Why De-embedding Is Necessary
When measuring a device under test (DUT) with a VNA, the measurement includes the effects of the test fixture, cables, connectors, and PCB launch structures. De-embedding mathematically removes these parasitic effects to obtain the true DUT S-parameters. Without de-embedding, you are measuring fixture + DUT, not just DUT.
De-embedding Methods
| Method | Standards Needed | Accuracy | Complexity |
|---|---|---|---|
| Open-Short | Open, Short | Good to ~10 GHz | Low |
| Thru-Reflect-Line (TRL) | Thru, Reflect, Line | Excellent | Medium |
| LRRM | Line, Reflect, Reflect, Match | Excellent on-wafer | Medium |
| Network subtraction | S-params of fixture halves | Good | Low |
Network Subtraction (S-Parameter Level)
The simplest de-embedding approach: measure S-params of each fixture half separately (left half: ports 1 to DUT port A; right half: DUT port B to port 2). Convert to T-matrices (cascade matrices), then:
T_DUT = T_fixture_left⁻¹ · T_measured · T_fixture_right⁻¹
Convert T_DUT back to S-parameters. This is the T-matrix (transmission matrix) de-embedding method.
T-Matrix and S-Matrix Conversion
From S to T: T11 = −det(S)/S21, T12 = S11/S21 T21 = −S22/S21, T22 = 1/S21 From T to S: S11 = T12/T22, S12 = (T11·T22 − T12·T21)/T22 S21 = 1/T22, S22 = −T21/T22
Practical De-embedding Workflow
- Measure the complete fixture + DUT assembly → save as
dut_in_fixture.s2p - Measure left fixture half (open at DUT plane) → save as
fixture_left.s2p - Measure right fixture half → save as
fixture_right.s2p - Apply T-matrix de-embedding formula above (Python script or EDA tool)
- Load de-embedded result in RF View to verify DUT performance
RF View Use in De-embedding Workflow
RF View helps at multiple stages:
- Load raw fixture measurement to assess fixture quality (S11, S21 of fixture alone)
- Compare raw vs. de-embedded DUT S-parameters side by side
- Batch-load de-embedded files from multiple DUT samples for statistical analysis
- Verify de-embedding is working: the de-embedded fixture-only measurement should approach an ideal thru (S21 → 0 dB, S11 → −∞ dB)
Common Pitfalls
- Port reference plane errors: ensure fixture halves are measured to the correct reference planes
- Fixture repeatability: poor connector mating reproducibility introduces errors that de-embedding cannot fix
- Frequency aliasing: de-embedding assumes minimum-phase behavior; resonant fixtures can cause artifacts