Non-linear circuit modelling

[Duncan_Widman]

Hi - I work at Applied Wave Research (we make MWOffice), and I wanted to clear up some things I saw in this thread:

- We no longer support Volterra simulation in version 6. Volterra series analysis was introduced to handle simulations that caused trouble with standard Harmonic Balance simulators, like amplifiers with weak nonlinearities operating far below compression. The simulator was removed because it became redundant - our new HB simulator in v6 can solve the same problems faster than the Volterra simulator. Of course, you can still use the same set of Volterra models, and you will get exactly the same answers when you simulate.

- I saw some comments comparing our HB simulator to Ansoft and ADS...I'm pretty sure they were made by people who had not tried out the latest version of the software. The new HB simulator is much faster, converges much better, and can handle much larger circuits than before. It also has many more options for tuning convergence and speed. Check out the following link for some tests we ran on different mixer circuits:

http://www.appwave.com/products/simulators/HBImprovements.html

 

[Duncan_Widman]

Hi - I work at Applied Wave Research (we make MWOffice), and I wanted to clear up some things I saw in this thread:

- We no longer support Volterra simulation in version 6. Volterra series analysis was introduced to handle simulations that caused trouble with standard Harmonic Balance simulators, like amplifiers with weak nonlinearities operating far below compression. The simulator was removed because it became redundant - our new HB simulator in v6 can solve the same problems faster than the Volterra simulator. Of course, you can still use the same set of Volterra models, and you will get exactly the same answers when you simulate.

- I saw some comments comparing our HB simulator to Ansoft and ADS...I'm pretty sure they were made by people who had not tried out the latest version of the software. The new HB simulator is much faster, converges much better, and can handle much larger circuits than before. It also has many more options for tuning convergence and speed. Check out the following link for some tests we ran on different mixer circuits:

http://www.appwave.com/products/simulators/HBImprovements.html

 

[wasm]

Just a few questions:

Do you know if there is a commercial software that can do equivalent circuit linear model extraction?

If it exists, what kind of measurements are required?
weak gate forward conduction?
open channel?
....

Ic-cap which kind of model implements? only non-linear? It is bundled in the ADS suite?


Thanks in advance

wasm

 

[madengr]

http://www.optotek.com/lasimoa.htm

 

[wasm]

thanks madengr

it does what I need... I don't know if it does it well ;-)

We have an home made extracting tool for linear and non linear model, but human presence is required to obtain good results (i.e. for obtain a physical model)

Did you try it?

I need a trial version of this commercial tool to evaluate it (I will try to ask to MMICAD developers)

 

[madengr]

I bought it a few years ago howver did not have the time to use it much. I pieced together a pulsed bias, s-parameter, and load pull system and planned on using it to extract models for large power FETs. It has some bugs so be sure to get a demo and try it. It does not seem they are actively developing it.

 

[mwmmboy]

In my opinon the better method to simulate the nonlinearities strongly depends on frequency range.
Time domain method (Spice)are used mainly at low frequency because it is difficult to account for ditribuited effect in time domain and because the time domain methods were the first devolped. They are good also for large circuits.
Frequency domain methods (Harminic balance)are better for higher frequency were the distribuited elements a much more important and with a good modelling in the frequency domain. Many frequency domain simulator are designed fior small circuits ather are designed for large circuits.
There are some difference so I thik you have to try what is the better solution for your problem.

Anyway the most important thing is the accuracy of the device model and this is the real critical aspect in nonlinear simulation.


Bye