jslee
Junior Member level 1
Hello Everyone,
I am simulating a small inductor which has inductance of around 50 pH. The Ansoft HFSS is adopted in my simulation.
The outside dimension is only 30 um roughly. A guard ring (PEC) is added to around the structure to let port 1 and port 2 have same reference.
The simulated inductance is around 100 pH. Actually, the guard ring contributes large inductance (~50%), compatible to the desired one.
My question is how to simulate a small inductor accurately. Do I need to do deembedding to remove the guard ring effect?
Another question is where the true reference is.
We add port 1 and port2, respectively. They are referred to the metal trace (guard ring) under the signal trace (thick metal).
However, this two references are actually not same so that the guard ring has impact on the inductance value.
If now we consider to measure the inductor using two probes (like GSG probes). We assume the two probe's ground are same. It seems that simulation does not
match the real measurement case. Am I wrong?
By the way, it seems not possible to use HFSS for full chip EM simulation.
You need a reference for each port. However, the interconnects, even some inductors, on chip sometimes are floating (not circuitous).
It is not easy to set up the port without introducing undesired effect.
Ansoft Designer also has problem as the operating frequency is high. Designer's gap ports are referred to a ground plane, which is far away (300 um)
from the signal path. As operating frequency is higher so 300 um is close to a wavelength, the Ansoft designer starts to have issues.
The port is far away from the ground plane. The port can not represent a TEM mode. The basic assumption of MOM used in Designer seems not to work properly.
Is anyone using Ansoft Designer and face the problem as me?
Thank you.
I am simulating a small inductor which has inductance of around 50 pH. The Ansoft HFSS is adopted in my simulation.
The outside dimension is only 30 um roughly. A guard ring (PEC) is added to around the structure to let port 1 and port 2 have same reference.
The simulated inductance is around 100 pH. Actually, the guard ring contributes large inductance (~50%), compatible to the desired one.
My question is how to simulate a small inductor accurately. Do I need to do deembedding to remove the guard ring effect?
Another question is where the true reference is.
We add port 1 and port2, respectively. They are referred to the metal trace (guard ring) under the signal trace (thick metal).
However, this two references are actually not same so that the guard ring has impact on the inductance value.
If now we consider to measure the inductor using two probes (like GSG probes). We assume the two probe's ground are same. It seems that simulation does not
match the real measurement case. Am I wrong?
By the way, it seems not possible to use HFSS for full chip EM simulation.
You need a reference for each port. However, the interconnects, even some inductors, on chip sometimes are floating (not circuitous).
It is not easy to set up the port without introducing undesired effect.
Ansoft Designer also has problem as the operating frequency is high. Designer's gap ports are referred to a ground plane, which is far away (300 um)
from the signal path. As operating frequency is higher so 300 um is close to a wavelength, the Ansoft designer starts to have issues.
The port is far away from the ground plane. The port can not represent a TEM mode. The basic assumption of MOM used in Designer seems not to work properly.
Is anyone using Ansoft Designer and face the problem as me?
Thank you.
