Beyond Fast Fault Detection: Current Sensing for AI Data Center SSCBs

This article is published by EEPower as part of an exclusive digital content partnership with Bodo’s Power Systems.

Article co-authored by Asahi Kasei Microdevices’ Yosuke Koiwa, Takahisa Shikama, Takaya Higa, and Taiga Kyosaki.

Artificial intelligence is driving an unprecedented increase in data center power consumption. As GPU clusters scale from hundreds of kilowatts to multi-megawatt installations, conventional AC power architectures face growing challenges in efficiency, scalability, and fault management. To support this transition, leading power infrastructure companies are evaluating solid-state transformers (SSTs), solid-state circuit breakers (SSCBs), and high-voltage DC (HVDC) distribution architectures that can deliver power more efficiently to future AI factories.

Image used courtesy of Freepik

Within these next-generation power systems, current sensing is becoming a critical enabling technology. While industry discussion often focuses on SiC and GaN power semiconductors, the effectiveness of power conversion, fault protection, and system reliability ultimately depends on how accurately and how quickly current information can be acquired and processed. For SSCBs in particular, current sensors are no longer simple monitoring components. They are becoming part of the protection architecture itself.

SSCBs Introduce New Requirements for Current Measurement

Traditional industrial systems often treat current measurement and protection as separate functions. Measurement supports control, energy monitoring, efficiency optimization, and diagnostics, while protection is implemented through relays, fuses, or dedicated protection circuits. SSCBs fundamentally change this relationship because semiconductor switches must interrupt fault currents directly. Figure 1 illustrates a typical SSCB architecture in which the current sensor serves both protection and monitoring functions.

In DC systems, fault currents can rise rapidly, and there is no natural zero-crossing as in AC systems. Therefore, SSCBs must detect abnormal conditions and react within microseconds. Furthermore, in AI data centers, increasing demand for monitoring rapid load fluctuations of GPU clusters and maintaining power quality is placing new requirements on current sensing.

Consequently, current sensors are expected not only to support protection functions, but also to provide measurement data that enables greater visibility into overall system behavior. Rather than simply detecting overcurrent conditions, it is becoming increasingly important to capture high-fidelity current waveforms. Therefore, the current sensor must support two objectives simultaneously: accurate current monitoring for control and diagnostics, and fast fault detection for system protection.

Figure 1. Current sensing for SSCB. Image used courtesy of Bodo’s Power Systems [PDF]

Challenges and Opportunities in Current Sensing

Historically, designers have had to trade measurement accuracy against response speed. Improving accuracy often requires filtering and signal conditioning, which may reduce responsiveness. Conversely, optimizing only for speed can sacrifice measurement quality. AI data center power systems make this compromise increasingly difficult to accept.

To address this challenge, many protection-oriented current sensors implement a dedicated overcurrent detection path based on an internal comparator. By decoupling protection functions from the measurement signal path, this approach achieves fast response times while maintaining accurate current measurement capabilities. This approach is simple and fast and remains effective for many conventional applications.

However, future SSCB developers increasingly view protection algorithms as a source of system-level differentiation. Advanced protection schemes may consider not only current magnitude but also current slew rate, transient behavior, accumulated energy, operating modes, and predictive fault indicators. Figure 2 compares these two approaches.

Conventional protection architectures rely on an internal comparator that converts the measured current into a binary fault signal, whereas high-bandwidth current sensing preserves the waveform information and allows protection decisions to be implemented externally using programmable logic.

A fixed comparator output can indicate that a fault has occurred, but it does not preserve the underlying waveform information. Once the sensor converts the current signal into a binary fault flag, valuable information before, during, and after the event can be lost. For next-generation SSCBs, the challenge is therefore not only detecting faults quickly, but also retaining access to high-fidelity current information that can support advanced protection strategies.

Solution: High-Bandwidth Current Sensing

Modern magnetic current sensing technologies are enabling designers to combine high accuracy, high bandwidth, fast transient response, galvanic isolation, and compact form factors within a single sensing solution.

This capability is particularly valuable in AI data center power systems, where SSCBs may be deployed throughout the distribution network and where downtime costs are extremely high. A high-bandwidth current output allows transient events to be observed accurately while maintaining the measurement performance required for efficiency monitoring, diagnostics, and system control. As a result, designers no longer need to compromise between measurement accuracy and response speed. Modern sensing technologies make it possible to achieve both simultaneously.

Access to high-fidelity current waveforms enables a fundamentally different approach to protection. Rather than relying solely on fixed comparator thresholds, protection decisions can be implemented externally using FPGA logic, digital protection controllers, DSPs, microcontrollers, or dedicated comparator circuits.

