The physical characteristics of a measurement set-up using a VNA inherently give rise to imperfections in the raw data output. An ideal calibration would cancel out the effect of these imperfections. The imperfections can be found in these parameters:
- Match: because it is a broadband instrument, the VNA offers a raw match which can lead to errors of more than 1-dB. Correcting mismatch will greatly reduce the size of this error;
- Directivity: the directional coupler on a VNA allows the separation of the incident signal to the DUT from the reflected signal from the DUT. Even though VNA manufacturers use high-quality couplers, there will always be some residue of unintended coupling between the signal paths. This can have an effect on tests which are intended to measure reflected signals of very low amplitude;
- Frequency response: when calibration cables are used for measurement purposes, they can have an effect on the overall frequency response. Again, these imperfections must be calibrated out.
The purpose of calibrating a test set-up, then, is to nullify the effect of match, directivity and frequency response errors. In fact, this kind of calibration would better be termed a ‘vector error correction’; it is quite different from the periodic (normally annual) manufacturer’s calibration required to maintain the quality rating of a VNA.
In turn, the manufacturer’s calibration is actually a verification. The procedure requires the VNA to be calibrated using known calibration standards. Ideally, these should be dedicated calibration standards for the VNA in question, as supplied in the manufacturer’s calibration kits. The calibration will also use customer-owned calibration cables, if they are available. The VNA and cables are then tested to specified verification standards, and S-parameter measurements are taken.
If the results meet the desired specification, then the test has verified that the VNA, calibration standards and calibration cables are good.