5
\$\begingroup\$

I have an extremely simple circuit that involves a NMOS, three resistors, and a voltage source. The circuit and info are shown below. Please understand that I could probably analyze this circuit while driving on the interstate, at midnight, before I hit the next mile marker. I should also mention that this is Example problem 3.3 on page 147 of Microelectronics Circuit Analysis and Design: Fourth Edition. It's not homework, its curiosity.

Why does PSpice generate an answer that is 1 Volt higher than the expected VDS voltage and .5mA less than what it should be for the drain current? I assume PSpice is adding a default value model property that is not considered when the author of the book wrote the example. But that brings question two: If the author can write an example by supplying only two values and get the correct 'ideal' answer, why does PSpice &^%*(^ it up? I realize this circuit may never cure cancer, or land me on the moon. But what if I design a clock tree or buffer that is of some importance? 1 volt becomes important. I feel like I am constantly double checking PSpice results, frankly because I don't trust it.

UPDATE 2012-08-16
I get the correct results if I double the transconductance parameter (KP/Kn) from .1mA to .2mA. Any clue why?

Are there any suggestions?

Am I missing something?

Any constructive criticism would be much apprecitated.

Problem From Book

Calculate the drain current and Vds or a common source circuit with an n-channel enhancement-mode MOSFET. Find the power dissipated in the transistor. For the circuit shown assume R1=30kOhm, R2=20kOhm, RD=20kOhm, VDD=5V, Vtn=1V, and Kn=0.1 mA/V^2

Answer from Book

ID = .1mA, Vds = 3V, Vg = 2V

The Circuit

enter image description here

The Known MOSFET Parameters

Here is a link to PSPICE Model Parameters

enter image description here

\$\endgroup\$
5
\$\begingroup\$

Reading into the authors explanation of the transconductance value (Kn) it was discovered that this value is not the same transconductance value PSpice uses (KP). The author wastes a paragraph of ink explaining that the value Kn used in the book is commonly known as the transconduction parameter, however the reader should refer to the value as the conduction parameter. The author defines the conduction parameter as:

1: Kn = (W*un*Cox)/(2L) and Cox = eox/tox

THEN at the bottom of the page it is written that "we can rewrite the conduction parameter in the form":

2: Kn = (k'n/2)(W/L)

The following page defines k'n as the process conduction parameter:

3: k'n = un*Cox

By default Pspice defaults the width and length to .1uM. Therefor equation two becomes:

4: Kn = k'n/2

Based on the evidence that the PSpice output was off by a factor of 2, k'n represents the value PSpice considers to be the transconductance parameter. The author of the book considers this value to be the process conduction parameter. This result was confirmed with other examples from the book.

I think I am more confused now than before. How do I know what value to use when I look at a spec sheet!?!?

\$\endgroup\$
4
\$\begingroup\$

Most likely the problems is in model. My guess is that is textbook and pspice model "do not speak the same language". That is equations are slightly different and parameters although use the same/similar letters mean different things.

Also spices (presumably pspice too) implement different mosfet models. You do not specify which model you are using but most likely default one is "level=1" model.

For description of level=1 model check page: http://www.ece.uci.edu/docs/hspice/hspice_2001_2-154.html.

From first glance I see following problems: a) You do not specify W and L, who know what is default value - this itself could be an explanation for mismatch b) VTO is not the same as vth - again please check mentioned page.

Regarding your general remark about simulation/pspice and trust. This is only tool, in essence nonlinear differential equation solver with nice interface for EEs application. What you put in is what you get out. The garbage in - garbage out rule applies to it too! The quality of the model (that it describes real devices accurately -> in your case textbook device :) ) and knowledge of its limitation are critical.

\$\endgroup\$
  • \$\begingroup\$ Thanks for your reply. It forced me to do some more reading! I found the solution, somewhat :/ \$\endgroup\$ – atomSmasher Aug 16 '12 at 9:00

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.