AC vs DC power in data center

There has been a debate going on for some years if the traditional AC or new DC power distribution is best approach to power a data center. The DC power side has been pushing their technology with claims of quite considerable power savings. In many published articles, expected improvements of 10% to 30% in efficiency have been claimed for DC over AC. I have had my doubts of the numbers on their promises. Now there is some new data on AC vs DC issue available.

White paper compares AC vs. DC power distribution for data centers article tells that a new white paper from APC by Schneider Electric provides a quantitative comparison of high efficiency AC vs. DC power distribution platforms for data centers. The latest high efficiency AC and DC power distribution architectures are shown by the analysis to have virtually the same efficiency, suggesting that a move to a DC-based architecture is unwarranted on the basis of efficiency

A Quantitative Comparison of High Efficiency AC vs. DC Power Distribution for Data Centers paper demonstrates that the best AC power distribution systems today already achieve essentially the same efficiency as hypothetical future DC systems. It also tells that most of the quoted efficiency gains in the popular press are misleading, inaccurate, or false (like I have suspected to be for some time). And unlike virtually all other articles and papers on this subject, this paper includes citations and references for all of the quantitative data (which is very good).

The paper first describes that there are five methods of power distribution that can be realistically used in data centers: two basic types of alternating current (AC) power distribution and three basic types of direct current (DC) power distribution. These five types are explained and analyzed.

One AC and one DC, offer superior electrical efficiency. The paper focuses on comparing only those two highest efficiency distribution methods, which are very likely to become the preferred method for distributing power in future data centers. The data in this paper demonstrates that the best AC power distribution systems today already achieve essentially the same efficiency as hypothetical future DC systems.

The best AC system is based on the existing predominant 400/230 V AC distribution system currently used in virtually all data centers outside of North America and Japan. Increasing Data Center Efficiency by Using Improved High Density Power Distribution white paper gives details how it could be used in USA. It says that the use of the international 230/400 V distribution system instead of the USA standard 120/208 system can save 56% in the lifetime cost of the distribution system, and save floor space and weight loading.

The preferred DC system is based on a conceptual 380 V DC distribution system (consensus in the literature as a preferred standard) supplying IT equipment that has been modified to accept DC power. In the proposed international ETSI standard for DC distribution for data centers, the 380V DC system is actually created with the midpoint at ground potential to keep the maximum system voltage to ground to within +/- 190 V.

Based on the data I think the 400/230 V AC distribution system is the best way to go in data centers around the world.


  1. Tomi Engdahl says:

    Bureau of
    Indian Standards
    Redefining Electricity
    First International Conference on Low Voltage Direct Current
    New Delhi, India,
    & 27 October

  2. Tomi Engdahl says:

    ETSI EN 301 605 (Grounding and Bonding)
    - Summary

    ETSI EN 301 605 (Grounding and Bonding)

    Both system
    arrangement comply with relevant safety requirements

    IF the continuity of operation is placed in the forefront THEN the symmetrical IT
    with earthed high
    point is the first choice
    In cases where an
    IT system is used for reasons of continuity of supply, automatic disconnection is not usually required on the occurrence
    of a first fault (single fault) to an exposed
    part or to earth. This is valid on condition that an
    Monitoring Device (IMD)
    indicates the first fault by an audible and/or visual signal which shall continue as long as the
    fault persists.

    IF similar system
    arrangement as for today’s
    system is requested
    THEN the TN
    S system +400
    may be chosen

  3. Tomi Engdahl says:

    In this article, we look at low-voltage dc at 380 Vdc,
    the new industry specification and the single worldwide
    standard in data centers.