This allows system designers to develop, optimize, and continuously refine application-specific protection algorithms. Protection therefore evolves from a fixed hardware function into a programmable capability that can adapt to changing system requirements through software and algorithm updates. The value of this approach extends beyond protection itself.

The same current data used for programmable protection can also support diagnostics, predictive maintenance, digital twins, and software-defined energy management. In AI data centers, operators increasingly seek insight into power consumption during AI training and inference, load-profile changes during workload migration, repetitive transient events that may indicate degradation, and correlations between electrical behavior and system reliability.

Figure 2. Comparison of two methods of protection: (a) Conventional comparator-based protection with internal comparator current sensor; (b) Intelligent programmable protection with high-bandwidth current sensor. Image used courtesy of Bodo’s Power Systems [PDF]

As a result, current sensing is evolving from a simple protection function into a valuable source of system intelligence. Future SSCB architectures must support not only fast protection, but also effective utilization of high-fidelity current information throughout the power system.

Conclusion

The increasing adoption of SSCBs in AI data centers is redefining the role of current sensing. High-bandwidth current sensing technologies enable both accurate measurement and fast response while preserving access to high-fidelity current information. This allows protection decisions to be based not only on current magnitude, but also on transient behavior, current slew rate, accumulated energy, and other system-level information.

The value of this data extends beyond protection to diagnostics, predictive maintenance, digital twins, and power optimization. As a result, future SSCB designs will be defined not only by how quickly they detect faults, but also by how effectively they utilize current information to create more intelligent, flexible, and reliable power systems.

Technology Spotlight

AKM’s current sensor portfolio supports the design objective described in this article: combining accurate measurement with fast current visibility for solid-state protection. The CZ family is a compact, board-mount, fully integrated isolated current sensor solution that enables high-speed, high-accuracy analog current sensing. For SSCB designers, this type of analog current output supports both fast protection logic and accurate current monitoring without relying solely on a fixed overcurrent flag.

Figure 3. SSCB (eFuse) system diagram including the CZ39xx series current sensor. Image used courtesy of Bodo’s Power Systems [PDF]

The EZ family is a non-contact current sensor solution capable of measuring up to kiloampere-class currents flowing through external busbars. It features two independent analog outputs, providing greater flexibility in system design. For example, one output can be used for accurate current monitoring over a narrower current sensing range, while the other, covering a wider measurement range, can be dedicated to high-speed transient observation or protection processing by a microcontroller, FPGA, DSP, or external comparator circuit.

In both cases, the central value is not simply detecting an overcurrent condition internally. The value is preserving usable current information so that SSCB designers can implement their own protection algorithms, diagnostics, and future firmware improvements. For AI data center power systems, AKM enables fast protection and accurate measurement not only by flagging faults, but by giving designers current information they need to act on.

Figure 3 shows the proof-of-concept intelligent SSCB (eFuse) system, developed in collaboration with Silicon Austria Labs, to overcome the limitations of conventional thermal fuses.

The system integrates three protection mechanisms: Short-circuit protection (comparator), dI/dt detection (differentiator circuit), and Overcurrent protection (integrator circuit). These three detection circuits generate the optimal current interruption trigger according to the specific fault condition, enabling significantly more flexible protection control than conventional fuse-based solutions.

Figure 4. SSCB (eFuse) system characteristics: (a) characteristic curve of eFuse system; (b) short-circuit protection measurement. Image used courtesy of Bodo’s Power Systems [PDF]

This capability is enabled by AKM’s CZ Series current sensors, featuring a bandwidth of 4 MHz and a response time of 100 ns, allowing accurate detection of fast transient and surge-current events. Figure 4 shows the performance characteristics of the proof-of-concept SSCB (eFuse) system. In the short-circuit protection mode, the system detects an overcurrent condition and generates a current interruption trigger within less than 160ns from the overcurrent detection point. For lower current levels, a slower response is intentionally implemented to avoid unnecessary protection actions, providing a more selective protection behavior than the short-circuit protection mode.

This article originally appeared in Bodo’s Power Systems [PDF] magazine and is co-authored by Miho Onuma, Technical Expert of Current Sensors, Asahi Kasei Microdevices Corporation; Yosuke Koiwa, Technical Expert of Current Sensors, Asahi Kasei Microdevices Corporation; Takahisa Shikama, Strategy & Business Leader of Magnetic Sensor Products, Asahi Kasei Microdevices Corporation; Takaya Higa, Field Application Engineer, Asahi Kasei Microdevices Europe GmbH; Taiga Kyosaki, Field Application Engineer, AKM Semiconductor, Inc.