Issue link: https://iconnect007.uberflip.com/i/1163814
86 DESIGN007 MAGAZINE I SEPTEMBER 2019 ish is greater for a GCPW circuit than for a mi- crostrip circuit. In this case, the comparison is between a tightly coupled GCPW circuit (Fig- ure 5b) and a microstrip circuit (Figure 5a). If the comparison had been between a loose- ly coupled GCPW circuit and microstrip, the results would have been somewhere between the microstrip and GCPW responses shown in Figure 5. A very loosely coupled GCPW circuit can behave very much like a microstrip circuit in terms of loss behavior. Trying to account for the losses of a final plated finish as a function of frequency can be rather difficult because many loss mechanisms for these circuits are frequency-dependent as is the loss impact of the final plated finish. One frequency related issue is skin depth and that is how much of the conductor will be used by the RF current at a given frequency. When the fre- quency increases, the skin depth will get thin- ner, and that will naturally cause more con- ductor loss. Skin depth is also impacted by the conductivity of the metal conductor. Copper has excellent conductivity, but most final plat- ed finishes have less conductivity than copper. The following function gives a reference to the skin depth formula and the conductivity of dif- ferent metals: As the skin depth formula suggests, an in- crease in frequency results in a decrease in skin depth. A decrease in conductivity will cause an increase in skin depth. Finally, an in- crease in permeability results in a decrease in skin depth. The effects of a lossy plated finish on the composite conductivity of the sidewalls of a signal conductor can be difficult to estimate, especially considering changes with increased frequency and how skin depth decreases with frequency. But it can be remembered that at lower frequencies (less than 500 MHz), the composite conductivity at the sidewalls of a signal conductor is a combination of copper- nickel-gold. As frequency increases, the skin depth will decrease, and the composite con- ductivity at the sidewalls will be a combina- tion of nickel-gold. At much higher frequencies with very thin skin depth, the composite con- ductivity will be dominated by the gold. So far, the plated finishes for PCBs have been referred to as "lossy plated finishes" because they in- crease the loss of a copper conductor beyond its unplated performance. A plated finish that would not be considered lossy would be one that does not increase the loss of a copper con- ductor, such as immersion silver (with conduc- tivity higher than copper). For ENIG, nickel exhibits about one-quar- ter the conductivity of copper; since it is less conductive than copper, it will suffer greater conductor losses. The presence of nickel can cause a doubling or even a tripling effect on conductor losses. First, the nickel will cause more conductor loss due to its conductivity be- ing less than copper. Second, with increased skin depth in nickel, the RF current will use more nickel, resulting in greater loss. A third factor has to do with the ferromagnetic nature of nickel and how it will normally suffer some amount of magnetic loss in combination with the other two loss components. The potential ferromagnetic effects of nickel are difficult to quantify. In general, ferromag- netic properties change dramatically with fre- quency, from lower microwave frequencies to a few GHz. The higher relative permeability (μ r ) of nickel will result in some decrease in skin depth, somewhat offsetting the increased skin depth of nickel due to poor conductivity. In addition to these complications, the nickel used in ENIG is not pure nickel but is typical- ly doped with phosphorous. Suppliers of ENIG will adjust different characteristics of the nick- el alloy for different reasons. Due to the many issues associated with the magnetic properties of nickel, it can make a correlation between models (for simulation) and measurements less accurate.