Issue link: https://iconnect007.uberflip.com/i/1163814
90 DESIGN007 MAGAZINE I SEPTEMBER 2019 tween the bare copper circuit and the EPIG fin- ish is still noteworthy (Figure 10). Since the test vehicle was using a slightly higher loss material and the copper surface was slightly rougher, the bare copper circuit loss in this study of EPIG finish is higher than for the bare copper shown in previous studies in this article. The copper was a low-profile, reverse-treated copper with minimal surface roughness variation. The insertion loss differ - ence between the bare copper circuit and the EPIG circuit has the least difference of any other gold finishes shown in this study. Even though this article has focused on in- sertion loss, it seemed judicious to mention a slightly different issue which can be important in material characterization and is related to the plated finish study. Microstrip ring resona- tors are commonly used to characterize mate- rials at microwave and sometimes millimeter- wave frequencies; however, if ENIG is applied, the nickel variation can have an impact on the capacitance in the gap coupled area of a ring resonator. An evaluation was done, in paral- lel to the evaluation shown in Figures 7 and 8, where gap coupled ring resonators were evalu- ated using the same sheet of laminate to make the circuits, and it was found that nickel thick- ness variation does impact the resonant fre- quency of a ring resonator. This means if this structure was used to calculate the Dk of the material, which is common, the extracted Dk would be altered due to nickel thickness vari- ation. A summary of this information can be seen in Figure 11. The materials used to make the microstrip ring resonators were made from the same large sheet of material cut in half. One half panel had circuits made with thinner nickel for the ENIG, and the other half panel had circuits made with thicker nickel for ENIG. The nick- el variation makes some small differences in the loss, Q, and bandwidth, but makes a pretty significant difference in the center frequency and the extracted Dk. By using the same large sheet of material, that will minimize material differences there is yet an extracted Dk differ- ence of 0.023. The ring resonators having dif- Figure 10: These curves show the insertion loss of microstrip circuits using a gold final plated finish which does not have nickel.