Meaning
Mathematical scattering networks provide the linear transfer functions needed to isolate a device under test from surrounding fixture parasitics. Engineers apply de-embedding matrices during high-frequency vector network analyzer measurements to cascade inverse matrix operations against measured raw data. Systematic matrix inversion removes attenuation and phase shift introduced by printed circuit board traces or coaxial transitions.
The boundary of validity stops at the reference planes established during calibration, beyond which non-linear behavior or unmodeled cross-talk cannot be removed.
Network Formulation
Linear network theory models test fixtures as two-port scattering parameter blocks. Operators invert the transmission matrices of launch structures to extract intrinsic device parameters through matrix multiplication.
Fixture Subtraction
Physical test boards contain microstrip lines, vias, and SMA connectors that distort high-frequency signals before reaching the silicon die. Utilizing de-embedding matrices allows RF test engineers to shift the calibration plane directly to the device pads. Calibration standards define the exact electrical length of fixture traces.
Incorrect line length definitions propagate phase errors directly into corrected measurements.
Residual Uncertainty
Measurement errors arise when physical fixtures deviate from ideal models used during calculation. Unmodeled radiation and ground plane discontinuities alter trace parameters during test cycles. If port impedance varies, de-embedding matrices introduce artificial gain spikes or phantom resonances.
Verification requires measuring known check standards to confirm that mathematical subtraction leaves no residual artifact.