Data centers have already become the backbone of our new digital economy. Not only do they provide the computing power, storage and networking resources required by emerging technologies, they also make many of today’s most critical tasks possible. 

But as a new era of connected systems emerges — one enabled by generative AI and machine learning — the power demands of next-generation data centers will only increase. New technologies, such as those supporting 224 Gbps-PAM4, will help to improve efficiency, but the pace of new and expanded data centers stresses the need for high quality, reliable power solutions to optimize data center energy management — most importantly, during times of power failure and surge.  

The performance of a data center can be measured in its uptime. After all, downtime is costly and disruptive, leading to lost revenue, decreased productivity and damaged brand reputations for both the data center owner and the customers relying on the data center for their day-to-day operations. A recent study performed by the Uptime Institute found that over 60% of unplanned downtime results in losses of at least $100,000 for the businesses impacted. And losses in the $1 million range are on the rise. This is only a financial loss. If a critical application goes down, such as a connected medical device relying on data center operation to ensure timely notification of a fluctuation in patient vital signs, lives may be on the line.  
 
For the manufacturers and suppliers of uninterruptable power supplies (UPS) — critical parts of an effective data center battery backup system — there’s little room for failure. So now the question is, how can design engineers tackle these higher loads and ensure uptime for critical applications, at the times UPSs are needed most? 

A UPS is defined by its topology, with three basic types — offline (or passive) standby, line interactive and online double conversion — each sharing the same fundamental characteristic. In all cases, a UPS utilizes a battery and an inverter to power the connected systems and devices during failure. What differs are the means by which the battery is charged, how the battery takes over and to what degree the UPS maintains or improves power quality. 

Online double conversion UPSs are the preferred choice for data centers as they provide an interruption-free transition to backup power along with continuous protection from surges, spikes and power degradation. Unlike offline standby UPSs which take several milliseconds to kick on during an outage, or line interactive UPSs which only provide a seamless experience for low load applications, an online UPS operates continuously. The mains is first fed into a rectifier, cleaning the signal as it’s converted into direct current (DC) and charging the battery. It is then converted back to alternating current (AC) by an inverter before being delivered to connected systems. During day-to-day operations, an online UPS ensures power quality is maintained to protect the equipment and power delivery. And in a failure, the battery takes over instantaneously and operates until drained.

Three-phase online double conversion UPSs where energy is delivered at 415V rather than single-phase at 230V are even more preferred for large and mission-critical data centers. Although they introduce further complexity in the design of the UPS, they’re more reliable, efficient and scalable than single-phase UPSs.  

For emerging applications like autonomous driving and long-distance robotic surgery, data center uptime is nonnegotiable. This means ensuring UPSs perform to expectation when in the most critical moments. Here are design considerations that can improve UPS performance:

• Size – Although critical to protecting data centers from financial loss, UPSs do not contribute to revenue generation. If UPSs take up excess footprint on the data center floor, facilities cannot use that space for equipment such as servers or storage. The use of miniaturized, yet high power components can ensure design engineers allocate space within the UPS where it’s needed most without sacrificing airflow — reducing the overall size of the system and freeing up floorspace for revenue generation. 
• Heat – Heat poses a unique risk to the longevity of UPS systems as it can shorten the lifespan of a battery. Proper airflow is just one means of reducing heat — another is ensuring the components within the UPS are designed for high-power applications. For instance, laminated busbars can improve heat dissipation and connectors designed for low voltage drop minimize heat generated at the contact interface. A cooler UPS is a more reliable UPS. 
• Field Serviceability - Although UPS systems are designed to properly operate in the heat, dust and vibration typically found in data centers, that doesn’t mean they run perfectly smoothly all of the time.  A loosened connection point will come up and can quickly lead to catastrophic results from ingress to disconnect. Connectors designed for blind-mating ensure a connector is properly seated in hard-to-reach locations, while terminal position assurance (TPA) keeps the connection in place. 
• Power quality – There are many factors that can influence the power quality within the UPS, but it often starts with the connectors and cables. Connector design techniques, such as reducing contact resistance, can increase current-carrying capacity without producing more heat. Additionally, reducing even slight misalignment can improve quality. For instance, Molex’s PowerPlane Busbar connectors are designed to allow for 40% more points of contacts than comparable solutions, improving performance and reliability. Further choices in alloys and plating, such as copper and silver, can also improve quality. And it goes without saying it’s critical to ensure that an interconnect solution is designed for the power required.  

Data centers are not only the backbone of our new digital economy. They also influence the reputation of companies that rely on them. When operating under normal circumstances, a data center is out of the public eye. But when a data center goes offline, the service impacts can be catastrophic. While data centers ensure continued operation of connected systems and devices, UPSs ensure continued operation of the data centers.  

Molex has 80+ years of expertise providing interconnect products for applications where reliability isn’t optional. Our multi-disciplinary engineering expertise is backed by our leading portfolio of high-power solutions including busbars, connectors and cable assemblies. We’re regularly enabling UPS manufacturers to build solutions for not just the data centers of today, but also the next generation of hyperscalers. Click here to learn more about our data center power management solutions for UPS and beyond.