    The Advantages of 380 Vdc
    subsequent r
    &D was able to fully articulate the
    benefits and advantages of 380 Vdc for the data
    center. t
    hey show that 380 Vdc:

    is 28% more efficient than 208 Vac systems at that

    is 7% more efficient than 415 Vac

    results in 15% less up-front capital cost in production

    requires 33% less floor space

    has a 36% lower lifetime cost

    is 1,000% more reliable

    uses 20–100 times less copper than –48 Vdc

    introduces no harmonics and is safe

    in fact, lower capital and operating cost is the reason that
    380 Vdc is the ideal voltage today.
    the third secret of mak
    ing a cost-effective, efficient voltage distribution standard is
    to stay below 420 V so as to use parts that share the volume
    economics with desktop personal computer P
    s. t
    hat is
    why the industry selected 380 Vdc (and a specification that
    requires operation up to 400 Vdc and must survive exposure
    of up to 410 Vdc).

    table 2. Analysis of IEC protection against electric shock safety requirements
    shows that distributing ±190 Vdc with a midpoint ground can provide 380 Vdc that has a
    comparable safety margin to ac at 208–250 Vac.

    380 Vdc is a hazardous energy, no more or less than ac at
    these voltages. Whether it is more or less hazardous than ac
    should not be the issue; it is enough to note that both ac and
    dc are hazardous at these voltages.
    national and interna-
    tional standards exist for the proper safety procedures for
    both ac and dc, and they need to be followed.
    at the same time, the data do exist, and they tend to
    show dc is actually safer than ac at the same voltages

    iec 23
    e W
    g 2
    analyzed; according to these data, 380 Vdc distributed
    at ±190 Vdc is actually as safe as ac voltages from 208
    Vac up to 250 Vac

    Debunking Common Myths

    DC is only 1% or 2% more efficient:
    False. t
    foregoing data clearly support that 380 Vdc is 7–8%
    more efficient than low-voltage ac will ever be.

    DC requires big conductors and can only go a cou
    ple of meters for a reasonable cost:
    While that may
    be true of dc at 12 or 24 or 48 V, it is certainly not true
    at 380 V

    Running off the battery bus eats up all your effi
    False. While it is true that the 380-Vdc P
    is 1–2% less efficient when it has to tolerate the wide
    260–400-Vdc input, this was judged more than worth
    while to get a 1,000% reliability gain

    Arc flash is an unacceptable hazard with 380 Vdc:
    False. t
    he connectors have been specifically designed
    with features to fully enclose the arc for 5–10-
    currents. c
    onnectors with magnetic arc breakers
    and switched interlocks exist in the market today
    for greater currents. a
    nd the iec
    -309 connector is
    already fully rated for up to 450 Vdc.

    AC is safer because the voltage crosses zero 50–60
    times a second:
    this is only true if the current is not
    leading or lagging. a
    nd if that were true for servers,
    they wouldn’t have PF
    c circuits.

    An ac server PSU is more efficient than a dc PSU:
    this can never be true for a given design. s
    imply clip
    the bridge and PF
    c circuits from said ac supply, and
    it is now a 1–2% more efficient 380-Vdc supply.

    What to Expect Next
    a worldwide industry consensus is building around
    380 Vdc, and it is being led by the 95-member
    alliance. t
    he group’s vision of l
    Vdc in the four areas
    of commercial buildings—occupied space, data center,
    building services, and outdoors—has led to initial indus
    specifications for 24 Vdc and 380 Vdc (distributed as

  4. Tomi Engdahl says:


    The ETSI standard states that the voltage drop in the entire 48 V DC loop has to be limited
    to 1V [12]. This leads to oversized cables, especially on long distances in order to meet

    At its origin, the telecom industry relied on 48 V DC technology, for safety reasons in
    wireline communication networks. As the years went by, this voltage was kept because of the large
    installed base, which led to lower equipment cost.

    AC and DC topology comparison

    The choice that was made to mitigate the losses in this ve
    ry high power density
    environment was to increase the voltage, and reduce the amount of conversion steps (only two in
    the 380 V DC topology) in order to mitigate the conversion and conduction losses associated with
    the AC setup.

    The voltage was
    determined to be 380 V DC because it is a voltage that was already present in
    the power supply at the time, thus facilitating the deployment of the technology. 380 V DC is
    also considered a low voltage from a regulatory standpoint.

