When I published the appnote on filler and ground effect for a 60GHz balun, there were many comments and questions on the ground cutout. It seems that many designers didn’t realize the possible loss in performance from a ground cutout that is too small.
That’s enough reason to adress the issue and a possible solution again in more detail now!
The new testcase, a 500pH octagon inductor designed for 28 GHz, shows a performance loss of 25% in Q factor and 10% reduction in L when adding a tightly spaced ground frame below. This is caused by eddy currents in the closed ground frame. Countermeasures like slots can reduce that effect.