Issue link: https://iconnect007.uberflip.com/i/1163814
88 DESIGN007 MAGAZINE I SEPTEMBER 2019 frequencies where the composite conductivity of ENEPIG and ENIG are basically the same due to the skin depth using about the same amount of copper-nickel-gold, or at slightly higher frequencies, the same thickness of nickel-gold. Once the frequency increases to the point where the nickel is contributing less for ENEPIG, the nickel will still be a significant factor for the ENIG because ENIG has thicker nickel. As Figure 6 shows, finishes with ENEPIG and ENIG both have more loss than a finish with immersion tin. Reviewing the metal conductiv- ities presented earlier, it would be expected that immersion tin would have greater loss than these other finishes. The insertion-loss relation- ship between ENEPIG and ENIG has been con- firmed by other studies; it has been assumed that the added magnetic losses of nickel used with ENEPIG and ENIG finishes makes them lossier than an immersion-tin finish. However, it is possible that the loss differences are relat- ed to thickness and/or skin-depth effects; im- mersion tin is extremely thin and skin-depth effects would not be significant until much higher frequencies. Microstrip circuits have other components of loss which result in increased loss with in- creasing frequency, and these loss components can make it difficult to separate the effects of immersion tin on loss. Some of the loss com- ponents not related to plated finish include the fact that the Df of the substrate material in- creases with increasing frequency and the radi- ation loss increases and the electric fields con- dense more at higher frequencies. Condensed fields will cause a narrower ground return path which increases the conductor loss, adding to the total loss behavior of the circuit at higher frequencies. The impact of variations within the plated finish on loss must also be considered. The nickel layers used in ENIG finishes can suffer large circuit-to-circuit thickness variations: it is possible for the nickel layer to vary from 50 to 250 μin. (1.27 to 6.35 μm). A study target- ed finishes with low and high nickel thickness and everything else remaining the same. The test vehicle remained the same for this study, but a different material was used with low loss and rolled copper. The material was a ce- ramic-filled PTFE laminate with a Dk of 3.0, Df of 0.001, and copper surface roughness of 0.35 μm. The microstrip insertion loss curves shown in Figure 7 depict significant differences in in- sertion loss with variations in nickel thickness that is within the thickness range which could occur from one circuit build to another. ENIG is typically not used above 60 GHz; however, as can be seen in the table of Figure 7, there are significant differences in loss at lower fre- quencies as well. As designers transition from applications op- erating at microwave frequencies to circuits at higher, millimeter-wave frequencies, they of- ten consider the use of GCPW to replace mi- crostrip transmission lines. The benefits of us- ing GCPW at millimeter-wave frequencies in- clude less dispersion, less radiation, and the possibility to suppress spurious wave propa- gation modes more effectively. However, there are several PCB-related issues which impact the consistency of RF performance for GCPW Figure 7: These insertion loss curves compare microstrip circuits with bare copper to microstrip circuits with ENIG finishes with different nickel thicknesses.