AnsweredAssumed Answered

Momentum simulations and measurements not agreeing

Question asked by Chemic on Mar 29, 2009
Latest reply on Mar 30, 2009 by ALEPEREZ
Good Morning.

I am using Momentum to simulate our product's printed circuit board. The system is quite complex, it is a GPS L1 receiver module (two sided FR4, 100um), glued on another PCB. The two circuits are then connected together with wire-bonds, which I add into the circuit cosimulation in ADS. The resulting assembly is then connected to another FR4 PCB (the user equipment) with a connector, that I model using via holes.

This is the simulation stackup:

Free Space (OPEN)
------ -STRIP (copper)
FR4 (100um)
------ -STRIP (copper)
Glue (60um)
------ -STRIP (copper)
FR4 (100um)
------ -STRIP (copper)
AIR (1mm)
------ -STRIP (copper)
Free Space (Open)

I import DXF files from our PCB cad, but I cannot model the metal thickness, as the process size is far too big for my 32 bit machine. Using edge meshing and pushing the mesh to 40 cells/wavelength, I belive I have an acceptable tradeoff between accuracy and complexity.

As a benchmark, I measured and simulated the crosstalk between two tracks into the design. Both are coming through all the layers. Into the simulation I included also the lumped components and wire-bonds that were placed into the measurement.

Here comes the problem: the simulation shows exceptional fidelity to the measurements up to about 1GHz, but then it diverges from the measurement (the difference is more than 20dB, whitout any apparent correlation into the curves).

Could it be a problem due to metal thickness problem?

Thanks for the answer