Background
With the global surge in AI data center construction and NVIDIA’s promotion of 800V HVDC power architecture, DC 800V Busbar Trunking Monitoring systems are becoming the new standard for next-generation AI power distribution.
The Middle East is accelerating AI infrastructure deployment—the UAE has unveiled a 5GW AI campus, and Saudi Arabia has announced a $5 billion NEOM AI Data Center.
This solution is specifically designed for the DC 800V small busbar power architecture, enabling comprehensive electrical monitoring from Feeder to Tap‑off.
Pain & Value
Your Pain Points
High-density AI racks create unprecedented power demands, yet the overall health of the busway power distribution system remains unknown, with no real-time early warning capability.
Multiple meters are required to separately monitor voltage, current, temperature, and leakage, resulting in high costs and complex supplier management.
Numerous tap-off boxes are widely distributed, cabling is complex, and fault localization is difficult in large-scale deployments.
Our Acrel Value
AMB100-D-HV provides a “monitor-analyze-warn-locate” four-in-one system, delivering end-to-end transparency from feeder to tap-off and enabling risk foreseeability.
One AMB100-D-HV integrates all functions (voltage, current, power, energy, 4×NTC temperature, 2×leakage, 4×DI, 4×DO) – one-stop procurement.
Supports LAN, RS485, and ring redundancy networking modes, flexibly adapting to various deployment scenarios for reliable and maintainable communication.
Solution
HVDC 800V Busbar Trunking System Monitoring for AIDC
Device Layer
AMB100-D-HV instruments are installed in feeder boxes and tap-off boxes, collecting voltage, current, power, energy, 4-channel NTC temperature, 2-channel leakage, 4 DI, and 4 DO. Each instrument features dual LAN + dual RS485 interfaces.
Network Layer
Supports three networking modes:
(1) LAN downstream collection – high speed, suitable for large-scale tap-off boxes;
(2) RS485 downstream collection – low cost, compatible with legacy equipment;
(3) RS485 ring redundancy – self-healing upon link failure, ideal for critical loads.
Management Layer
The feeder box aggregates data from all tap-off boxes via LAN2 or RS485-2 to the environment monitoring system (EMS). Supports Modbus TCP/RTU, HTTP WebServer, NTP time synchronization, and local HMI via LAN3.
AMB100-D-HV delivers comprehensive, reliable, and flexible monitoring for DC 800V busway systems, fully supporting the three networking modes for AI data center deployment.
Networking
01. High‑speed Ethernet + dual upstream
Application: New data centers with many tap‑off boxes (>32), requiring fast data refresh (≤1s); environment monitoring system needs both IP and serial interfaces.
Tap‑off box requirement: Each tap‑off box instrument must have an Ethernet port (AMB100‑D‑HV has dual LAN).
Cabling: Network cable, ≤100m per segment, can be extended via switches.
High Speed
Dual Uplink
Easy Expansion
Mixed Protocol
02. Traditional RS485 + dual upstream
Application: Retrofit projects with many existing RS485 tap‑off boxes, limited budget, low sensitivity to downstream speed.
Tap‑off box requirement: Each tap‑off box instrument needs only one RS485 port (the instrument has 2 RS485 ports, use one).
Cabling: Twisted pair, max 1200m, daisy‑chain connection.
Low Cost
Long Distance
Simple Cabling
Legacy Friendly
03. Ring redundant RS485 + IP‑only upstream
Targets:Sudden arc flash faults inside the cabinet.
Application: Critical load circuits (e.g. AI training clusters, core network equipment) requiring extremely high communication link reliability, no single point of failure allowed, and environment monitoring system is fully IP‑based.
Tap‑off box requirement: Each tap‑off box instrument uses both RS485 ports (one in, one out) to form a physical ring.
Cabling: RS485‑1 and RS485‑2 need to be wired separately to form a loop; slightly more complex, but fault self‑healing time is milliseconds.
Ring Redundancy
Self Healing
High Reliability
IP Only
Application Scenarios
Abu Dhabi’s 5GW Stargate AI campus (first 200MW in 2026). Our solution provides real-time DC 800V busway visibility for massive GPU clusters, preventing downtime.
NEOM’s 1.5GW AI factory with DataVolt (2028). HUMAIN targets 1.9GW by 2030. End-to-end busway monitoring delivers reliability for sovereign AI.
Gulf legacy data centres face power capacity limits. AMB100-D-HV supports RS485 downstream collection for seamless, cost‑effective migration without full overhaul.
Dubai & Abu Dhabi rank top data centre markets. New 5G Industry 4.0 zones across Qatar & Saudi require compact, reliable power monitoring – our flexible networking modes fit perfectly.
Make Power Visible, Risks Predictable, and O&M Smart.
Contact us to build your customized, high‑reliability
DC 800V Busway Monitoring solution.
