google.com

jump to navigation

Effects Of Corners In PCB Traces April 6, 2007

A novel SPICE simulation technique was developed to investigate the effects of corners in PCB traces. The PCB traces were modeled by breaking them down into tiny squares or finite elements. A SPICE sub-circuit was made for the finite element and a small square symbol with one pin on each face was also created for it. Many such symbols were then placed in a schematic (like tiles) to graphically and electrically represent a PCB trace. A comparison was then done between a straight trace, one with a 90 degree square corner, and another with a 90 degree corner beveled at 45 degrees. (more…)

Comments

1. james - April 8, 2007

1. Novel to use SPICE perhaps, but representing spatial
elements as LC circuits began in 1944 (see J. R. Whinnery
and S Ramo, “A new approach to the solution of
high-frequency field problems”, Proc. IRE, vol. 32, pp.
284-288, May 1944. and also… G. Kron, “Equivalent circuit
of the field equations of Maxwell” Proc. IRE, vol. 32, pp.
289-299, May 1944). Before computers were widely available,
the LC circuits were actually built and measured! In the
1970s, the LC circuits were replaced by a network of
transmission lines (TLM).

2. I suspect that even for a width/height of four, the
edge-effect is important and a 2D model is a bit
approximate. Also you’d want more than four elements in the
width to represent the concentration of the currents on the
edges.

3. It’d be nice to have the results as return loss in the
frequency domain (for all us RF/microwave people).

Sorry if that all sounded a bit negative! It’s an inventive
use of SPICE. You’ve inspired me to go and do a 3D
simulation to check my believes regarding microstrip bends.

2. Brad - April 8, 2007

Hi James,

Thanks for the thought provoking comments. I had a feeling this would be a hot topic for discussion. Here are responses to your comment:

1. Granted, the underlying math and L-C circuitry is not the novel part. Using “tiles” in a SPICE / schematic capture environment to graphically represent finite elements rather than using a full-blown FEA program is the part that seemed novel to me. Easy access to free SPICE programs is the motivation for using this approach.

2. Yes, a resolution of only 4 is kind of low. I’d hoped to spend some time investigating higher resolutions when I get a chance. As far as current concentrating on the edges, I believe the current will be mainly concentrated on the surface closest to the return path. In the case of microstrips in close proximity to a ground plane, this means the current will tend to be spread on the surface and not very concentrated on the edges.

3. Agreed - return loss would be another good way to look at it.

Let us know what the 3-D results say.

Brad