    380 V DC topology g
    ot rid of all phase balancing issues, as well as harmonics present in the
    legacy AC system. Also, DC offers much better integration of renewable resources
    , such as PV
    , wind power
    and fuel cells
    , that are natively DC.

    of the improvements resulting from the use of 380 V DC
    in the data center industry can
    be applied
    to a cell site facility

    Cabling modeling
    Cabling was selected following the principles of the ETSI EN 300 132-2
    for the 48 V DC
    system (less than 1 V of voltage drop in the entire loop).
    The regulations for a 380 V system were unclear up to two years ago. The Japanese Electric
    code states that the voltage drop rate has to be 1.5% or less for a branch circuit, or 2.5% or less
    including the main line. These specifications are defined for facilities carrying sensitive ICT


    The voltage originates from the DC source at 169 V DC, before being boosted to a voltage
    of 380 V DC. This voltage is subsequently
    to a voltage of 48 V DC to feed the loads at the
    top of the tower and in the shelter.

    As we would have expected, the 380 V DC system is more energy efficient than the 48 V DC system. The hybrid topology is the least efficient of the three systems

    The ETSI EN 300 132-
    2 makes a 48 V DC loop energy efficient at the expense of great
    amounts of copper.
    The 380 V DC system is energy efficient and costs less

  5. Tomi Engdahl says:

    Telecommunications cable

    Telecommunications power cable products, as described in Telcordia GR-347, consist of a stranded copper conductor used in AC/DC circuits up to 600 V that are insulated with non-halogen, limited smoke, polyolefin materials that are heat-resistant, moisture-resistant, and flame-retardant. These cables are provided as either Class B (standard) or Class I (flexible) products.

    Telecommunications power cable is intended for use in AC/DC distribution circuits, wireways, racks, and conduits installed in telecommunications Central Offices (COs), transmission stations, cell-tower sites, and other remote sites. These environments are normally dry, but cables may be placed in partially covered or protected porches, crawl spaces, or in underground vaults where water and high moisture levels can occur.

  6. Tomi Engdahl says:

    Rack Power (PDU) terms and technology

    This video is a primer on the terms and technology behind the rack Power Distribution Units (PDU) that are used in data centers.

    What Functionality Do I Need in a Power Distribution Unit?

    Power Distribution Units can offer a variety of features such as local displays, remote monitoring capabilities and even outlet control. There are six main PDU functionalities available in the market today, and it’s important to understand the benefits of each. Today we will be discussing PDU functionalities, focusing on their main features, and deciding what features are important to your application.

  7. Tomi Engdahl says:

    Key grounding and voltage
    considerations in the data center

    Designing a data center’s power system consists of numerous decisions about the
    components in the power path. In most of the world, there are two primary voltage
    schemes (three-phase) available, which are based on either the North American
    480/208/120 V (600/208/120 in Canada) or the 400/230 V system in used Europe and
    some parts of Asia.

    we are generically referring to the 400/230 V system (this is the midpoint voltage that represents 380/220 V through 415/240 V).

    some data centers are exploring the use of direct current (DC) to improve overall efficiency

    Rack-level power density and distribution

    Here in the North America, the common use of 120 V worked fine when a rack used 1-2 kW
    per rack and a single 20 A circuit was all that was needed (two for A-B redundancy). With
    the advent of blade servers, which typically require 208 V or 230 V circuits and use five or
    more kilowatts as well as racks full of 1U servers, the new baseline is now 5 kW per rack.

    almost all IT power supplies are now autosensing and universal voltage-capable (100-250 V)
    more efficient at 208 V or 230 V than at 120 V (or even lower at 100 V in Japan)

    by making three-phase power available in the rack, you will increase the available power bya factor of 300%, yet increase your cable conductor count and its cost by only 66%

    by deploying three-phase 208/120 V power to the racks, you can supply either 208 V single
    -phase or 208 V three-phase power and also provide 120 V for older or specialized IT gear

    consider using three-phase connectors such as NEMA “Twist-Lock”

    In the 400/230 V system, all output circuits are 230 V single-phase (from any phase to neutral and ground).

