Issue link: https://iconnect007.uberflip.com/i/1163814
SEPTEMBER 2019 I DESIGN007 MAGAZINE 87 Measured Results and Discussion I have conducted several studies on plated finishes over the past few years with the coop- eration of two ENIG suppliers and a PCB fab- ricator. Before reviewing some of the measure- ment results of those studies, it is helpful to detail the test vehicle used for the measure- ments. The preferred test vehicle consisted of a 50-Ω microstrip transmission-line circuit on a selected circuit material as the device under test (DUT) with measurements made per the differential length method [3] . The microstrip circuits were fabricated on thin, low-loss substrates with smooth copper. These material characteristics help minimize dielectric losses while exaggerating any con - ductor loss differences among different plated finishes used on the DUTs. It should be noted that all copper foils used in the PCB industry have some normal surface roughness variations which result in variations in conductor loss - es from one circuit to another when using the same circuit laminate. Copper surface rough- ness and its variation will impact insertion loss and the phase response of a circuit [4] . By us- ing smooth rolled copper, the surface roughness variations were minimized as much as possible. Microstrip circuits were used in the measure- ments instead of GCPW even though GCPW might have been thought to be more sensitive to loss differences among circuits with different plated finish- es. But studies have shown [1, 5] that GCPW suffers more performance variability due to PCB fabrication processing and normal copper plat- ing thickness variations, and con- ductor trapezoidal effects can cause significant loss variations from cir- cuit-to-circuit when using the same substrate, making it more difficult to separate the effects of final plat- ed finishes from the fabricated cir- cuits. Microstrip circuits are less impacted by PCB fabrication vari- ables, making microstrip a more suitable choice of transmission-line format to study the effects of plated finishes on circuit loss. Initial studies were performed using 50-Ω microstrip transmission-line circuits based on 5-mil, ceramic-filled PTFE laminate with a Dk of 2.94, Df of 0.0012, and rolled copper with an average surface roughness of 0.35-μm RMS. Multiple plated finishes were used with a summary of the insertion-loss curves for the different finishes provided in Figure 6. The reference (top) curve in Figure 6 is for a microstrip circuit with bare copper. As can be seen, the loss curves for the OSP and immer- sion silver finish are approximately equivalent to the circuit with bare copper. Other studies have shown similar results, where OSP and immersion silver do not significantly influ - ence the insertion loss. The ENIG plated finish shows the greatest amount of loss, and these results have been confirmed in other studies of this nature. The electroless nickel electro - less palladium immersion gold (ENEPIG) plat- ed finish has less loss at higher frequencies; however, at lower frequencies, this finish and ENIG exhibit similar effects on loss. Other studies have confirmed this trend. A plausible reason for the frequency-loss relationship is that ENEPIG finishes use thinner nickel than ENIG finishes. Since conductivity is a composite effect and changes with frequency, there is a range of Figure 6: Insertion loss curves of multiple final plated finishes using the microstrip differential length method.