The wireless revolution has changed us into fast-food information addicts. Methods of delivering insight and software knowledge to customers have changed, and in response we’ve pioneered AWR TV™ which we introduced in January. An always-on, always accessible site, AWR TV enables you to watch what you want, where you want, when you want, and to get the information you seek through short and concise video demonstrations.

Continuing AWR’s commitment to partnering with best-in-class tool providers to give our customers the best possible design flow, we announced AWR Connected™ for Mentor Graphics last September and changed the way high-frequency PCB co-design flows work by obsoleting file translation. Now we’ve announced AWR Connected for Anritsu’s VectorStar VNA and have changed the way design and test domains interface. We’ve made Microwave Office® software a standard component of Anritsu’s new VectorStar™ MS4640A vector network analyzer, omitting the leg work (literally) of taking data from the VNA to the Microwave Office design environment and vice versa, and making it the first microwave instrument in the world to physically provide a full suite of design tools within its firmware.

There is no denying that change is all around us these days. At AWR we are embracing change for as much as we can in the affirmative - by keeping our eyes focused on you, our customer, and reminding ourselves daily that our corporate mission is simply to make you more successful by providing you with tools that dramatically increase the productivity of your GHz electronic product development. By working along side our customers and partners, we can make a difference. Isn't that change worth believing in?

When it comes to jogging, I normally recommend a crawl-walk-run approach, so let’s try that here and see where it gets us. You’re experiencing one of those errors that is more common than you think because microstrip circuit models are so powerful and easy to use that we often forget the assumptions under which they are valid. First off, we assume that we have a good ground that is at least 3-5x wider than the width of the microstrip line—maybe someone will ask Tech Talk about that someday. Secondly, the microstrip models themselves are derived assuming good TEM quasi-static propagation. So let’s delve into that a little bit.

Microstrip is not what we have in mind when we want to go and analytically solve Maxwell’s equations. The “easy” solutions to Maxwell’s equations in rectangular coordinates normally deal with infinite or semi-infinite planes of material and material discontinuities in perhaps one dimension in cross-section. These simplifications provide us with solutions that show TEM plane-wave propagation. With microstrip, however, we have a conductor of finite width and thickness sitting on top of one dielectric and embedded in another, which gives a solution that involves all sorts of modes, but at frequencies low enough relative to the geometry of the microstrip that we can talk about approximate, or quasi-static, TEM propagation as the dominant mode.

The fallout from this for you, my friend, is that whenever we have a discontinuity, like a bend, the quasi-TEM approximation breaks down, and, while it is encapsulated within the MBEND or MBEND90X model for the discontinuity, the model nonetheless assumes that a good quasi-TEM signal is presented at the port(s) of the model. So you need to have the MLIN (or equivalent) on the “output” of the bend be appropriately long enough, where the appropriate length is at the very least a function of frequency.

It is easy to visualize what’s going on in your design by setting up a current animation using AXIEM (View "Effective Design – Microstrip Bend" Video on AWR TV). With the two bends so close to each other, the propagation coming out of the one bend does not have enough distance to re-establish itself in a completely good quasi-TEM mode, hence the propagating current wavefront is on an angle. The conditions under which the bend model is valid are therefore violated for the second bend and your simulation does not match your AXIEM or measured results.

So next time you go for a jog, remember “crawl-walk-run” and make sure you pick an appropriate distance.