    In North America, we commonly use 208/120 V to end-user equipment using standard plugs and receptacles.
    At 480 V, the danger of Arc Flash is substantially greater

    Will European voltages work in U.S. data centers?
    In Europe, only single-phase 230 V is distributed to plug-in devices via standard IEC C13-
    and C19-type receptacles and plugs, at up to 16 A. However, three-phase 400 V power is also commonly available via the larger IEC type 309 receptacles at up to 60 A. Also
    in Europe, 400 V work in the panel is commonly done (with appropriate safety gear), since
    that is the basis of all their power distribution systems.

    It is clear that European data centers will continue to use the 400/230 V system since it is already native to their overall existing power system

    In North America, several vendors now offer 400/230 V products as a higher-efficiency alternative to traditional 208/120 V distribution systems.
    the North American high-utility voltage would be transformed only once to down 400 V (instead of 480 V).
    hope of gaining a 2-5% potential increase in energy efficiency

  8. Tomi Engdahl says:

    INTELEC 2012
    400V DC Microgrid Small Scale Demo System for Telecom and Datacom Applications

    EN 300 132-3-1 normal service voltage range of 260V – 400V DC

    Safety concerns for 400V DC distribution have been systematically addressed with a high resistance mid-point ground (HRMG)
    a. On the bulk power source by midpoint resistive grounding with fault detection
    b. On the distribution system by using ±190V instead of 0V-380V
    c. On the equipment front end conversion, with fully isolated, safety extra low voltage (SELV)-output high voltage bus
    Midpoint resistive grounding with fault monitor provides for a safe and reliable distribution line; fully isolated, SELV output high voltage bus converters used as equipment front-end provide for an extra layer of safety for the operators.

    Existing 48V DC loads can be transitioned to 400V DC distribution by implementing simple, minimally invasive adapters.

  9. Tomi Engdahl says:

    Why Use AC Instead of DC at Home??

    Isn’t AC more dangerous than DC?? So why do we use AC instead of DC to power our homes? Did we go wrong somewhere?

  10. Tomi Engdahl says:

    Rack Power (PDU) terms and technology

    This video is a primer on the terms and technology behind the rack Power Distribution Units (PDU) that are used in data centers.

  11. Tomi Engdahl says:

    The Top 3 Reasons for Transformer Failures

    The culprits are all sins of omission: a lack of craftsmanship, high-quality materials, and good design.

    When it comes to big-ticket items, power transformers are near the top of the list. So, when they fail prematurely, it’s all the more painful: Damages can far exceed the cost of a replacement. The added expenses may include the loss of production time, damaged credibility, and regulatory fines and civil lawsuits.

    “After a transformer failure, the first thing out of the customer’s mouth is inevitably, ‘Hey, it’s just a year old! What happened?’”

    But it’s no mystery, according to Jones. Ultimately, you get what you pay for.

    Plant engineers, facilities managers, general contractors, and specifying electrical engineers can learn a lot from the “post-mortem” experiences of a CSI tech such as Jones. In most cases, the premature failures of transformers could have been avoided, and the culprit is often an inadequately designed or constructed unit.

    “You want equipment that is cost-effective, not cheap,” says Jones. “Cheaper transformers often cost you more in the long run, especially if [they are] critical to your business process or data center. Then the extra $10,000 or $20,000 for a better unit represents inexpensive insurance.”

    “Probably the most influential factor in transformer longevity is the level of craftsmanship, the attention to detail in the manufacturing process, and the quality control,” says Jones. “This is often overlooked in today’s rush to automate every manufacturing process.”

