Linearity extension of UHV class lightning impulse dividers up to 3000 kV

Recent research has raised questions regarding the validity of the current linearity extension methods for lighting impulse voltages beyond 1500 kV. Factors identified include the unexpectedly large influence of proximity, corona and generator source impedance. The lowest uncertainty claimed by National Metrology Institutes for lightning impulse reference measurement systems up to 1000 kV is 0.5 % for the test voltage value U_t, and 1 % for both the front time T₁ and the time to half value T₂. Using state-of-the-art methods for a linear extension, earlier research has reported U_t non-linearities of 1 % up to 2700 kV, and more than 3 % above 3000 kV. The non-linearities in front times were significantly larger, and the reasons for that are not understood. A three-week measurement campaign was carried out at the high voltage laboratory at TU Delft in October 2022 to tackle the problem. The campaign started with calibration of nine measuring systems at 400 kV: five based on resistive divider with maximum voltages ranging from 500 kV to 2000 kV, and four based on damped capacitive divider with maximum voltages from 1000 kV to 4000 kV. In the second step these dividers were compared with each other in a step-up procedure while measuring the charging voltage of the generator.