Determining tradeoffs among various laminates.

Should you use wider trace widths? And how will you know?

An engineer recently asked me about the relationship between trace width and insertion loss, while adjusting dielectric height to maintain a 50Ω single-ended impedance.

At a high level, five variables are at work here, including trace width, copper weight, dielectric height, Dk and Df. Include frequency and resin content, and we’re really talking about seven variables. (Then there are stripline vs. microstrip configurations, which change things a bit, as well as percent-copper, which impacts prepreg thickness, and copper roughness.) We’ll keep things simple, for discussion’s sake, and address some of these factors in future columns.

Lossy transmission-line effects become significant signal integrity concerns at clock frequencies above roughly 1GHz and for interconnect lengths that exceed 12 inches. Assuming we’re talking striplines and grabbing data from a stackup I was looking at [Megtron 6 (G)], we’ll use an insertion-loss comparison between two different trace width/dielectric height combinations. (Dks are similar, but a little bit different, depending on resin content.)