Issue link: https://iconnect007.uberflip.com/i/1085192
FEBRUARY 2019 I DESIGN007 MAGAZINE 43 weight on the plane layer due to having to go to a single-ply prepreg construction. As evidence of this for internal planes larger than a half ounce or three-eighths of an ounce, fabricators like to use a minimum of two plies of prepreg. Bear in mind that even the thinnest preg ply used as two-ply may be too thick to deal with the 0.004" or below trace widths called out on the surface for impedances requiring a re- layout to increase the space on the surface and the trace widths, and still meet a two- ply construction to keep your planes at one ounce • Material type to get closer to the Dk you were modeling the trace and space at so no changes or very few changes need to be made to the trace widths, which may also require a new layout • Copper weight reductions if a given trace has to be increased, digging into your available usable space These are really the only things a fabricator can do to get close to your desired impedances on their end without requesting a new design with wider spaces, material deviations, and the overall thickness and tolerance requirements. But what are some things a designer can do to help the fabricator if those changes are needed? One thing is to not use the same trace width for different scenarios on the part (i.e., using a 0.004" trace for single-sided at 50 ohms, and part of a 100- or 90-ohm differential pair). Dif- ferentiating them by a tenth or even a thou- sandth of a mil will allow the fabricator to uniquely pick out those structures that need to resized and not affect the others of the same width. This has been a practice known to de- signers for years. Additionally, using the same trace width for copper fill area as the imped- ance traces is not a good idea because if one needs to change to meet the desired impedanc- es, the fill area will also be affected. Now, let's talk about mask color changes from what was initially negotiated based on the same 0.004"/0.004" design rules for trace and space. If the original request calls for red sol- der mask and the part dataset is received with it changed to green mask, red mask is thicker. Thus, impedances are based on the red mask are calculated, so changing to a green mask may mean the 0.004" traces would have to in- crease due to the thinner mask color. Even as little as a half-mil change may throw out your ability to increase the trace as a true 0.004" trace and space design. Further, going with a lighter copper weight to start to not have to re-lay out the design with wider space will not work because reducing the copper weight means increasing the trace width to get the de- sired impedance. What is the bottom line of this column? Simi- lar to ISO—where we must do what we say and do what we do—as a board designer, you need to make sure all changes (e.g., material type, trace widths or spaces, mask colors, or even surface finish) from what was original- ly negotiated with the fabrication house be- fore asking them to calculate impedances not to have to re-layout the design. Save yourself some grief by having a conversation with the fabricator about all of the variables so you will not be asked to re-lay out the design, taking additional time. As always, thank you for reading. DESIGN007 Mark Thompson is in engineering support at Prototron Circuits. To read past columns or contact Thompson, click here. To download your copy of Thompson's eBook, The Printed Circuit Designer's Guide to… Producing the Perfect Data Package, click here. Even as little as a half-mil change may throw out your ability to increase the trace as a true 0.004" trace and space design.