    “Look where they put transformers these days: in tight spaces when the buildings are first being erected or on the roof where you’ll need a crane to replace it,” s

    At that point, this manufacturer put a single wrap of Kevlar, whereas a high-quality transformer would have double or triple layering of insulating material because it’s a high-stress area.”

    The manufacturer tried to avoid the rap by explaining that their transformers needed a snubber, a capacitor-filter-resistor network that connects in parallel with the primary winding to absorb high-voltage transients.

    “But a snubber is a $25,000 add-on,” explains Jones. “If you have a cheaply manufactured transformer and need to buy this extra equipment, where are the savings in that? The university had 10 of these transformers, so they were facing a quarter-million-dollar jump in costs. Now who is going to pay for that? It becomes a big fight.”

    Jones also discussed the importance of the iron core material. Pure original—as opposed to recycled—magnetic silicon steel is best. Also, the thinner the core steel pieces, the better.

    Commonly used M6 steel has a thickness of 0.014 in. per piece, whereas the M3 steel is only 0.009 in. thick. To cover the same volume or area, you have more pieces with M3. The more pieces, the lower the no-load losses and the higher the efficiency.

    “High heat contributes to transformer failures,”

    Whether wet- or dry-type, the way coils are wound around the transformer’s core greatly affects its durability. Because of increased axial forces acting at the corners of rectangular-wound transformers, energy gets wasted and noise is created. With round-wound designs, however, voltage stresses are lower, so they stay cooler, run quieter, and present less risk of short-circuit with the sheet wound secondary.

    Beyond the improved reliability factor, the round-wound designs further increase efficiencies and save costs in real time as the plant consumes less electricity.

    “In the past, they made transformers that were somewhat overdesigned in terms of capacity and durability,”

  12. Tomi Engdahl says:

    AC vs DC Switching Demonstration with Arcs

    The difference between switching AC and DC.
    With DC, the arc formed as the switch contacts open is far more difficult to extinguish, and can cause significant damage.

    Explains why relays have different voltage rating for AC and DC.

  13. Tomi Engdahl says:

    Rack Power (PDU) terms and technology

    This video is a primer on the terms and technology behind the rack Power Distribution Units (PDU) that are used in data centers.

    Selecting a rack PDU

  14. Tomi Engdahl says:

    AC/DC Electrical Panel/Wiring Set Up For a Tiny House or Camp Cabin

  15. Tomi Engdahl says:

    The importance of circuit protection in electrical distribution system design

    The electrical engineer is responsible for designing power distribution systems for buildings. Understanding the full circuit-protection requirements will enable the engineer to design the safest and most reliable electrical distribution systems for buildings.

    Nearly every article in the NEC includes some form of circuit protection, stressing the importance of the issue. The basic goals of circuit protection are to 1) localize and isolate the condition or fault and 2) prevent and minimize any unnecessary power loss. There are several types of abnormal conditions that may occur throughout a building’s life, in which an electrical system must be designed to correct or overcome. These include overloads, short circuits, under/overvoltages, transient surges, and other power issues, such as single-phasing of 3-phase systems and reverse power-phase rotation.

    An overload is caused by an excessive demand from utilization equipment that is higher than its rated capacity. System overloads can be tolerated for a short period of time before corrective action must be taken. Short-circuit faults, on the other hand, are caused by failed electrical components. Since the damage can be immediate, the faulted part of the system must be isolated as quickly as possible. Several types of faults exist including arcing line-to-line faults, line-to-ground faults, and 3-phase bolted faults. Many faults start out as intermittent, arcing faults with variable impedance and relatively low magnitude currents, characterized by the uncontrolled release of energy.

    The service entrance equipment offers the first step in protecting against thermal overloads and faults where circuit-protection devices are introduced into the system. Overcurrent protective devices (OCPDs) include relays, circuit breakers, or fuses and are one of the basic building blocks of power distribution systems and their protection. At the most basic level, these devices are inserted into the distribution system to “break,” isolate, or disconnect the circuit if there is an overload or short-circuit condition. These devices have been used since the late 19th century and continue to be applied today.

