how to improve the L, Q and the efficiency?

[mustangyhz]

I have simulated a transformer proposed in the following paper, but I got very low L, Q and efficiency.
Can you tell me what's wrong with my workspace, mml_wrk.zip, in ADS2011?

 

[vfone]

In your ADS simulation you placed the P1 and P2 ports one on primary and one on secondary windings of the transformer.
I think you have to place first P1 and P2 on primary (and measure the inductance and Q) and later place on secondary and do the same.

 

[volker@muehlhaus]

Mistakes:

1. Wrong boundary condition. Your transformer in Momentum is placed at 10µm distance above a solid ground plane (bottom cover in Momentum substrate). This metal plane under the inductor dramatically changes inductor magnetic fields, this can't work. There is no such ground plane in your reference paper.

Solution: Remove the ground plane. You can use differential ports with explicit (-) pins to place the port reference. The (-) pins then go to the one end of the spiral that is connected to the ground via at the moment.

2. Your inductor is inserted into the conducting silicon with no isolation between metal and silicon substrate. This is wrong, and different from the model in your reference paper.

As I mentioned in your other post, in Momentum you can't model this very special case where the inductor inserted into the silicon substrate with a thin isolating oxide layer as isolation. That requires 3D volume meshing (FEM solver) and can't be done with the Momentum solver. Also, it is tricky to create the isolating oxide. That setup will take hours even if you are an ADS expert.

 

[mustangyhz]

Mistakes:

1. Wrong boundary condition. Your transformer in Momentum is placed at 10µm distance above a solid ground plane (bottom cover in Momentum substrate). This metal plane under the inductor dramatically changes inductor magnetic fields, this can't work. There is no such ground plane in your reference paper.

Solution: Remove the ground plane. You can use differential ports with explicit (-) pins to place the port reference. The (-) pins then go to the one end of the spiral that is connected to the ground via at the moment.

2. Your inductor is inserted into the conducting silicon with no isolation between metal and silicon substrate. This is wrong, and different from the model in your reference paper.

As I mentioned in your other post, in Momentum you can't model this very special case where the inductor inserted into the silicon substrate with a thin isolating oxide layer as isolation. That requires 3D volume meshing (FEM solver) and can't be done with the Momentum solver. Also, it is tricky to create the isolating oxide. That setup will take hours even if you are an ADS expert.

1、I used the ground plane(10 µm substratemml) under the inductor instead of the BCB layer in my reference paper, what's more, if I remove the ground plane,I can not map metal layer between the 300 µm substrate and the bottom cover, then I can not define the layer42 intruded into substrate.
2、You told me that I should consider using the ADS 3D FEM solver instead of Momentum. Is there any example about inductor or transformer 3D FEM simulation? I don't know the detailed steps.

 

[volker@muehlhaus]

1、I used the ground plane(10 µm substratemml) under the inductor instead of the BCB layer in my reference paper

I see that you did that, but this is a completely different problem with completely different behaviour now. Makes no sense to me.

what's more, if I remove the ground plane,I can not map metal layer between the 300 µm substrate and the bottom cover, then I can not define the layer42 intruded into substrate.

As I wrote above, yes, this is more complicated because you now need explicit port reference (explicit ground pins). To get started, for differential inductor parameters you could also group pins 1 and 2 into one differential port, as discussed here:
https://muehlhaus.com/support/ads-application-notes/inductor-em-ports

And intruding the metal into the substrate (with no isolating oxide between them) is wrong because this creates a conducting path from inductor to the substrate. There is no solution for this in Momentum. You need the FEM solver and need manual work to model the isolating oxide. This can be done with derived layers, similar to the work that i showed here:
https://muehlhaus.com/support/ads-application-notes/ads-fem-3d-passivation
That is a lot of work and I can't describe all that in detail, sorry :-D

2、You told me that I should consider using the ADS 3D FEM solver instead of Momentum. Is there any example about inductor or transformer 3D FEM simulation? I don't know the detailed steps.

My "real" examples are in real technologies that are confindential and can't be shared. Contact Keysight support for help, they might have an example to share.