPEDANCE CONTROL, REVISITED continues Figure 1: Traceable 50 ohm air line standard with sMa adaptor. the pulse