Q:limits of spice simulation

M

Michael Schuster

Guest
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

Any link or comment (especially if you've faced a certain problem) is
welcome.

Thanks in advance

Michael
 
Michael Schuster wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.
Click to expand...
What do you mean by limits?

For example, the main error sources for a simulation are the models.
Models are pretty much independent of the simulator, although some
simulators have some minor variations on how the handle such models.

Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
 
Michael,

where can I find some infos about some limits of
spice simulation? I need it for a comparision of
different simulators.
Click to expand...
Maybe you're trying to get at the trade between
accurate transistor-level simulation with a global
Jacobian as used in SPICE and partitioning solvers
and/or gate-level solvers which don't have SPICE
accuracy but the capability to simulate orders of
magnitude larger designs. You might look at the
sales collateral from vendors selling these
solvers. I remember seeing one in the Anagram
brochure. It was a 2D plot of accuracy vs.
circuit size with a regions marked that were
viable for various solver methodologies.

--Mike
 
In article <f6742f8c.0408172247.3e578aeb@posting.google.com>,
Michael Schuster <michi@flyflaps.de> wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

Any link or comment (especially if you've faced a certain problem) is
welcome.
Click to expand...
Here's a few:

In general, simulators never disprove something the models assume.

Spice models usually do not sumulate behavour outside the normal operating
conditions of the part. Things like the output side characteristics of an
op-amp when Vcc is much below operating would be an example.

Thermal issues are generally not included. This can lead to missing real
life problems like thermally caused low frequency distortion.

Op-amp models often do not include things like power supply rejection.

Some models, of things like swither chips, assume that the chips ground is
connected to the global ground.

Models of noise are usually very simplified views of what really happens.


Simulator software can't really handle much of a mixed digital and analog
system. Take a system with a few op-amps, an ADC, a 512 macro cell CPLD,
a DC-DC converter, and an 8051 as an example. It is fairly hopeless
trying to model the whole thing without making so many simplifying
assumptions that your model is mostly a fairy tail.



--
--
kensmith@rahul.net forging knowledge
 
"Kevin Aylward" <salesEXTRACT@anasoft.co.uk> wrote in message news:<r8DUc.155454$28.137622@fe1.news.blueyonder.co.uk>...
Michael Schuster wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

What do you mean by limits?
I mean e.g. the following:
Click to expand...

I want to simulate behaviour of a resistor which characteristic (U:
Voltage, I current) is given by
I = f(U)
or
U = g(I)

Can I feed these characteristics directly into spice or would I have
to convert e.g. g(I) to its inverse function in order to model this
resistor with spice?

Or a component is given by:
I = d/dt(I) * d/dI(f(I))
or
U = d/dt(U) * d/dI(g(U))

Possible or not?
Thanks for your answers!
Michael
 
On 19 Aug 2004 05:40:20 -0700, michi@flyflaps.de (Michael Schuster)
wrote:

"Kevin Aylward" <salesEXTRACT@anasoft.co.uk> wrote in message news:<r8DUc.155454$28.137622@fe1.news.blueyonder.co.uk>...
Michael Schuster wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

What do you mean by limits?
I mean e.g. the following:

I want to simulate behaviour of a resistor which characteristic (U:
Voltage, I current) is given by
I = f(U)
or
U = g(I)

Can I feed these characteristics directly into spice or would I have
to convert e.g. g(I) to its inverse function in order to model this
resistor with spice?

Or a component is given by:
I = d/dt(I) * d/dI(f(I))
or
U = d/dt(U) * d/dI(g(U))

Possible or not?
Thanks for your answers!
Michael
Click to expand...
Read up on "Behavioral Modeling" in the PSpice manual. Powerful, yet
easy.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
 
Michael Schuster wrote:
"Kevin Aylward" <salesEXTRACT@anasoft.co.uk> wrote in message
news:<r8DUc.155454$28.137622@fe1.news.blueyonder.co.uk>...
Michael Schuster wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

What do you mean by limits?
I mean e.g. the following:

I want to simulate behaviour of a resistor which characteristic (U:
Voltage, I current) is given by
I = f(U)
or
U = g(I)

Can I feed these characteristics directly into spice or would I have
to convert e.g. g(I) to its inverse function in order to model this
resistor with spice?
Click to expand...
Yes. You can make a model using the B source, or a gm source.

Or a component is given by:
I = d/dt(I) * d/dI(f(I))
or
U = d/dt(U) * d/dI(g(U))

Possible or not?
Click to expand...
Yes. Any standard spice will allow you to model this simply by using
capacitors/inductors in a little ".subckt". XSpice based ones (with is
just about all of them) will also allow a second method, that is by
using laplace S transforms. Look up the help in what ever spice you use
e.g. http://www.anasoft.co.uk/Spice3F5Manual.html


Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
 
michi@flyflaps.de (Michael Schuster) wrote in message > I want to simulate behaviour of a resistor which characteristic (U:
Voltage, I current) is given by
I = f(U)
or
U = g(I)

Can I feed these characteristics directly into spice or would I have
to convert e.g. g(I) to its inverse function in order to model this
resistor with spice?

Or a component is given by:
I = d/dt(I) * d/dI(f(I))
or
U = d/dt(U) * d/dI(g(U))

Possible or not?
Click to expand...
This would be simulator dependent. I know with Micro-Cap, you can put
in an equation directly into a resistor to define the resistance, and
I believe PSpice does the same although I'm not 100% sure on that.

Alex
 
On 19 Aug 2004 13:43:59 -0700, engr4fun@yahoo.com (Engr4fun) wrote:

michi@flyflaps.de (Michael Schuster) wrote in message > I want to simulate behaviour of a resistor which characteristic (U:
Voltage, I current) is given by
I = f(U)
or
U = g(I)

Can I feed these characteristics directly into spice or would I have
to convert e.g. g(I) to its inverse function in order to model this
resistor with spice?

Or a component is given by:
I = d/dt(I) * d/dI(f(I))
or
U = d/dt(U) * d/dI(g(U))

Possible or not?

This would be simulator dependent. I know with Micro-Cap, you can put
in an equation directly into a resistor to define the resistance, and
I believe PSpice does the same although I'm not 100% sure on that.

Alex
Click to expand...
Yes, PSpice can do equations directly.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
 
Thanks for your answers

Michael
kensmith@green.rahul.net (Ken Smith) wrote in message news:<cg0ao6$108$1@blue.rahul.net>...
In article <f6742f8c.0408172247.3e578aeb@posting.google.com>,
Michael Schuster <michi@flyflaps.de> wrote:
Hello,
where can I find some infos about some limits of spice simulation? I
need it for a comparision of different simulators.

Any link or comment (especially if you've faced a certain problem) is
welcome.

Here's a few:

In general, simulators never disprove something the models assume.

Spice models usually do not sumulate behavour outside the normal operating
conditions of the part. Things like the output side characteristics of an
op-amp when Vcc is much below operating would be an example.

Thermal issues are generally not included. This can lead to missing real
life problems like thermally caused low frequency distortion.

Op-amp models often do not include things like power supply rejection.

Some models, of things like swither chips, assume that the chips ground is
connected to the global ground.

Models of noise are usually very simplified views of what really happens.


Simulator software can't really handle much of a mixed digital and analog
system. Take a system with a few op-amps, an ADC, a 512 macro cell CPLD,
a DC-DC converter, and an 8051 as an example. It is fairly hopeless
trying to model the whole thing without making so many simplifying
assumptions that your model is mostly a fairy tail.



--
Click to expand...
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top