Partial Discharge (PD) measurements are of great importance to enable the monitoring and diagnostics of HV systems. The requirements of the Paris Agreement and climate goals have fuelled the increase in penetration and demand of HVDC for offshore wind. The HVDC Gas Insulated Switchgear (HVDC GIS) is a reliable technology to support the necessary electrical infrastructure. Nevertheless, some in-service failures may occur. These failures can occur in the insulation system and thus developing a measurement system for PD detection is essential for monitoring and diagnostics. To monitor and diagnose the HVDC GIS, a novel Magnetic Antenna (MA) is being developed to operate in the high-frequency (HF) (30-300 MHz) range. The well-established UHF method for the GIS is typically used due to its high sensitivity and resilience to electromagnetic interference. However, the UHF method is unable to calibrate to apparent charges as this information is in the low frequency (up to 30 MHz) until HF range. The knowledge of charge calibration indicates the discharge type which is important in DC as DC does not have phase-resolved information as with AC. The appropriate frequency range of the MA should enable the measurements of the apparent charge and localize the defects when monitoring and diagnosing a HVDC GIS setup. The overarching goal is to develop a measurement system to measure PDs in the HF range in a GIS setup. For this purpose, MAs are created and investigated. A workbench has been built and developed to characterize the MAs and measure its frequency characteristics. A 380 kV GIS measurement setup has been developed. This enabled the measurement and acquisition of data of the discharges using the MAs. The Threshold Peak detection (TPD), Energy Criterion (EC), and Phase Method (PM) localization methods are investigated and implemented for localization of the source of defects. The PM is unable to localize the pulse due to its sensitivity to noise and reflections. The TPD and EC are both suitable with the TPD being the preferred method due to its 95% accuracy of localizing the defect within ± 1.5 m.