    The most basic levels of circuit protection include fuses and thermal magnetic-type circuit breakers.

    With a standard fuse, or thermal magnetic device, you have basic circuit protection, but due to limited flexibility, they only offer basic protection from significant arc flash dangers. A thoughtful design assures the downstream feeder breaker has enough time to “clear” before the fault condition pushes the upstream breaker into its trip curve. This is referred to as selective coordination.

    A means to produce a more reliable and coordinated system is to add intelligence to a circuit breaker in the form of integral trip units and protective relays. Another type of circuit breaker is an electronic trip-adjustable circuit breaker. This breaker has a long time-operating region, a long-time delay, a short time pickup, a short time delay, and finally, an instantaneous pickup. These parameters are adjustable over a given range.

    A technology that provides further reduction in the let-through energy for a fault in the region between two electronic-trip circuit breakers can be accomplished through ZSI (zone-selective interlocking). ZSI consists of wiring two circuit breaker trip units together so a fault is cleared by the breaker closest to the fault in the minimum amount of time possible. They operate such that if the downstream circuit breaker senses a fault, it sends a restraining signal to the upstream circuit breaker. The upstream circuit breaker will then continue to time out as specified on its characteristic curve, tripping only if the downstream device does not clear the fault. The primary goal is to switch off the fault current within the shortest time possible while impacting the least amount of connected equipment.

  16. Tomi Engdahl says:

    The future of the utilities sector

    Utilities companies are facing some big challenges as customers seek to become more self-reliant through the installation of on-site generation solutions.

  17. Tomi Engdahl says:

    Selectively coordinated overcurrent protection for power systems

    When designing a safe and reliable power distribution system, it is imperative to consider life safety and equipment protection

    overview of key terms from 2017 Edition of Figure 1: An electromechanical overcurrent relay installed in a switchgear at a federal facility. All graphics courtesy: CDM Smith NEC Article 100 includes the following:

    Overcurrent: Any current in excess of the rated current of equipment or the ampacity of the conductor. This may result from overload, short circuit, or ground fault.
    Overload: The operation of equipment in excess of the normal, full-load rating or in excess of rated ampacity (conductors) when it persists for a sufficient length of time to cause overheating. A fault, such as a short circuit or ground fault, is not considered overload.
    Short-circuit current: An overcurrent resulting from a fault of negligible impedance between live conductors having a difference of potential under normal operating conditions.
    Ground fault: An unintentional electrically conducting connection between an ungrounded conductor of an electrical circuit and the normally non-current-carrying conductors, metallic enclosures, metallic raceways, metallic equipment, or earth.
    Selective coordination: Refers to the localization of an overcurrent condition to restrict outages to the circuit or equipment affected. This is accomplished by the selection and installation of overcurrent protective devices (OCPDs) with their ratings or settings for the full range of available overcurrents, from overload to the maximum available fault current, and for the full range of OCPD opening times associated with those overcurrents.

  18. Tomi Engdahl says:

    Teardown of an APC Switched Rack PDU – AP7921

    In this video we take a look inside my APC AP7921 Switched Rack PDU. I’ve taken a look inside more basic PDUs before which are essentially rackmount extension leads, I was interested in seeing how this would compare.

    The only problem is the processor is too slow to support https and ssh using the latest firmware. Older firmware uses poor encryption.

    What’s really bad with PDU manufacturers is that they subcontract out the software and design super pretty graphical interfaces to appeal to what I imagine is their principal audience, but ignore how painful this makes them to use from automated scripts.

    The industrial PDUs that you can still control through simple telnet or serial port commands are more and more rare.

    The display board looks to be the same as on the full rack 3 phase pdu.

  19. Tomi Engdahl says:

    WWT – APC Rack PDU

    An 6 min video discussing APC’s broad range of Rack PDU solutions and those installed in the ATC.

