A high-speed converter captures brief power-rail disturbances in AI data centres, enabling continuous remote monitoring, faster fault diagnosis, improved power reliability and reduced downtime in computing infrastructure.
AI data centres can experience costly interruptions when brief power-supply disturbances escape conventional monitoring systems. A new high-speed analogue-to-digital converter (ADC), the ADC34DSO75 from Texas Instruments (TI), addresses this challenge by capturing transient electrical events at nanosecond resolution, helping operators identify faults before they escalate into prolonged outages.
Operating at a sampling rate of 3 gigasamples per second (GSPS), the converter enables continuous, low-power, event-based remote monitoring of power rails. This allows data centre operators to detect short-lived voltage disturbances that may affect AI servers, accelerators and associated power-delivery systems.
Unlike conventional monitoring approaches that may require technicians to connect oscilloscopes or other diagnostic equipment on-site, the ADC34DSO75 supports remote event detection. Its combination of high-speed sampling and on-chip digital signal processing enables the capture of transient faults as they occur, improving visibility into power-system behaviour and simplifying troubleshooting.
The key features are:
- Supports event-triggered capture of transient electrical faults.
- Integrates advanced digital signal processing for fault analysis.
- Enables monitoring without conventional on-site oscilloscope setups.
- Designed for low-voltage, high-current power-rail telemetry.
- Targets continuous monitoring in hyperscale AI computing facilities.
The converter is particularly relevant to AI computing infrastructure, where high-current, low-voltage power rails must maintain stable operation despite rapidly changing workloads. Brief electrical glitches can affect system stability and contribute to unplanned downtime. Continuous monitoring helps engineering teams investigate these events, identify potential causes and improve power-delivery reliability.
The device also supports power-system optimisation by providing detailed information about electrical disturbances. Operators can use this telemetry to investigate power-quality issues, assess system behaviour under changing loads and improve the reliability of high-density computing installations. Remote diagnostics can reduce the need for physical intervention, which is especially useful in large-scale facilities containing numerous server racks and distributed power systems.
The ADC34DSO75 is part of a broader set of semiconductor solutions targeting AI data centre infrastructure. These include the LMI800PS isolated hot-swap controller, designed to improve high-voltage power delivery, and the LMK6B bulk acoustic wave oscillator, which supports low-jitter timing for high-speed data transmission.
Together, these technologies address three important infrastructure requirements: detecting fast electrical faults, protecting power-distribution networks and maintaining timing integrity across high-speed interconnects. As AI workloads increase demands on data centre power and networking systems, high-resolution telemetry can help operators improve fault visibility and reduce the risk of expensive service interruptions.
Click here for the original announcement.
