{"id":3753,"date":"2026-09-11T03:00:18","date_gmt":"2026-09-11T03:00:18","guid":{"rendered":"https:\/\/acrel.pl\/?post_type=case&p=3753"},"modified":"2026-09-11T03:00:37","modified_gmt":"2026-09-11T03:00:37","slug":"application-of-the-acrel-aim-t500l-insulation-monitoring-device-in-dc-800v-systems-of-ai-computing-centers","status":"publish","type":"case","link":"https:\/\/acrel.pl\/ru\/case\/application-of-the-acrel-aim-t500l-insulation-monitoring-device-in-dc-800v-systems-of-ai-computing-centers\/","title":{"rendered":"Application of the Acrel AIM-T500L Insulation Monitoring Device in DC 800V Systems of AI Computing Centers"},"content":{"rendered":"\n
With the explosive growth of artificial intelligence and the computing economy, AI data centers (AIDCs) are placing ever-higher demands on power supply systems. The conventional 400 V low-voltage AC distribution architecture has reached its physical limits in the face of high-density GPU server clusters, and the 800 V high-voltage DC (HVDC) power supply architecture based on solid-state transformers (SST) is becoming the industry consensus. With its higher voltage class and lower line losses, the 800 V DC system provides power support for megawatt-class high-density cabinets.<\/p>\n\n\n\n However, while improving energy efficiency, HVDC systems also bring new safety challenges. The 800 V DC bus operates in an unearthed (IT) mode, with neither the positive nor the negative pole directly earthed. In such a system, a single earth fault is extremely difficult to detect \u2014 the system may keep running in a faulty state without being noticed. Once a second earth fault occurs, a short-circuit loop is formed instantly, which can cause equipment fire, shutdown, or even the outage of the entire computing cluster. The team standard T\/CMEEEA 261-2026 Technical Requirements for HVDC Systems in Data Centers explicitly requires 800 V DC systems to be equipped with online insulation monitoring devices.<\/p>\n\n\n\n The Acrel AIM-T500L DC insulation monitoring device is designed specifically for such scenarios. It is suitable for DC 0~800 V and AC 0~690 V or lower voltage systems. Together with the ASG200 test signal generator and the AIL200-12 insulation fault locator, it can also locate fault circuits and is compatible with DC systems of 800 V and below. This article describes its core value in DC 800 V and lower-voltage systems of computing centers from three dimensions: technical principle, system application, and electrical design.<\/p>\n\n\n\n 2.1 Characteristics of the 800 V DC Power Supply Architecture<\/strong><\/strong><\/p>\n\n\n\n The 800 V DC power supply system of a computing center usually uses a solid-state transformer (SST) to convert 10 kV medium-voltage AC into 800 V DC, which is then distributed to GPU server cabinets through DC distribution cabinets. The system adopts a three-wire (POS\/RTN\/PE) bus architecture, with neither pole earthed.<\/p>\n\n\n\n 2.2 Insulation Safety Risk Analysis<\/strong><\/strong><\/p>\n\n\n\n In an 800 V HVDC system, insulation safety faces multiple challenges:<\/p>\n\n\n\n First, the high-frequency electromagnetic environment accelerates insulation aging. The SST uses SiC power devices and high-frequency PWM modulation technology. The high-frequency, high-voltage, and high dv\/dt operating conditions impose continuous stress on insulating materials and accelerate insulation aging.<\/p>\n\n\n\n Second, DC earth faults are highly concealed. When a single earth fault occurs, the system can still operate normally, making it hard for conventional protection devices to detect, and manual inspection is even less likely to find it.<\/p>\n\n\n\n Third, HVDC faults have severe consequences. Once a double earth fault occurs, the short-circuit current may instantly destroy power modules and cause the outage of the computing cluster, potentially leading to major losses.<\/p>\n\n\n\n
II. Special Requirements of DC 800V Systems in Computing Centers<\/strong><\/strong><\/h2>\n\n\n\n