Issue link: https://iconnect007.uberflip.com/i/461203
February 2015 • The PCB Design Magazine 41 cycle, material and the amount they will press out the resin. Another area where the variability of PCB manufacturing affects the cou- pon differently than the circuit is etching. Typically, the process of copper plating and etch- ing varies in etchback around the PCB based on density (i.e., denser center areas of the actu- al circuit will plate a bit less and etch a bit finer than the outside edge where the impedance cou- pon is located). Copper plating has the nasty effect of plating areas of less copper density at a higher rate than on more dense areas of the PCB. The thickness of the copper is one of many variables in determining final impedance calculations. The etch rate across the PCB is also affected by circuit density, with typically less dense outside areas where the impedance coupon is located being slightly overetched, and line width is a major factor in impedance numbers. If the impedance coupon is not the correct impedance, does that mean the PCB is bad? Not really, as most PCB manufacturers will do a cross-section and verify the trace width, thick- ness and dielectric spacing, and if they are built as designed, one must assume that the coupon is incorrect. I did a test with one coupon on two identical impedance measuring units. They were over 10% different; one operator had his hand close to the circuit, and one machine was on a stainless steel countertop. But even after correcting the difference in setups and measur- ing techniques, the units were still 6% differ- ent. As the coupon was just over the specified impedance, the PCB manufacturer did a cross- section and everything was within design spec- ifications. The customer was notified of the dif- ference and the subsequent cross-section retest, and the boards worked as planned. Many designers do not know that the im- pedance in a PCB trace varies significantly as it winds its way inside the PCB circuit. Traces close to ground planes, other traces running along the same path—even the thickness of the soldermask all change the impedance at that particular point of the trace. There can be a 2x impedance difference in two close points along a trace, if one point is close to a glass bun- dle and a second point along the same trace is a few thou- sandths of an inch away over a high-resin section. However, the average impedance along the trace will be close to the designed ohms. The balance between what the designer needs in his matching impedance traces and what a PCB shop can deliver is really not that far apart. As a designer, you should be knowledgeable of the limitations and actual usefulness of the coupons we and the PCB shop uses use to obtain the desired impedance. If you have a difficult design, contact your PCB manufacturer. I know that they will gladly run the design criteria through their different calculators to help solve the problem. So there you go, Dan. I hope you caught all that. PCBDESIgn COnTROLLED IMPEDAnCE: A REAL-WORLD LOOK AT THE PCB SIDE continues bob and me If the impedance coupon is not the correct impedance, does that mean the PCB is bad? not really, as most PCB manufacturers will do a cross-section and verify the trace width, thickness and dielectric spacing, and if they are built as designed, one must assume that the coupon is incorrect. " " Bob Tarzwell is a PCB consul- tant who has spent 50 years in the PCB industry, inventing technology and building al- most every type of PCB. he is the co-owner of DB Publishing, publisher of the PCB 101 and Quality 101 handbooks. For more informa- tion, visit www.dmrpcb.com. Dan Beaulieu is a well-known industry consultant and co- owner of DB Publishing. his column It's Only Common Sense appears Monday morn- ings in the I-Connect 007 Daily Newsletter. he can be reached at danbbeaulieu@aol.com.