Nearly all systems-on-chip (SoCs) these days are mixed-signal, with increasingly complex analog/mixed-signal (AMS) IP blocks. Meanwhile, analog blocks increasingly contain digital control logic. Yet analog and digital design are still done in relative isolation, using very different methodologies and tools. The time has come for a comprehensive methodology for the design and verification of mixed-signal SoCs.
With the publication of the Mixed-Signal Methodology Guide this summer by Cadence, a documented mixed-signal implementation and verification methodology is finally here. This hands-on guide uses a number of practical examples to demonstrate the use of analog behavioral models, mixed-signal verification, RF verification, digitally-assisted analog verification, physical implementation, electrically-aware design, and more. Mixed-signal experts from Cadence and other companies collaborated to develop the content. You can learn more next week at the Cadence booth (#1930) at the Design Automation Conference (DAC 2012), and you can reserve a book copy here.
Much of the book is devoted to verification, and there is a lengthy chapter that outlines a mixed-signal verification methodology. I was particularly interested in this discussion because I've written in the past about the need to bring advanced "digital" techniques such as metric-driven verification, random test generation, assertions and coverage into AMS verification (in this report on a DAC 2011 panel, for instance). The Mixed-Signal Methodology Guide not only shows you why to do this, but more importantly how to adopt and make use of these advanced verification techniques in a mixed-signal setting by leveraging mixed-signal simulation and analog behavioral languages.
The mixed-signal verification chapter provides hands-on information about topics including:
- Mixed-signal simulation as the foundation for verification
- Design partitioning, simulation planning, and regressions
- Assertion-based verification using Verilog-AMS, Property Specification Language (PSL), and SystemVerilog Assertions (SVA)
- Mixed-signal metric-driven verification
- Verifying low power intent in mixed-signal designs
Another chapter provides a detailed discussion of AMS behavioral modeling. It first identifies the types of modeling one might use, including discrete digital modeling, continuous analog modeling, mixed-signal modeling, and real value modeling. It then uses a programmable gain amplifier (PGA) example to show how one would code analog, digital, mixed-signal, and real value models. The chapter concludes with some "best practices" for constructing these types of models.
The book also presents a practical methodology for verifying RF designs. The chapter shows how designers can develop behavioral models for performance analysis in the early stages of a design project, and later develop more accurate models for functional verification. The book discusses both passband models and baseband equivalent models.
A chapter on verifying digitally-assisted analog designs explains why digital control logic is increasingly needed at advanced process nodes. It presents some design examples and shows how to verify them. Another chapter outlines an electrically-aware design methodology for advanced process nodes. This methodology represents a paradigm shift in which electrical analysis and verification move forward in the design flow, providing "in design" verification.
Other chapters include IC/package co-design for mixed-signal systems, mixed-signal physical implementation methodology, and data management for mixed-signal designs. There's also an introduction that presents mixed-signal design trends and challenges. In my view, anyone working with analog/mixed-signal IC design on any level, from IP creation to SoC assembly, can benefit from this book. For more information about the Mixed-Signal Methodology Guide, click here.