|
What Is your principal job function as a consultant?
My principal function is to enable designers to utilize nonlinear transistor models with circuit simulators, such as Advanced Design System, for the development of MMIC and hybrid circuits.
Sometimes this merely involves providing parameter extraction to a standard transistor model and other times it involves customization of the model to more accurately reflect the behavior of the customer's devices. A few years ago, the effort to improve circuit performance was driven by military needs. Now, competitive commercial products are the driving force.
How important is the transistor model to the success of the circuit design?
The most critical elements in the performance of transistor circuits are the transistors themselves. Designers might have their own transistor fabrication facility or use an external foundry. In either case, they need an accurate model to develop the best circuit design for a specific application.
The "best" RFI/MMIC design means superior performance characteristics as well as high yield and low cost.
Why is it difficult to develop a good model?
One major difficulty is the lack of self-consistent device data. This may occur because a designer takes DC I/V data on one group of transistors and S-parameters on a second group. To develop accurate models, one really wants both DC and RF data on the same devices. Another issue is device variation within the same process line. One may need a model for the average device and models for limiting cases.
A particular foundry line of transistors may not behave very much like any standard model. Or, they may have DC behavior similar to a standard model but the RF behavior is quite different.
In that case, some customization of the standard model is required. A third area of concern is the effect of heating upon the performance and the characterization of the transistors. For power amplifier applications, the self-heating effects can significantly alter the efficiency of the DC to RF conversion.
|
How is Agilent EEsof's IC-CAP helpful?
IC-CAP enables the user to collect ED and RF data in a systematic fashion on each device tested. This provides consistency between ED and RF data. It can operate the tests to be pulsed or CW (continuous wave), enabling the separation of voltage-dependent effects and self-heating effects. The DC and RF testing is flexible and even lets users configure specific tests.
The parameter extraction facility of IC-CAP is equally flexible. One can extract to specific standard models, such as the Gummel-Poon, or to new models for which equations are user-defined. Optimization of the "fit" between data and model is available.
The statistical package within IC-CAP permits selection of the average model and models for the limiting cases for Gaussian or non-Gaussian distributed statistics. This is called non-parameter analysis in IC-CAP. In the case of normally distributed parameter values, IC-CAP produces standard statistics, such as mean values, standard deviations, and correction coefficients.
What do you like best about IC-CAP?
Perhaps the best overall feature of IC-CAP is its flexibility. The user is in no way limited by the standard models and may develop unique testing and parameter extraction routines for new, customized, nonlinear models. The PEL language is similar to HP BASIC and is relatively simple to program.
Is IC-CAP difficult to learn?
As with all software, there is a learning curve. However, since the release of IC-CAP 5.0, the software environment has a simplified, more intuitive user interface that greatly facilitates ease of use and designers' productivity.
Because IC-CAP is easy to use, it saves tremendous amounts of engineering time when developing nonlinear models for specific applications. In many cases, what used to take several days now takes only hours to do.
|
Can you summarize your evaluation of IC-CAP?
IC-CAP is the most useful software package that I have seen in many years. Its capabilities, flexibility, and speed have made my work far more efficient. In a company environment, I believe that an investment in IC-CAP soon returns a profit in significantly reduced engineering time that used to be spent testing transistors, programming computer control of equipment, and parameter extraction routines.
|