How to simulate Insertion Loss of cmos transformer in ADS Momentum

hi I'm a fresher in designing on-chip transformer, I want to simulation the IL in momentum, how to simulation? the testbench is?

could you help me? many thx

Evaluate S21.
If needed, evaluate S21 in diffential and common modes.

Could you explain how to evaluate transformer's common-mode using Momentum?
For differential-mode, defining differential ports will do the job but how about common-mode.

I am asking this because usually solid ground is not used for transformer; therefore ///GND/// of Momentum should not be used. Also ground reference ports will not lead to accurate results especially if patterned ground shield on the poly layer is used.
This leaves us with no choice but using differential ports. The problem is common-mode could not be evaluated in this case.

Do you have a solution in Momentum?

The problem is getting accurate single-mode S parameters in this case is not possible.
As I explained I have patterned ground shield on poly.
One way is assigning single ports on each terminal of inductor and associating ground reference ports defined on poly layer with these single ports.
But unfortunately aforementioned method is not leading to accurate results. I mean differential S parameters of this case should be same as differential S parameters obtained using differential ports; but unfortunately they are not equal.

I think ground reference ports are working well only on perfect conductor layers.

What is your opinion?

Treat poly layer as signal port not ground reference.

It is a natural result. Consider equivalent circuit for DUT.

Consider that I want to simulate an inductor with a patterned ground shield on poly.
If I assign two single (or internal) ports on every terminal of inductor and another single (or internal) port on poly, then these ports need to have a either implicit (///GND/// in Momentum) or explicit (ground reference ports) ground.
But there will be no ground in this case and as a result S parameters from this three-port network will not have a physical meaning.

It's natural that differential L and Q are not same as L and Q which are obtained from single-mode S parameters. (considering equivalent circuit for DUT as you mentioned)
But if the mixed-mode S parameters are calculated from the single-mode S parameters and then differential L and Q are obtained from these mixed-mode S parameters; differential L and Q should be same though slight discrepancy is acceptable.
I mean either we use one differential port or two single ports, differential L and Q should be same. (in the two single ports case, L and Q are obtained from mixed-mode S parameters and not from single-mode S parameters)

申明：网友回复良莠不齐，仅供参考。如需专业帮助，请学习易迪拓培训专家讲授的ADS视频培训课程。