Thread: why use models for transistor ac analysis

Hi everyone.Why do we use models like re model or hybrid model etc when talking about transistor in the a.c domain? How come we use direct calculations in the d.c domain but cannot do the same for a.c analysis?Thanks for the help

Originally Posted by fine

Hi everyone.Why do we use models like re model or hybrid model etc when talking about transistor in the a.c domain? How come we use direct calculations in the d.c domain but cannot do the same for a.c analysis?Thanks for the help

To get you to figure out the answer on your own, go ahead and (try) to perform a "direct calculation" in the ac domain for a single-transistor circuit. If you manage that, look at what you've done and assess whether you'd be able to extend that method to circuits of more than one transistor.

I did try it and got wrong answers. So why can't we get the correct answers without these models?

Originally Posted by fine

I did try it and got wrong answers. So why can't we get the correct answers without these models?

You have neglected the possibility that you simply made a mistake. Don't blame the model issue so quickly.

Transistors are nonlinear devices, so the equations are...nonlinear. That makes analysis complicated. If your math chops are not well developed, you won't get far.

To make analysis tractable -- particularly for circuits and systems containing many transistors -- one can use linearized approximations to the full nonlinear behavior. Mathematically, the approach is equivalent to expanding the nonlinear relationships in a Taylor series, and discarding all but the DC and linear terms. You've already stated that you know how to find the DC values, so all that remains are the linear relationships. And that's what "small signal" models are all about.

Being approximations, though, you have sacrificed accuracy for simplicity. One therefore must always keep in mind that the linearization error grows as you move beyond the point around which you performed the linearization.

So it's not that you can't analyze without the simplified models, but that you will greatly increase the likelihood of making it to the (approximate) answer if you do use such models.