Electronic-design-automation (EDA) software includes optimizers, which often help RF/microwave designers meet new criteria. For a microwave filter, for example, the optimization case could include goals for in-band insertion loss and return loss, cutoff frequency, and out-of-band rejection. Unfortunately, however, the optimization engine must consider a large number of criteria to create what are often size-limited and rather random solutions. In a four-page application note titled “Improved Circuit Design Flow Using Modelithics Passive Models,” AWR Corp. presents a method to raise design efficiency while keeping an eye toward cost-effective, first-pass design success.
Accurate passive-component models are key to this solution, as they will enable yield analysis via a tolerance setting. The solution combines Modelithics' Global Models or passive resistive-inductive-capacitive (RLC) components with AWR’s Microwave Office software. In doing so, it is able to map individual components’ impact on subsystem performance. A tradeoff can then be made regarding component values and tolerance (with low tolerance being more expensive and high tolerance less so).
A standard low-pass-filter (LPF) design is employed to demonstrate the new approach. After the synthesis of an ideal filter, the ideal components are replaced with Global Models. Transmission-line effects are added eventually as well. To compensate for non-idealities and parasitic effects, an optimization of component values is then performed. The resulting nominal filter design should match the design goals well, although the design can be further optimized.
This flow can be improved by adding tolerance and yield analysis before fabrication. To run statistics with Modelithics Global Models, a tolerance parameter is enabled. It is then assigned to the anticipated manufacturing distribution and yield analysis is performed. As a byproduct of yield analysis, the Microwave Office simulation can be set up to output sensitivity graphs for all of the various parameters. Using those graphs, the user can quickly tell the parameters to which the design performance is or is not sensitive. This information also reveals what parameters can be changed for improved yield. Starting with an ideal design, the note shows that it is possible to achieve a nominal design with good agreement to measurements. That design can then be adjusted for improved yield while accounting for anticipated manufacturing tolerances.
AWR Corp., 1960 E. Grand Ave., Suite 430, El Segundo, CA 90245; (310) 726-3000, www.awrcorp.com.
