Accurate modelling of grid-connected converters is essential for system-level stability analysis in power systems with large penetration of inverter-based resources. This paper applies robust control theory to define a matrix relative error that quantifies the discrepancy between modelled and measured converter impedance. Properties of the matrix relative error and the impact on uncertainty bounds in stability analysis are investigated. The validation approach is demonstrated on a Siemens Gamesa offshore wind turbine converter impedance obtained from an analytical state-space model, an electromagnetic-transient (EMT) simulation and test-rig measurements. The EMT and test-rig converter impedance is estimated empirically using a frequency-scan method applied at a selected operating point. The results show that the model and measured converter impedance match within 10 \% matrix relative error for most frequencies. Derived uncertainty bounds are illustrated through a weak-grid stability study, highlighting the importance of applying robust control techniques to manage model uncertainty.