Careful observation of the global voltage levels yields an overall range of 100-240VAC at either 50 or 60Hz, which leads one to believe if we have a single power supply capable of supporting the full voltage/frequency ranges, then it is universally-compatible with any AC source, but this is not the case. See the example image of a power supply safety label, which one can find on any certified, shipping, international solution in figure 2.
It may seem natural to want to support as wide of a range as possible, but as with anything in life (especially power) there are tradeoffs we must make to optimize a solution for a given application or use space. There are also tolerances that must be built into all designs to account for non-ideal situations. In terms of voltages, this can be related to protections for overvoltage scenarios (typically related to personal safety in addition to protecting equipment), undervoltage scenarios (maximizing uptime, also for equipment protection), and balancing of phase currents in multiphase solutions. For line frequency, this can be related to power quality and grid stability. How these protections/mechanisms are achieved and why are beyond the scope of this discussion, but are covered extensively in the
RECOM “AC/DC Book of Knowledge: Practical tips for the User” [3] document freely available.
When we take a common tolerance figure of ±10%, it quickly becomes obvious how we can define a universal range of 90-264VAC, 47-63Hz that is seen on many power supply safety labels. This was merely a quick example to demonstrate how we get from so many international standards to consolidated ranges for more universal support, but is very general and does not go into the motivations for any given mains specifications. There are other support ranges specific to military and/or
industrial applications, such as a 400Hz standard for aircraft/shipboard power. Multiple, single-voltage sources can also be separated by phase angle to maximize power delivery, while minimizing currents, which can be seen in three-phase AC systems.
Ultimately, most end systems and loads will run off of DC power (AC motors being the glaring exception), which is why there are even more standards for DC voltage supplies than AC, though not typically for facilities/building-scale distributions. High voltage is defined as >1,000/1,500V (AC/DC, respectively) though pretty much anything >=60VDC is considered a higher voltage for safety purposes (human contact), also known as Safety Extra Low Voltage (SELV). While no one standard exists (actually numerous worldwide) for what is commonly known as the high-voltage data center (HVDC, not to be confused with direct current), there are many different standards to call out a distribution architecture in the 300-400VDC range.
The logic is if server/networking hardware and supporting infrastructure is all designed to support a universal AC input with a power-factor-correcting (PFC [4])
AC/DC power supply, then all the same equipment can handle the DC voltage derived from the rectified, AC-input waveform, thus justifying the mitigation of a conversion stage (and all that is gained by its removal). 24VDC distributions can be common for industrial settings with small relays/breakers/motors and smaller systems optimized for a standard, mechanical form factor, such as the DIN rail [5] standard. Other well-known DC distributions include the universal serial bus (USB, 5-20VDC) and
Power over Ethernet (PoE, 44-57VDC), also combining power with data conductors in hybrid cables.
The choice of a main distribution voltage for a facility is driven by many factors related to decisions driving capital and operational expenditures (CAPEX/OPEX, respectively), not merely what equipment is supposed to plug into it. Safety is nearly always a key factor determining distribution architectures and must be considered based on worst-case expectations for operator exposure, conductor-to-conductor spacing, and constraints of the operating environment. Consolidation of voltage distribution bus architectures offers many advantages in the streamlining of equipment purchases (CAPEX) and efficient utilization of equipment/machines (OPEX). The less conversion stages from upstream source (i.e. – utility grid, energy storage, etc.) to end load (i.e. – system, ASIC, motor, etc.), the more to be gained in terms of streamlining equipment purchases and taking advantage of economies of scale. Commonality can also help mitigate net load dynamics, which enable optimization of energy efficiency by narrowing unpredictability and therefore enhanced opportunity for intelligent power management (IPM [6]) techniques.
A common mains or distribution carries far more benefits than can be comprehensively reviewed here, but should be recognized is some other categories. The ability to have a more predictable maintenance schedule and fewer part numbers to manage can offer significant savings in short- and long-term expenses. Reduced number of parts to replace/manage carries many obvious advantages that ripple from saving user cycles at the point of consumption to reducing overhead and shipping costs for replacements.
As we transition to
Smart Buildings and factories of the future, getting the best of both configurability and agile changes with common form factors is critical for success. From a Quality perspective, systems (particularly components and motors) will last longer when they can function in more-constrained, predictable operating/environmental conditions and maintenance cycles. These first-order benefits lead to an immense list of second-order benefits, depending on far deep one wants to analyze. For instance, a common distribution may mitigate costly back-up power and/or
energy storage solutions that would otherwise need to act as energy buffers for intermediary voltages. If the entire power delivery solution gets just a couple percent more efficient in commutating input-to-output power, then there is justification from CAPEX savings that ripple all the way from the point of load up to the power plant.
