After a full two-port calibration there is always residual error, and its size is set by the uncertainty of the standards you calibrated with. A measurement quoted without that bound is a number, not a result.
of VNA error: systematic, random, and drift or stability. Calibration corrects the first. The other two persist and have to be budgeted. Copper Mountain Technologies
loaded incorrectly is a silent failure. The instrument then reports a confident, corrected, wrong number. Tektronix, RF calibration
at the connector move the reference plane. Mechanical discipline is part of the measurement, not preparation for it. Tektronix, RF calibration
None of this is exotic. It is bookkeeping on error terms. But it is bookkeeping that gets skipped, and the failure mode is a corrected measurement that looks authoritative and is not.
Six that close the biggest gaps. 5 in the full rf and electrical engineering library.
The gap it closes. Applies the error-correction model explicitly and carries residual uncertainty through, so the measurement arrives with its bound attached rather than as a bare number.
See it run in your browser →The gap it closes. Friis makes the first stage dominant, so a gain or noise-figure error early in the chain propagates to the whole receiver. Per-stage contribution shows where the budget actually goes.
See it run in your browser →The gap it closes. A link budget is a long chain of gains and losses where one sign error or one omitted term changes the margin, and the total alone will not reveal it.
See it run in your browser →The gap it closes. Closed-form L-section solutions come in pairs with different bandwidth and component-value consequences. Picking one without seeing both is a coin flip.
See it run in your browser →The gap it closes. Touchstone files carry reference-impedance and format conventions that silently misread if assumed, turning a good dataset into a wrong plot.
See it run in your browser →The gap it closes. Substrate permittivity and thickness set characteristic impedance, and a board that arrives on a different stackup than assumed is a respin.
See it run in your browser →These implement published closed-form models and standard error-correction mathematics. They do not replace a calibrated instrument, a measurement, or an accredited calibration laboratory. They let you check a budget, a conversion or an uncertainty before and after you take the measurement.
Every tool above runs entirely in this browser tab. Nothing is uploaded, nothing is sent to a model, and each result cites the rule it applied.
Free tier, no card, no install. An account saves your work and lets you export a citable record of a run. See all 5 rf and electrical engineering tools.