In lightning impulse tests for DC cables

the measured voltage waveform has abnormal jitter and does not accord with the standards. To solve this problem

we theoretically analyzed the impulse test discharge circuit with its impulse voltage generator connecting firstly to the voltage divider and then to the test object（defined as "Way Ⅰ"）. We also simulated the circuit in Way I in Pspice

and compared and simulated the waveforms obtained from the capacitive voltage divider and the test object. The analysis indicates that the lead inductance between the capacitive voltage and the test cable leads to the deviance of the measured waveform from the actual waveform or even the abnormal jitter

if the lead inductance is large enough. Hence

we proposed a circuit that has the impulse voltage generator connected first to test object and then to voltage divider（defined as "Way Ⅱ"）. To verify Way Ⅱ

using various lead inductances

we compared the coincidence degree between the measured waveforms using the two circuits and the actual waveform

as well as the measuring error of some time parameters（front time

time to half-value）. The result indicates that using Way Ⅰ

with increasing lead inductance

the measured waveform deviation

the amplitude of abnormal jitter

and the measuring error of time parameters all increases. However

using Way Ⅱ

even with a relatively high lead inductance

the measured waveform has no abnormal jitter and it is nearly consistent with the actual one

while the measuring error is small. In addition

we conducted the experimental lightning impulse tests using Way Ⅱ. The obtained test waveform has no abnormal jitter and it is consistent with the simulation result.