X hits on this document

PDF document

The Designer’s Guide Community - page 16 / 31





16 / 31

Principles of Top-Down Mixed-Signal Design

A Rigorous Top-Down Design Process


Simulation and Modeling Plans

An important focus in a good top-down design methodology is the development of a comprehensive verification plan, which in turn leads to the simulation and modeling plans. The process begins by identifying particular areas of concern in the design. Plans are then developed for how each area of concern will be verified. The plans specify how the tests are preformed, and which blocks are at the transistor level during the test. For example, if an area of concern is the loading of one block on another, the plan might specify that one test should include both blocks represented at the transistor level together. For those blocks for which models are used, the effects required to be included in the model are identified for each test. This is the beginning the modeling plan. Typi- cally, many different models will be created for each block.

It is important to resist the temptation to specify and write models that are more compli- cated than necessary. Start with simple models and only model additional effects as needed (and as spelled out in the modeling plan). Also, the emphasis when writing mod- els should be to model the behavior of the block, not its structure. A simple equation that relates the signals on the terminals is preferred to a more complicated model that tries to mimic the internal working of the block. This is counter to the inclination of most designers, whose intimate knowledge of the internal operation of the block usually causes them to write models that are faithful to the architecture of the block, but more complicated than necessary.

It is also not necessary to model the behavior of a circuit block outside its normal oper- ating range. Instead, you can add code in a model that looks for inappropriate situations and reports them. Consider a block that supports only a limited input bias range. It is not necessary to model the behavior of the block when the input bias is outside the desired range if in a properly designed circuit it will never operate in that region. It is sufficient to simply generate a warning that an undesirable situation has occurred.

Following these general rules will result in faster simulations and less time spent writing models. However, the question of how much detail is needed in each model is a delicate one that must be answered with great care. It is important to understand the imperfection in the blocks and how those imperfections affect the overall performance of the system before one can know whether the effects should be included in a model. Also, it is not always true that a pure behavioral model will be superior to a more structurally accurate model. Often making the model more structurally accurate makes it more predictive, and also may make it easier to include some secondary effects due to parasitics.

The simulation plan is applied initially to the high-level description of the system, where it can be quickly debugged. Once validated, it can then be applied to transistor level simulations.

A formal planning process generally results in more efficient and more comprehensive verification, meaning that more flaws are caught early and so there are fewer design iter- ations.

5.2 System-Level Verification

System-level design is generally performed by system engineers. Their goal is to find an algorithm and architecture that implements the required functionality while providing adequate performance at minimum cost. They typically use system-level simulators, such as Simulink [22] or SPW [3], that allow them to explore various algorithms and

16 of 31

The Designer’s Guide Community


Document info
Document views53
Page views53
Page last viewedMon Oct 24 14:34:08 UTC 2016