Why De-Embedding Is Needed
VNA calibration moves the reference plane to the cable ends. Any test fixture, PCB launch structure, or bonding wire between the cable and the DUT adds parasitic effects that must be mathematically removed — this is de-embedding. Without it, the measured S-parameters include both the DUT and the fixture, giving incorrect results.
Two-Port De-Embedding Methods
| Method | Standards Required | Best For |
|---|---|---|
| Open-Short | Open + Short structures | On-wafer probing, RFIC |
| Thru-Only | Direct connection | Balanced/differential lines |
| ABCD Cascade | Characterized fixture S2P | PCB launch removal |
| TRL | Thru + Reflect + Line | Moves cal plane to board edge |
Open-Short De-Embedding Procedure
1. Measure OPEN fixture (DUT removed): S_open → Y_open 2. Measure SHORT fixture: S_short → Z_short 3. Measure DUT in fixture: S_raw → Y_raw De-embed shunt parasitics: Y_dut1 = Y_raw − Y_open De-embed series parasitics: Z_dut = 1/Y_dut1 − (Z_short − Z_open_series) Convert Z_dut → S_dut (true device response)
ABCD Cascade De-Embedding
[ABCD_dut] = [ABCD_fixture1]⁻¹ · [ABCD_measured] · [ABCD_fixture2]⁻¹ Requires: characterized S2P (or ABCD) of both fixture halves Convert final [ABCD_dut] → S-parameters
Common De-Embedding Errors
- Fixture resonance: Fixture S-parameters have resonances near operating frequency — extrapolation fails
- Wrong reference plane: Open/Short structures don't match DUT launch geometry exactly
- Multimode propagation: PCB transitions excite non-TEM modes above ~10 GHz
RF View: RF View reads de-embedded Touchstone files exported from VNA software or EDA tools. Use the multi-file overlay to compare raw vs de-embedded S-parameters on the same chart.