How to simulate a frequency dependent resistor?

[Greg]

Hi,
how can I simulate a frequency dependent resistor (skin effect) in a
simulation of an amplifier?
Does anyone have a solution for that?
Greg

 

[Kevin Aylward]

Hi,
how can I simulate a frequency dependent resistor (skin effect) in a
simulation of an amplifier?
Does anyone have a solution for that?
Greg

Yes. Well if it is the standard R=k.sqrt(f) for AC only

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

 

[Paul Burridge]

Hi,
how can I simulate a frequency dependent resistor (skin effect) in a
simulation of an amplifier?
Does anyone have a solution for that?

What do you mean by "in a simulation of an amplifier"?

 

[Fred Bartoli]

Hi,
how can I simulate a frequency dependent resistor (skin effect) in a
simulation of an amplifier?
Does anyone have a solution for that?
Greg

I've used this quite successfully:

http://weewave.mer.utexas.edu/MED_files/MED_research/Intrcncts/Skin_Effect_Ldr/MTT_96_skn_ldr.html

 

[Jim Thompson]

I've used this quite successfully:

http://weewave.mer.utexas.edu/MED_files/MED_research/Intrcncts/Skin_Effect_Ldr/MTT_96_skn_ldr.html

Yep. That's the classic paralleled conducting cylinders model. I
have someone else's paper on it around here somewhere but I can't find
it ;-)

...Jim Thompson

 

[Greg]

Hi,

first, many thanks for the answers!

Of course I tried to give the resistor an equation with the variable
freq, but the software sets freq only to the starting frequency and
does not update it during the frequency sweep :-(
Tried f, fr, frequency also, but they stay at their predefined value
and cause errors if not predefined.
Software is some IC simulation spice from tektronix.

The RL ladder network emulating coaxial sheets seems clever, but it
changes L with frequency also, which I have to avoid as the L in the
model is constant with frequency.

http://www.eettaiwan.com/ARTICLES/2001AUG/PDF/2001AUG27_AMD_AN1541.PDF

Unfortunately my spice also does not have laplace sources so I am bound
with the Rskin=k*sqrt(f) model which also does not work as I described
above.

Now how to make R dependent and L independent?

Greg

 

[Jim Thompson]

Hi,

first, many thanks for the answers!

Of course I tried to give the resistor an equation with the variable
freq, but the software sets freq only to the starting frequency and
does not update it during the frequency sweep :-(
Tried f, fr, frequency also, but they stay at their predefined value
and cause errors if not predefined.
Software is some IC simulation spice from tektronix.

The RL ladder network emulating coaxial sheets seems clever, but it
changes L with frequency also, which I have to avoid as the L in the
model is constant with frequency.

http://www.eettaiwan.com/ARTICLES/2001AUG/PDF/2001AUG27_AMD_AN1541.PDF

Unfortunately my spice also does not have laplace sources so I am bound
with the Rskin=k*sqrt(f) model which also does not work as I described
above.

Now how to make R dependent and L independent?

Greg

Get a better spice ?

...Jim Thompson

 

[Greg]

Unfortunately this is not an option. Its a design system package for
chip design with non-transparent customized libraries so I could not
even switch to a different spice if I wanted. Also another simulator
with that capabilities (suited for IC design, e.g. ability to easily
converge on circuits with hundrets of transistors in it) would cost a
small fortune.

Isn't there any other way of simulating that?

Greg

 

[Jim Thompson]

Unfortunately this is not an option. Its a design system package for
chip design with non-transparent customized libraries so I could not
even switch to a different spice if I wanted. Also another simulator
with that capabilities (suited for IC design, e.g. ability to easily
converge on circuits with hundrets of transistors in it) would cost a
small fortune.

Isn't there any other way of simulating that?

Greg

If there's no behavioral way to do a frequency-dependent resistor,
then you're left with a lumped model, such as Fred Bartoli pointed
out.

BTW, a frequency dependent resistor CANNOT exist without a reactive
term.

...Jim Thompson

 

[Fred Bartoli]

If there's no behavioral way to do a frequency-dependent resistor,
then you're left with a lumped model, such as Fred Bartoli pointed
out.

BTW, a frequency dependent resistor CANNOT exist without a reactive
term.

More, the inductance *does* change with frequency, between two asymptotic
values. Always.
Look for internal/external inductance.

 

[Kevin Aylward]

Unfortunately this is not an option. Its a design system package for
chip design with non-transparent customized libraries so I could not
even switch to a different spice if I wanted. Also another simulator
with that capabilities (suited for IC design, e.g. ability to easily
converge on circuits with hundrets of transistors in it) would cost a
small fortune.

Well, as noted, my XSpice engine in SuperSpice does handle the sqrt(f)
for resisters. Sure, it aint the best in the world for convergence, but
I have been using it solidly for ic design and most of the times it
converges fine for 100+ transistor circuits. You might have to set the
options stiffly, like itl4=5000, and the gmin though. However, apart
from LTSpice, its the only cheap simulator that handles most ic
constructs directly, like the M multiplier for mosfets etc. I might be
able to help with your custom lib. You can email me direct if you want.

Kevin Aylward
[email protected]
remove EXTRACT
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Jim Thompson]

More, the inductance *does* change with frequency, between two asymptotic
values. Always.
Look for internal/external inductance.

Yep. Somewhere back in time I posted my wirebond model that includes
skin effect.

Thank god for skin effect, otherwise wirebonds would have such high Q
that everything would ring.

...Jim Thompson