Grid-forming converters play a key role in grids with a high share of renewable energies. Clear definitions and verification procedures for grid-forming behaviour are a fundamental requirement for large-scale use of this technology. They ensure that grid codes are implemented in such a way that the converters behave in the most beneficial way for the grid. This study demonstrates how the core characteristic of grid-forming behaviour - the voltage source behaviour - can be conclusively verified. Measurements with real converters, which are considered as black boxes, are used for this purpose in order to guarantee the manufacturers maximum freedom to achieve the grid-forming properties.

A freely programmable converter with a rated power of 1 kVA is used for the measurements on a hardware test bench. The converter is tested with four different controls. Two of these are state-of-the-art grid-forming controls with different voltage controls. The other two cases form the reference group by using a grid-following control method. First with fixed setpoints and then with additional, higher-level P(f) and Q(V) controls. This allows the voltage source behaviour of grid-forming controls and the current source behaviour of grid-following controls to be compared to each other.

The novel method used in this paper differs from other approaches as the only input for the verification is the voltage measured at the converter's point of connection that is converted into a space vector in a rotating reference frame. No internal signals or knowledge of internal parameters are required. The identified voltage space vector, which is formed from the three-phase voltage, can be used to clearly determine whether the converter behaves like a controlled current or voltage source behind the filter impedance.

The scenarios used to determine this are on the one hand synthetic and on the other hand also feasible in field measurements on real systems. In the simpler method, the grid emulator in the laboratory provides a jump in the angle or amplitude of the grid voltage. Depending on how dynamically the voltage space vector follows this change in the voltage, it is possible to determine which converter behaviour is given.

The second approach generates the change in state by connecting or disconnecting serial or parallel impedances. These impedances could be provided in field measurements using fault-ride-through (FRT) testing equipment that is already in use.

The contribution of this study therefore lies in the application-orientated realisation of hardware tests. The aim is to make a novel proposal as to how grid-forming properties can be easily and reliably verified, while also preserving the manufacturer's freedom of implementation.