**5. Results**

Following, the measurement results are presented. As a summary, a classification of the occurred partial discharge phenomena is derived that helps to improve the understanding and interpretation of PD measurement at a protrusion under DC voltage stress. Therefore, the impulse current amplitude ˆ *I*PD, the time difference between two subsequent sequences Δ*t*, and the pulseless partial discharge current *I*mean were analyzed in dependence of the gas pressure and the applied voltage.

#### *5.1. Inception Voltage (Voltage Rising Test)*

In order to determine the inception of current impulses and the pulseless PD current in dependence of the polarity of the protrusion and the gas pressure, a voltage rising test (VRT) was performed (Figure 8). The rate of voltage increase was set to 0.5 kV · s<sup>−</sup>1.

**Figure 8.** Inception of current impulses and pulseless currents in dependence of gas pressure and the polarity of the protrusion in SF6.

The measurements determined different voltages for the inception of impulse currents and pulseless PD currents. At a negative protrusion, these voltages were very similar. Only slight differences occurred due to measurement uncertainties and the statistical behavior of the discharge inception. At a positive protrusion, the inception of pulseless PD currents took place at higher voltages than the inception of current impulses. That led to the conclusion that a pulseless partial discharge current, probably a superposition of many small electron avalanches, was superimposed on the PD impulses and emerged in dependence of the polarity and the electric field strength.

The inception voltage of the current impulses at a positive protrusion was almost twice the inception voltage determined for a negative protrusion. These differences could be explained with the different supply of starting electrons in dependence of the polarity of the protrusion. Due to field emission processes from the cathode, more electrons were present in front of the negative protrusion [19]. The starting electrons at the positive protrusion had to be generated by collision detachment of negative ions close to the needle electrode [38]. Another fact influencing the growth of a PD impulse in the vicinity of a protrusion was the direction of movement of electrons and ions. Electrons at a positive protrusion were accelerated in the direction of the protrusion, whereas electrons at a negative protrusion were accelerated in the direction of the oppositely charged electrode. This underlined the polarity dependencies. The changing gas pressure had no influence on the described polarity effects.

An increased gas pressure led to an increased inception voltage for both polarities, which corresponded to the assumptions derived in the Physical Fundamentals Section. The effective ionization coefficient *α*¯ decreased with an increasing gas pressure (Figure 1).

In the following sections, the presented results always refer to the inception of PD current impulses for each polarity-pressure combination, which was thus determined as the inception voltage *U*i in this investigation.

#### *5.2. Partial Discharge Impulse Current*

In the following, the investigations of the PD current impulses, in particular the impulse current amplitude ˆ*I*PD and the time difference between subsequent sequences Δ*t*, are described. The investigation of the positive protrusion could only be carried out with a limited voltage, since higher voltages were too close to the breakdown voltage with a high risk of damaging the measurement equipment.

#### 5.2.1. Amplitude of Partial Discharge Impulses

The amplitude of the partial discharge impulses depended on the polarity of the protrusion, the gas pressure, and the electric field strength (Figure 9).

In principle, a higher gas pressure led to a lower amplitude of the impulse current. The amplitude of the impulse current was in the same range for both polarities. For negative polarity of the protrusion and high gas pressures (*p*SF6 ≥ 0.5 MPa), the partial discharge current amplitudes remained almost constant. At a gas pressure of 0.1 MPa, an increased voltage led to an increase of the amplitude of the impulse current in one order of magnitude. If the voltage was increased further up to a voltage of *U* ≈ 2.6 · *U*i, no current impulses could be measured any longer, whereas the measurement with the transimpedance amplifier was still showing a pulseless current (cf. Section 5.3). Obviously, the partial discharge physics had changed. It could be assumed that a pulseless glow discharge built up at the protrusion. In the investigated pressure-voltage range, this behavior could only be observed under a gas pressure of 0.1 MPa and a negative polarity of the protrusion.

When increasing the voltage at a positive protrusion and a gas pressure of 0.5 MPa, a significant increase of the amplitude of the impulse current could be measured.

**Figure 9.** Mean value and maximum/minimum of the amplitude of the impulse current in SF6 depending on the voltage polarity and the gas pressure.

