Hybrid/Mixed EM and Circuit simulator?

时间：03-22 整理：3721RD 点击：

Hi all:

Is there any Hybrid EM & circuit simulator?

As you all know, EM solvers (FEM,FDTD,MOM)
try to solve PDE (Partial Diff. Equation-- Maxwell Equation).

SPICE like simulators (HSPICE, PSPICE..) are
solveing ODE (Ordinary Differential Equation).

Instead of geting Frequency domain data (S-parameter)
and do macromodeling to export to SPICE, do
people solve them at one shoot?

FDTD can include lumped elements, but
that is not what I am talking about.

I am saying given a system model with
A) lumped networks
B) EM structures
C) Nonlinear devices
can a general solver solve it?

So final challege:
Is there a hybrid Nonlinear PDE & ODE solver/algorithm?

Thanks
Div

Many software can do this but they uses different solver and methos to get the final thing done. Output of one method is used as input of others. I have not seen such general solver yet. Is that really exist??

For example check the IE3D output of an amplifier at
http://www.elektroda.pl/eboard/viewtopic.php?t=43185
where IE3D uses MOM to solve the transmision lines. Now the transistor is a nonlinear device whose characteristic is obtained with formulas of solid states. But we are using the output of it. And finally IE3D calculates the final output by using transmission line theorem which is basically electrical network property. So all different methods have been used to get the output.

maxwell9.0 can do what do you want.

Is it?? 8O

you should go to www.ansoft.com for more infomation

Can you please show the exact place where it is written that maxwell can simulate a nonliner device like FET ??

From what I could see, without having the program,
Ansoft Designer uses a Spice models of nonlinear devicesl like FETs,
while the rest of the circuit is simulated with the Method of moments.
I am a bit sceptical about this approach, but it may just work.

I think nothing solves problem in one pass as a combined simulator. But, I use EagleWare EMPOWER to do this, I don't think others can. It simulates including transistors/resistors/etc

speek@2000 should work also . see hxxp://www.s-i-g-r-i-t-y.com, it's based on FDTD

An appropriate software for linear/nonlinear/EM simulation:

@WR Microw@we Office with VSS
@gilent @DS2002
E@gleW@re Gene$y$

All this progs have 2.5D EM simulator, linear and nonlinear analysis of schematics with semiconductor devices like FETs, BJTs, PIN-diodes etc.

Best regards,
Kit-the-great

Hi all:

Thank you all for the disscussion.

I consider the probelm again.
Here is my two cents.

Since Maxwell equation is using Ex,Ey,Ez and
Hx,Hy,Hz, but Modified Nodal Equation (MNA, SPICE)
is using V and I (Voltage and Current).
The relation betwen these two groups is either
differential or integration-- for example, V=[int E dl]

If you want to combine them together and solve
it at once, you need to solve equations have both
differential and integrate operators. And the unknown
vector can either be [Ex Ey Ez Hx Hy Hz] or [V I].
It depends.

Now the problem is how to build equation and solve
this differential-integrate equation?

Give me your opinion and suggestion, please.

Div

the key is to add current density at your maxwell euqation

curl H = dD/dt + J
using J=I/area or J=sigma* E you can get euqation related to I,V

see paper on:
"Extending the two-dimensional FDTD method to hybrid electromagnetic systems with active and passive lumped elements"