This bug has been discovered by Margeride48. She brought to my attention several strange results simulating microstrip coupled lines. She wanted me to try to reproduce her results, just to confirm the wrong ones that she was getting from the simulator.
The following lines explain the anomaly in her words:
Focusing on coupled microstrips lines, have you tried thinner dielectrics ?
I’m simulating a 0.254mm I-Tera MT RF material.
S-parameter simulation values are going suddenly totally wrong when thickness is lower than 0.44 mm (eg 0.435mm).
Same for the Coupled Microstrip Line Calculator !
Impedances, losses, etc, become suddenly wrong.
On the opposite, the ‘single’ Microstrip Line Calculator leads to good results for 0.254mm thickness.
An EM simulation performed on the auto-generated PCB give good results with 0.254mm thickness for coupled microstrips.
We made a little investigation and I think that the best way of understanding the findings is to consult the document attached that summarizes the results we got.
The document contains several simulations and figures very complicated to describe in plain text.