    Rack PDU’s are becoming an important part of the overall data center design since the introduction of Converged IT and other high density applications. This video will explore considerations when choosing Rack PDU’s and covers topics of management types, KW capacity, mounting type and descriptions of outlet types. Having intelligent PDU’s can decrease the chance of downtime, which is still mainly caused by human error since often times open plugs on the PDU are used without consideration of what the overall PDU capacity is. These intelligent PDU’s can send out alerts and alarms showing current consumption as well as the benefit of outlet control to either remotely reboot devices or for security/capacity management reasons, turn off unused outlets until approved assets are installed.

  20. Tomi Engdahl says:

    Technical Pro PS9U vs. Cyberpower CPS-1220RMS PDU (And One is a Fire Hazard) [C39FD0152]

    I had the occasion to buy a Technical Pro PS9U from Amazon, precisely because it had a bunch of really bad reviews. Like fire-and-meltdown bad. This intrigued me, because I wanted a product just like this and, well, how bad could it be?

    Really bad, as it turns out.

    Then for the sake of comparison I take a look at the Cyberpower CPS-1220RMS PDU, which is the exact opposite type of PDU in terms of fire hazards.

  21. Tomi Engdahl says:

    Home Lab: DIY Network PDU – Raspberry Pi

    A re purposed relay board to Switch 240v sockets.

    Based off the code from this Instructable Project:

  22. Tomi Engdahl says:

    How to Choose a PDU: Basic PDUs

    Choosing a PDU to suit your needs.

    Need a Rack PDU tailored to your data center?

    Both HDOT and Alternating Phase are available through Server Technology’s “Build Your Own PDU” configurator. Build Your Own PDU takes a Switched, Smart or Metered 42-outlet High Density Outlet Technology (HDOT) PDU and allows you to build an HDOT PDU your way in Four Simple Steps. Choose a configuration. Download a spec sheet and request a quote in four simple steps.

  23. Tomi Engdahl says:

    Introducing the HDOT Cx Outlet

    Data center power requirements are hard to predict, and data center managers need to prepare for increasing power density as well as rapid advancements in server, storage, and network technology. The task of predicting what mix of C13 and C19 outlets required in a rack PDU is now moot.

    Server Technology’s new “two outlets in one” innovation in HDOT Cx

    The Best Data Center PDU Just Got Better

    Server Technology’s multiple award winning, DCS Awards 2018 Data Centre PDU Product of the Year , High Density Outlet Technology (HDOT), has been improved with our Cx outlet. The HDOT Cx PDU welcomes change as data center equipment is replaced. The Cx outlet is a UL tested hybrid of the C13 and C19 outlets. HDOT Cx accommodates both C14 and C20 plugs.

  24. Tomi Engdahl says:

    erver Technology creates hybrid outlet for flexible PDUs

    Legrand subsidiary Server Technology has introduced a new power outlet in its power distribution units (PDUs) which can act as a C13 or C19 socket, giving users more flexibility.

    The HDOT Cx range includes the new Cx outlet, so the strip can accept any combination of hardware connectors, and data center managers don’t have to keep re-ordering and replacing PDUs when they change or redistribute the equipment in their racks, Server Technology’s director of engineering Calvin Nicholson told DCD.

    Traditional server racks often have the C13 outlets for network switches near the top, and the heavy-duty C19 outlets for servers or storage near the bottom of the rack, Nicholson said, but these configurations can change: older blade server shelves used a C19 outlet, while ‘pizza box’ servers use a C13, and heavy duty Cisco switches usually need a C19. “In a colocation environment, the environment is changing all the time,” he said. “In the enterprise space, you have server and networking and storage racks, which typically, all use different PDUs, so the data center manager has to stock multiple units.”

    The Cx outlet is not a standard, of course, which may arouse some doubts among users, but Nicholson assured DCD that the product has safety approval from Underwriters Labs (UL) and Europe’s CE certification. The company has no plans to submit its Cx to the IEC for approval as a standard: “We could do that, and it would allow us to sell the outlet to other vendors,” he explained, “but we want to sell PDUs.”

