**3. Experimental Setup**

The experimental setup was located in a completely shielded room in order to achieve a high signal-to-noise-ratio (SNR), necessary for the measurement of low currents.

#### *3.1. Generation of High DC Voltage with Low Ripple*

Investigating the discharge physics under DC voltage stress requires a DC voltage with low ripple. This is necessary because a ripple factor *δU* of a few percent can lead to a phase dependent concentration of PD impulses in the voltage maximum and would therefore influence the PD behavior significantly [35].

For this reason, a symmetric Greinacher voltage doubler circuit (Figure 3) was used to generate high DC voltages [4]. It was fed with a power frequency *f* = 50 Hz and loaded by a current *I*. In contrast to common Greinacher circuits with a ripple of *δU* (Equation (4)), the smoothing capacitors *C*s were charged every half-cycle, leading to a lower ripple *<sup>δ</sup>U*symm according to Equation (5).

$$
\delta \mathcal{U} = \frac{I}{2 \cdot f \cdot \mathcal{C}\_s} \tag{4}
$$

$$
\delta \mathcal{U}\_{\text{symm}} = \frac{I}{4 \cdot f \cdot \mathcal{C}\_{\text{s}}} \tag{5}
$$

Hence, the ripple could be reduced by a factor of two. In addition, a high smoothing capacitance *C*s was used to further reduce the voltage ripple. In contrast to usual realizations of the symmetric Greinacher circuit [4], two high voltage transformers with a primary voltage shifted by 180° were used in this investigation.

Due to the maximum reverse voltage of the rectifiers used, the maximum output voltage of this voltage doubler circuit was limited to *U*DC max = ±250 kV. The measured voltage ripple was approximately 250 V at its maximum (*δU*symm ≤ 0.1%). The voltage measurement was performed using a calibrated ohmic voltage divider with a Highvolt MU17 peak voltmeter. A resistor *R*d and an inductance *L*d were placed in between the DC voltage supply and the test object *C*p in order to limit the current in case of a breakdown.

**Figure 3.** Circuit diagram of a symmetric Greinacher voltage doubler circuit for the generation of high DC voltage with low ripple.

#### *3.2. Electrode Arrangement and Test Vessel*

The high DC voltage supply was connected to the gas-insulated test vessel using a SF6-air bushing (Figure 4a). The test vessel used was a commercially available part of a 420 kV GIS with a sandblasted encapsulation. It allowed investigations up to an absolute gas pressure of 0.7 MPa.

**(a)** Photo of the exerimental setup with ① symmetric Greinacher circuit and ② test vessel.

**(b)** Weakly inhomogeneous electrode arrangemen<sup>t</sup> with protruding needle.

In order to model the electric field of a real gas-insulated system, a weakly inhomogeneous electrode arrangemen<sup>t</sup> made of aluminum was placed inside the test vessel (Figure 4b). The gap distance between the half-sphere and the plate could be varied between *d* = (0...100) mm. Hence, the degree of homogeneity *η* could be varied from 0.88 to 0.37. For the investigations presented in this paper, the gap distance was fixed at *d* = 60 mm. This resulted in *η* ≈ 0.52, which was close to real applications [33]. In order to investigate the PD behavior of the insulating gases, a protrusion was placed in the middle of the half-sphere. The used needles were made of 100Cr6 steel and had a tip radius of *r*i ≈ 22 μm. The length of the needle was *l* = 5 mm. To measure the partial discharge current *I*PD, the needle was separated from the grounded sphere using an insulating PTFE ring and directly connected to an SMA adapter. The polarities mentioned in this investigation refer always to the polarity of the protrusion.
