A New Look at Device Modeling

Lately I’ve been working with customers interested in using SystemVision to simulate either a real production design, or some sort of evaluation test circuit. To a customer, however, each ran into the “don’t have a simulation model” challenge for one or more devices in their system. Despite SystemVision’s large VHDL-AMS and SPICE model libraries, we didn’t have a suitable model match. And Internet searches didn’t turn up anything useful. So what is an engineer to do? The answer might be easier than you think.

SystemVision is a powerful and flexible modeling and simulation platform. Part of this power and flexibility is centered in the IEEE standard VHDL-AMS modeling language. Using VHDL-AMS, engineers can create device models of almost any type and complexity. A pretty good deal if you already know VHDL-AMS, but a pretty tough row to hoe if you first need to learn the language before creating any models. Fortunately, there is an easier way — Graphical Component Modeling.

Graphical Component Modeling (GCM) is a technique for building device models from simple, low-level building blocks. It’s kind of like having a modeling Erector Set: you have a box full of modeling pieces from which you can build an almost limitless variety of useful simulation models. With a little creative imagination and a bit of engineering inspiration, you can use GCM techniques to develop some pretty amazing models without writing a single line of VHDL-AMS code. In a way, GCM is like creating SPICE macromodels or Simulink control algorithm models, but with a couple of key differences.

First, GCM isn’t stuck modeling in a single discipline (think about the electrical-only limitations of SPICE). With SystemVision’s building-block library, it’s easy to create models in a variety of technologies, including models where the input technology is different from the output technology. Consider a pressure sensor, for example, where the input is a pressure, and the output might be a voltage or current. GCM simplifies device modeling, giving engineers a simulation toolbox full of building blocks for single and mixed-technology device modeling. Pretty cool stuff.

Second, GCM simplifies conserved system modeling using simple real number processing based on a library of math function building blocks. This is similar to control system-level signal-flow modeling for which Simulink is famous, but GCM also accounts for energy conservation in the device – not a common feature in signal-flow type models. How does this work? The answer is in SystemVision’s ability to efficiently and effectively model both conserved and signal-flow systems, which GCM uses to model physical behavior. The technique is simple: use signal-flow building blocks to model device functions, then wrap the signal-flow model in conserved building blocks to ensure energy conservation and accurate modeling of physical behavior.

Graphical Component Modeling truly is a powerful technique for developing sophisticated device models without the overhead of knowing (or learning) a new modeling language. In future posts I’ll introduce a few GCM-based model examples to illustrate GCM’s power and flexibility. Stay tuned…

In the mean time, post a comment and let me know what analog/mixed-signal/mixed-technology modeling challenges you see in your design work. GCM just might be a viable solution.