    Despite the Cx’s wider groove, C14 and C20 cables designed for C13 and C19 outlets respectively will not wobble when plugged into the new outlet, Nicholson said. C20 locking cables will “lock right in,” and sleeves are available for C14 locking cables.

    Some observers have predicted the death of the PDU, with solutions based on DC power distribution heavily promoted by groups such as Open Compute. Nicholson says demand for PDUs is still high – including from the webscale operators who might be expected to move to DC distribution: “People are doing DC distribution as a small percentage experiment.”

  25. Tomi Engdahl says:

    Server Technology Introduces the Cx Outlet

  26. Tomi Engdahl says:

    Power Supplies and Circuit Breakers Keep Faults in Check

    Sponsored by Digi-Key and Phoenix Contact: Industrial power supplies that incorporate features such as SFB circuit breakers provide a better level of protection and overall reliability.

  27. Tomi Engdahl says:

    A Birds-Eye View of Power Equipment in a Data Center

    How all the different equipment in a data center work together to provide emergency backup power supply

    UPS topologies: standby, line interactive, online UPS systems

    Explains the different types of UPS backup power systems, including line interactive, standby (offline) and online UPS topologies. Adjusts power voltage using a variac to demonstrate how a UPS supports a critical IT load. Live demonstration shows how on-line and line-interactive UPS technologies operate under varying power conditions. Check out how the line-interactive UPS technology will experience voltage drops, but the on-line double-conversion UPS technology can handle the same power variations without voltage drops.

  28. Tomi Engdahl says:

    Designing power systems for co-location data centers
    Engineers must consider many factors when designing power and electrical systems for data centers.

    Providing continuous, high-quality electrical power is among the most important obligations of the co-location data center owner. Electrical engineers must consider many factors when designing power/electrical systems for these co-location facilities. Issues such as backup, standby, and emergency power systems must be incorporated. Co-location facilities differ slightly in that they may have metered/submetered power systems or power based on occupancy measurements.

    Co-location data centers are facilities where data center rack space is rented out to companies that require secure and reliable data center space. They provide secure computing space for companies so that they can spend time on their core business rather than operating data centers. Co-location building owners’ business model requires that they push the envelope of design to remove unnecessary components, increase efficiency, and prioritize design decisions to meet the terms of their contracts. The design engineer for co-location owners must understand these drivers and design a system that supports the co-location business model.

  29. Tomi Engdahl says:

    Seven tips for transformer design in industrial buildings

    Proper knowledge of transformer design for both new and replacement installations is essential to building operation, specifically in industrial applications.

  30. Tomi Engdahl says:

    Home> Community > Blogs > EDN Moments
    Three-phase current field trial ends War of Currents, May 16, 1891–May-16–1891

  31. Tomi Engdahl says:

    Data center power in 2019

    The power challenge

    Preventing disruption to the systems in the data center building is critical; downtime means dollars lost and unhappy customers. The operator can rely on uninterruptible power supply systems and power distribution units that safely and reliably control the flow of electricity to sensitive equipment. Small to mid-sized businesses and residential buildings with back-up power generation may also be candidates for load management programs. Surpassing 10kW per rack is the norm, which will make in-rack power protection less viable. The usage of end-of-row UPS systems is coming.

    Intelligent power management (IPM)

    IPM is a combination of hardware and software that optimizes the distribution and use of electrical power in computer systems and data centers.

  32. Tomi Engdahl says:

    A short tutorial on power system designs in cloud computing

    According to, the global data center market is forecasting growth of 6.4% CAGR from $19.1 billion in 2020 to $26.1 billion by 2025. With the increasing growth in cloud computing demands comes the increasing demand on processing power. A recent study titled “Recalibrating global data center energy-use estimates” assessed that the worldwide power consumed by data centers in 2018 was 205 terawatt-hours, or 205,000,000,000,000 W-hr. