Simulation of an inductor with patterned ground shield in Momentum

[Hamid.Kiumarsi]

Dear experts,

I am trying to simulate an inductor with Patterned Ground Shield (PGS). PGS is on the "poly" layer and inductor is on the top metal.
Since the ground is patterned, I can not use the infinite ///GND/// of Momentum.
One way is to define two single ports for the inductor on the top metal and then adding ground reference ports for each single port on poly layer.
For checking that ground refernce ports will work well, first I didn't use patterned ground and just used a solid sheet of poly under inductor and defined ground reference ports on it. Then I also simulated this inductor when poly layer was removed and instead of it, infinite ///GND/// of Momentum with the conductivity value same as poly was used. Surprisingly simulation results of both cases was very different. (of course a slight discrepancy is acceptable but not like changing Q from 9 to 1.5 !)

After this trail, I wanted to assign one differential port to the inductor instead of 2 single ports. But the problem now is S11 (dB) is like -0.004dB for this differential port.

How should I accurately simulate an inductor with PGS using Momentum? Is there any point that I am missing in my simulation?

Regards,
Hamid

 

[BigBoss]

Defining any layer as GND in Momentum as you have done will give you high discrepancies in your simulations.
Instead, define POLY layer as a internal layer and make the global GND as a reference for all layer.Because Momentum assumes the global GND is everywhere and if you use a layer as GND layer, it will confuse.

 

[Hamid.Kiumarsi]

BigBoss

Thank you for your prompt reply,
The problem is: If I define poly as layer and have my patterned shield there, then this global GND (///GND///) of Momentum in where should be defined?
Actual patterned ground shield will be on poly and then beneath poly is substrate. If I put ///GND/// right under the substrate, then the drawn layout in Momentum and my actual inductor are not same.

Actual inductor layers are as below

Polyimide
---------------STRIP- Top metal
Oxide layer

--------------STRIP- poly >>>>>> patterned ground shield will be here
Substrate


Thank you in advance

---------- Post added at 15:41 ---------- Previous post was at 15:33 ----------

BigBoss, also what do you think about using differential port?
By using differential port then ///GND/// will not be needed?

 

[BigBoss]

Put a port to your POLY shield with same layer and make it grounded in your schematic simulation.
Global Ground is NOT a layer, it's just a infinite dimensioned ground definition in Momentum,you don't have it to tie to the ground, it will be presumed as ideal ground at infinite.
Otherwise, Momentum will confuse the ground waves that are taken into account in computation.

 

[grit_fire]

hello,

at first sight, why don't you put oxide between PGS and si-substrate?

---------- Post added at 20:43 ---------- Previous post was at 20:42 ----------

if you draw PGS on a slot layer, the conductive part will be grounded automatically in Momentum---hence the PGS is grounded.

This is what Momentum does. I asked their technical staff about this before and that is what they told me.

---------- Post added at 20:46 ---------- Previous post was at 20:43 ----------

and I am not sure if you can define the conductivity of a infinite ground plane in Momentum? which version of ADS are you using?

---------- Post added at 20:48 ---------- Previous post was at 20:46 ----------

and what is your inductor like? are the two terminals close to eath other? only in that case should you use differential port according to the documentation about "Ports" on ADS website.

---------- Post added at 20:49 ---------- Previous post was at 20:48 ----------

sorry that i have been to busy to reply you in the previous post

 

[Hamid.Kiumarsi]

BigBoss,

Thanks for comment. I did what you suggested. Since when using data items in schematic, they have a reference pin; after importing s3p of the inductor (port 3 is placed on poly) into the data item, this reference pin should be tied to GROUND of schematic.
Then pin 3 of the data item should be connected to the - terminals of port 1 and port 2 while + terminals of port 1 and port 2 are pin1 and pin2 of data item receptively. This scenario generates reasonable results, but in case that pin3 of the data item was connected to the GROUND of schematic, results were not reasonable.

---------- Post added at 07:04 ---------- Previous post was at 06:44 ----------

Dear grit_fire

Thanks for reply.

1) I didn't put oxide between PGS and Si substrate because beneath poly I have substrate according to the information of layers for the process that I am using.
2) Drawing PGS on slot layer is a good trick as you mentioned. But problem is in slot layer we con not define conductivity and its perfect conductor whereas poly is way different than perfect conductor.
3) I am using ADS 2009 Update1 and I defined conductivity of infinite ground plane.
4) Two terminals of my inductor are close to each other. But as I mentioned I didn't have much luck using differential ports. By defining deifferntail port, I had only one port and S11 (dB) of this port is about -0.004dB. I defined differential port according to the guidelines in mom.pdf from ADS. Even I tried to put a lumped inductor in schematic and extract its L and Q using differential port which also went wrong. Could you please explain in more detail that what should I do for very simple case of lumped inductor in schematic simulation.

Again thank you for giving your comments.

Hamid

 

[tony_lth]

I guess in HFSS it's better to simulate.

 

[Hamid.Kiumarsi]

I guess in HFSS it's better to simulate.

But optimization of momentum is way better than HFSS

 

[grit_fire]

But optimization of momentum is way better than HFSS

HFSS should be much slower. But maybe you can run it once to check if the results matched with momentum under the same layout and configuration.