Affordable analog and radio frequency integrated circuits design and optimization

The cost of analog/RF IC design has reached the point that fewer products can afford the long design cycle due to the inability to predict silicon realities. As this problem worsens for technologies below 100nm, the high cost of design and multiple manufacturing spins causes fewer products to have the volume required to support full custom implementation. Affordability is, therefore, crucial for the proliferation of high performance processes.; A novel back-end of line (BEOL) configurable circuit design methodology is proposed to offer reduced design and manufacturing costs, yet provide performance that is consistent with leading edge product requirements. The majority of mask layers are designed to be shared by all applications, therefore they can be optimized for manufacturability and accurately precharacterized in terms of devices and parasitics. A limited set of BEOL masks are then used for application-specific customization. It should be noted that the circuit topologies were optimized to facilitate application-specific configuration using a limited number of BEOL masks.; To demonstrate our methodology we present the design, layout, and test results for a BEOL mask configurable LNA and mixer circuit pair. For this RF front-end design example, only the final two (BEOL) via and three metal layers are used for application-specific customization. The BEOL customization is implemented for three front-end circuits: 1.5GHz GPS, 2.1GHz WCDMA, and 5GHz WLAN. Testing results reveal excellent performance that is comparable to what can be achieved with a full-custom design in the same process.; To formalized this methodology, we further developed an analog circuit optimization framework, ORACLE, which is a combination of reuse and shared-use by formulating the design problem as an optimization with recourse problem. Using a two stage geometric programming with recourse approach, ORACLE solves for both the globally optimal shared and application-specific variables. The ORACLE methodology is applicable to configurable designs in general, but here it is demonstrated for metal-mask configurable designs.; Given that even with shared masks and pre-characterization of more regular structure, there are still large scale process variations for scaled CMOS technologies, we further extend the optimization framework to formulate the analog and RF design with variability problem as a special type of robust optimization problem, namely robust geometric programming. The statistical distribution of the process parameters are captured by the confidence ellipsoid. Using such robust geometric programming with ellipsoidal uncertainty approach, design with a guaranteed yield bound can be achieved. Most importantly, the problem size grows linearly with number of uncertain parameters, and we demonstrate the efficiency for trading off between design costs and yield requirements. Numerical examples reveal significant improvement in the design cost (∼20%) using this approach as compared with corner-based optimization.