Why does ICE not use the offset methods of the R&S ZVx?

[marcvb]

The offset method on R&S ZVx (or any type of network analyzers) extends or retracts the calibration plane. In many large signal characterization setups tuners are being and are possibly positioned between the network analyzer and the device under test. Whether this is the best topology this is another type of discussions.
Anyway for such topologies the tuners need to be deembedded dynamically as they change impedance and this is not supported by regular network analyzers. As such in ICE we did provide deembedding by adding S-par blocks in the schematic / setup. Of course, the user needs to supply the S-parameters.
One of the many possibilities ... using a fixture with a short (usually better defined than the open) in the DUT plane:
- measure the reflection factor after calibration of the short in the ficture
- assume S11=S22=0
- the refl factor = S21 (-1) S12 (ideal short - or use a model if available)
- calculate S21 (using reciprocity and check whether the root selection was properly done)
- generate a s2p file
- load s2p file in deembedding block in ICE