I have designed a matching network using ADS for some on-chip circuitry. Now I need to model the effect of wire bonds from the chip to a PCB using CST. On the PCB I have grounded coplanar waveguides. But I am unable to match the S11 values from ADS in the CST simulation even without wire bonds.
I have a parallel RC termination on one end which I have entered using lumped element. I have also ensured proper meshing.
Could it be that CST is not very reliable with such waveguides in the 50-70 GHz range? Around 10 GHz I get a good match between the tools.
Thanks!
CST is perfectly good at 60 GHz. Any difference will be because your model is not like your actual circuit. The termination can make a difference, if the currents on the transmission line are not the same shape as the currents at the termination there will be reactive fields around the termination that look like inductors or capacitors. Bond-wire shape and thickness are also important parameters.
Thank you, Malcolm White
I have an RC termination from lumped elements. Even without the bond wire my results are off. As my matching stub is designed for this specific RC load, I expect an S11 < -15 dB (from ADS). Maybe I am using the lumped element wrong?
I am attaching a screenshot of how I have connected the lumped element between GCPPW line and ground.
You have excited, or loaded the coplanar waveguide assymetrically. You have excited it in a combination of odd and even modes. You need to excite and load it symmetrically.
Take care of parasitics too at that high a frequency range. Coupling effects of the bond wires, how they are looped and also the sizes, shapes and orientations of the lumped elements could cause differences in results between the simulation and prototype phase.
- Badges
- Science topic