less than 0.35 dB and a return loss of better than 15 dB was achieved from 1 to 16 GHz.
As the exact values of the permittivities of the materials we used are not known, the simulations were performed with the values of 4 for the glass permittivity [8], and 11.6 and 4000⍀ cm for the silicon permittivity and resistivity, respectively. However, we noticed that the simulation results are very sensitive to the permittivity values-in particular, the resonance frequency, which appears on the S 11 parameter. This explains the difference between the simulations and measurements for S 11 less than 16 GHz. Nonetheless, with this parameter being always less than Ϫ20 dB over this frequency band, the difference is weak.
As this transition is low loss, accurate measurements of devices in inverted line technology can be obtained. Three standards have been measured to extract the S-parameters of the transition [9]: two inverted lines embedded in CPW-inverted line transitions, with the first one of length 600 m (as presented here) and the other of length 1400 m, and an open-ended inverted line of length 700 m preceded by a transition. Thus, the S-parameters of the measured devices, connected to two transitions, can be de-embedded at the inverted-line plane.
5. CONCLUSION
A simple coplanar to inverted microstrip 3D-transition has been designed, measured, and simulated, using a micromachined silicon technology. The very good performances obtained for the inverted lines on silicon explain the interest in developing low-loss transi-tions for such structures. Performances for the back-to-back transitions with less than 0.4-dB insertion loss and more the 13.5-dB return loss were achieved over 1-18 GHz. The measurements obtained agreed very closely with the simulation results. This low-loss transition allows accurate measurement of inverted microstrip structures. It can also be used in complex devices such as RF MEMS integrating coplanar and inverted lines, or to mount inverted-line components in the flip-chip on coplanar waveguides.