Corrected Data
Understanding Corrected Data
When you plug a highly precise RF filter into a Vector Network Analyzer (VNA) and hit the sweep button, the raw voltages hitting the internal receivers are absolute garbage. The raw signal has been warped by the phase delay of the 3-foot test cables, distorted by the VSWR reflections of the SMA connectors, and corrupted by the frequency roll-off of the internal VNA couplers. If you looked at this "Raw Data," the filter would appear completely broken. The beautiful, pristine graph you actually see on the VNA screen is the Corrected Data.
Corrected Data is the magical result of VNA Calibration. During calibration, the engineer connects known standards (Short, Open, Load, Thru). The VNA measures the raw garbage and uses the known truths of the standards to calculate 12 massive mathematical error matrices (The 12-Term Error Model). These matrices perfectly map every microscopic flaw in the cables and VNA hardware.
The Matrix Inversion
When you finally plug in your prototype filter, the VNA measures the new Raw Data. The supercomputer instantly applies the 12 error matrices, running massive matrix algebra (inversion and cascading) in real-time. It mathematically strips away the test cables, strips away the internal VNA reflections, and completely isolates the actual, physical performance of the filter. This resulting Corrected Data is the absolute truth of your device.
S11Corrected = ( S11Raw - ED ) / [ ET + ES × ( S11Raw - ED ) ]
This equation is calculated simultaneously for every single frequency pixel on the screen in milliseconds. For full 2-port measurements, this expands into a massive 12-term matrix calculation.
Comparison
| Data State | What it Contains | Usefulness to Engineer |
|---|---|---|
| Raw Data | VNA hardware + Cables + DUT | Useless (Cannot isolate the DUT) |
| Error Terms | VNA hardware + Cables (Isolated) | Used internally by the software engine |
| Corrected Data | DUT ONLY (Perfectly Isolated) | The absolute truth (What is exported and saved) |
Frequently Asked Questions
How do I know if the VNA is displaying Raw or Corrected data?
Look at the bottom corner of the VNA screen. There is a specific UI indicator. If it says 'Cor' (Corrected), the massive math engine is running and the data is valid. If it says 'Off' or 'Unc' (Uncorrected), the VNA has disabled the math engine, and you are looking at raw garbage data. Never screenshot a VNA that does not say 'Cor'.
What happens if I bend the cables after calibration?
You instantly invalidate the Corrected Data. The error matrices were calculated based on the cables being in a specific physical position. If you bend the cable, the physical phase delay of the cable changes. The VNA is still applying the 'old' error matrices, meaning it is mathematically subtracting the wrong amount of phase. The resulting Corrected Data will now be polluted with massive artificial ripples and errors.
Can I extract the Raw Data from the VNA?
Yes, advanced metrologists often do this. You can tell the VNA to export the uncorrected Raw Data, and you can export the exact 12 Error Terms. Metrologists load this raw data into MATLAB or specialized software (like Metas VNA Tools) to run their own custom, proprietary error-correction algorithms, completely bypassing the VNA's internal math engine to achieve even higher precision or specialized non-linear corrections.