cadence importing ie3d

hi,

I'm designing a 1nH CTI (Centre-Tapped Inductor) with high quality factor. First, I create a square symmetrical CTI using AISITC and obtain a 1nH inductor with quality factor 7, then I export it into CIF file so that it can be imported into Cadence VLE for editing. In Cadence VLE, I cropped its angles to obtain an octagon CTI (I think octagon CTI has higher quality factor), and exported it into GDS file. The GDS file is then imported into IE3D for electromagnetic simulation. Is it the correct flow for the designing of an octagon CTI?

There are 2 ports for input and output of inductor, and 1 port connected to ground. But warning after the simulation in IE3D states that the PI-model might not be suitable for it, does it means that the PI-model for version 9.0 IE3D not suitable/support the 3-ports (odd-ports) design? This problem doesn't exist when I remove the ground port and run the simulation again, and the results are quite reasonable. Is it possible that the defining of metal layer conductivity is wrong? If so, how should I define the conductivity of metal layer based on its sheet resistance?

Thanks in advance!

Hi, carol82:

The flow is ok. However, it is a little bit cumbersome. Following are some comments on it.

1. IE3D can import CIF directly.
2. The latest IE3D Speiclal Flow Edition (after 11.5) has an automatic flow from Cadence Virtuoso. You can make the process automated and streamlined.
3. The latest IE3D 12.0 has implemented an automatic lumped element conversion program. You can specify to convert the 3-port (center-tapped) s-parameters into 2-port PI-network or 1-port R and L in series.

You should try to upgrade to the IE3D Release 12 for your works. It is much more powerful and provides much convenience to you.

Best regards,

Thanks for your reply, but I think my University might not able to upgrade to the latest version, is there any other way to solve it? Is it possible to do the lumped element conversion in Cadence Virtuoso?

Hi, carol82:

I am not sure about virutoso. I think Cadence SpectraRF should have it because it is a circuit.

You can also do it on the MODUA simulator on IE3D. You can just connect a SHORT to the 3rd port for the center tap, and simulate it to convert it to 2-port s-parameters. Then, you can convert the 2-port to PI network.

Best regards,

Hi,

I'm using the IE3D 12.0 trial version for CTI simulation and define 3 ports for the CTI, after the simulation I choose so that the option is 3-port converted to 1-port series RL and obtain the result as below.

P1 o-----Z1-------------Z2------o P2
|
Z3 ====> P1 o---R--L---o P2
| |
o P3 GND

Q = Omega L / R

R, L and Q are also referenced as R(1,3), L(1,3) and Q(1,3).

Freq(GHz) Q L(nH) R(Ohm)

0.5 9.8009223133e-001 1.0098224756e+000 3.2368901306e+000
1.5 2.9237766956e+000 1.0049325409e+000 3.2393944715e+000
2.5 4.8744670923e+000 1.0078732614e+000 3.2478701506e+000
3.5 6.8213191869e+000 1.0119685891e+000 3.2624703500e+000

Does the L and Q indicates the real value of the CTI or it is just a inductance at port 1 refers to port 3? How should I get the value for Z1, Z2 and Z3 through simulation?

Thanks!!!

Hi, carol82:

Yes. These are the correct values when your center tap is completely symmetrical to the ports 1 and 2. When it is symmetrical and you are feeding ports 1 and 2 differentially, you will get the same result no matter what load you are putting in port 3 (no source).

IE3D V12 allows you to find the equivalent circuit automatically. You just need to pick the right topology you want the s-parameters to be fitted.

Best regards,