#### 5.2.2. Time Difference between Subsequent Sequences

The observed time differences between subsequent sequences varied in a huge range between a few microseconds and several minutes (Figure 10). The scatter for almost every voltage-pressure constellation was rather high. Due to the breakdown danger, it was not possible to perform measurements at 0.7 MPa for a positive protrusion.

In general, the time differences between subsequent sequences at a positive protrusion were smaller than at a negative one, and a higher voltage led to a decreased time between subsequent sequences. At a gas pressure of 0.1 MPa, a higher voltage led to a higher time difference between subsequent impulses. The space charge in front of the protrusion became more stable, preventing the inception of further impulses. Above 2.6 · *U*i, no impulses could be measured any longer at the negative protrusion. As known from the measurements under Section 5.3, only a pulseless current occurred. No statement can be given for a positive protrusion above 2 · *U*i, since the voltage could not be increased further due to the danger of a breakdown. At a positive protrusion and a gas pressure of 0.5 MPa, a remarkable change of the PD behavior could be observed, when increasing the voltage from 1 · *U*i to 2 · *U*i: the amplitude of the impulse current increased by one order of magnitude (cf. Figure 9), whereas the time between subsequent sequences decreased by approximately seven orders of magnitude (Figure 10). Considering the time dependent PD current (Figure 11) at inception voltage, only one small current impulse could be measured per sequence (Figure 11a); whereas at 2 · *U*i, one high current impulse in the range of several milliamperes was followed by several small current impulses with an amplitude of several hundred microamperes (Figure 11b).

Every recorded current impulse was followed by smaller impulses. The time differences <sup>Δ</sup>*t*imp between the first high impulse and the subsequent smaller impulses was in the range of several hundred nanoseconds (Figure 11c); its minimum was defined according to Table 2. A higher gas pressure led to lower time differences. At a gas pressure of 0.7 MPa, these subsequent impulses already occurred at inception voltage, whereas at gas pressures of 0.1 MPa and 0.5 MPa, this behavior could only be observed at twice the inception voltage. At gas pressures of 0.1 MPa, only approximately 5% of the recorded sequences contained these subsequent impulses.

This transition from a streamer discharge to a streamer discharge with subsequent impulses was in accordance with the literature [6,32,33,38]. The subsequent impulses were probably generated in the discharge channel of the first discharge streamer. Due to the present space charge of the first discharge, the critical volume decreased, leading to a smaller amplitude of the subsequent impulses [32]. This behavior occurred mainly at a positive protrusion due to the different drift velocities of ions and electrons (cf. Table 1), resulting in a different space charge distribution compared to a negative protrusion.

**Figure 10.** Time differences between subsequent sequences in SF6 depending on the gas pressure and the applied voltage.

**Figure 11.** Comparison of time dependent partial discharge current and time differences <sup>Δ</sup>*t*imp between subsequent PD events at a positive protrusion in SF6.

At a negative protrusion and gas pressures *p* ≥ 0.5 MPa, the time difference between subsequent sequences decreased by three orders of magnitude (Figure 10), and the amplitude of the impulses remained almost constant with increasing voltage (Figure 9). A behavior such as under positive voltage could only be observed for less than one percent of the recorded sequences at higher gas pressures (*p*SF6 ≥ 0.5 MPa). The occurrence of these subsequent impulses at a negative protrusion became more probable with higher voltages.

#### *5.3. Pulseless Partial Discharge Current*

Measurements with the transimpedance amplifier showed the pulseless direct current share of the discharges (Figure 12).

This current was mainly related to the movement of slow positive and negative ions, which drifted along the electric field lines. The amplitude of the pulseless PD current was magnitudes below the impulse current amplitudes. As described in Section 5.1, the inception of the pulseless current was dependent on the polarity of the protrusion. At a negative protrusion, pulseless currents and impulse currents incepted at once at the same voltage during a VRT; whereas at a positive protrusion, the pulseless current incepted at almost twice the inception voltage of the impulse current. An increased voltage led to a higher pulseless PD current, regardless of the polarity of the protrusion and the gas pressure. A higher gas pressure led to a lower pulseless partial discharge current, but the differences at a negative protrusion were less pronounced compared to a positive protrusion.

**Figure 12.** Pulseless partial discharge current dependent on the voltage, the polarity of the protrusion, and the gas pressure (limit of the measureable current: 100 μA).
