Battery-backed uninterruptible power supplies (UPSs) are invaluable in protecting sensitive equipment in data centers, medical facilities, factories, telecoms hubs, and even homes against short-term grid spikes and outages. In the case of longer outages, they provide necessary short-term power to allow a managed and orderly shutdown.
UPSs can be either ‘online’ or ‘offline.’ In an offline UPS, the load connects to the grid. When power fails, the battery switches in—this process can take around 10 ms to complete, limiting the suitability of this type of UPS for some applications. Online UPSs, however, connect the load to the grid via the battery, so no switching is required, supporting use in all applications.
Generally speaking, the modular approach is favored by designers and users, meaning that lower power UPSs (in the range of 10 kVA to 50 kVA) are combined to meet the application’s needs. This allows for the quick and easy sizing of solutions and benefits from economies of scale in manufacture.
However, as with any power design, there are challenges in designing UPSs. Some critical areas to be considered include size, input regulation, output, battery management, and topology.
Size is important, especially in applications such as data centers where space is at a premium. Because UPS does not store data, it does not directly contribute to revenue. Traditionally, the transformer has been one of the largest components, but with the availability of enhanced semiconductor technology, this can be omitted—thereby saving space. This technology is capable of delivering hundreds of kVA in a standard-size cabinet.
Because an online UPS uses high frequency (20 kHz to 40 kHz) PWM to perform a double conversion (from AC to DC and back again), there are a few issues with input glitches.
The inverter stage defines the UPS output, which also significantly impacts the overall efficiency of the UPS. The output is generally a sine wave similar to the grid voltage, and if sensitive equipment connects to the UPS, it requires the generation of a pure sine wave. This requires high-frequency operation of the switching devices (IGBTs/MOSFETs) in the output stage and filtering to reduce any EMI generated by the switching.
In a typical UPS, multiple stacked batteries create a string managed by a battery management module. For the batteries to perform at their peak and have an extended lifetime, the design must ensure adequate load balancing, voltage, and current protection, charge and discharge control, thermal management, fan control, monitoring, and communications.
The most crucial decision in designing a UPS is the selection of the most appropriate topology for the application, balancing performance and cost. While UPS design sometimes uses a two-level topology such as a three-phase half-bridge (6-switch), three-level topologies (T-NPC, A-NPC, or I-NPC) generally offer better efficiency and reduce EMI.
The choice of semiconductor material for the switches and diodes is also crucial, with newer wide bandgap (WBG) materials such as silicon carbide (SiC) offering greater efficiency and enhanced robustness with higher operating frequency and temperature.
onsemi has produced a technical white paper that looks in detail at the design challenges and considerations for UPSs, including an overview of UPS types and an explanation of the critical specifications. It considers the various topologies available and describes the tradeoffs that drive decision-making in the design process. The paper concludes with an overview of the new SiC devices (and associated drivers) that every modern power engineer should consider.
You find onsemi’s UPS design tips here.