1
\$\begingroup\$

In my internship I have to create an amplifier with low noise, high impedance and that works in the 100 to 400MHz band.

While searching for a transistor, I found the ATF 53189 (datasheet), which seems to be suitable for my project. The problem is that I can’t design the model of this transistor in QUCS (Quite Universal Circuit Simulator). Instead, I'd have to model it with ADS (Advanced Design System). But my laboratory needs it with QUCS.

Below is the model image.

So my question is: Can I model the transistor ATF 53189 in QUCS? How?

Model transistor ATF 53189 in ADS logiciel

\$\endgroup\$
  • \$\begingroup\$ What kind of BJT model does QUCS use? \$\endgroup\$ – gsills Mar 21 '14 at 16:55
  • \$\begingroup\$ HA. Now that I look at this on a larger screen I see that you're asking about a FET not a BJT. Sorry about the irrelevant question. \$\endgroup\$ – gsills Mar 23 '14 at 21:42
2
\$\begingroup\$

Not having used QUCS, this will be a very generic answer.

It looks like QUCS, in addition to having a standard MOSFET template (looks like Shichman-Hodges), also supports Verilog-A. First thing to check is if the ADS model from Avago can be ported or converted to Verilog-A. If not, you will need to extract parameters from either the datasheet or measurements to map into either:

  • QUCS standard model (Shichman-Hodges)
  • Roll your own Verilog-A lumped physical model
  • Roll your own Verilog-A behavioral equation based model

The QUCS site has some papers describing creation of Verilog FET models, like: "Verilog-A compact device models for GaAs MESFETs"

It's not clear how complete the model needs to be, and what it needs to include. You might want to modularize the model. Put the device into one model, but all the package parameters into another model that the device can be plugged into later. That way the package part might be more generic.

Remember that all models are wrong. The only accurate model is the thing itself, and there is not much advantage in that. So, try to make whatever model you use as numerically light as possible, while still being accurate enough to be useful in the region of interest. Keeping things focused on a region of interest will help the model from becoming so numerically ponderous that it fails to converge for any practical circuits.

Excuse me if I've just stated what is too obvious here. Good luck.

\$\endgroup\$
  • \$\begingroup\$ Hello, thank you very much for this information I will do my best to solve this problem and find the model required for my project. if I will face other problems I would like to share them here it's so useful to have advices from you. \$\endgroup\$ – eng.M.AHMED Mar 25 '14 at 18:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.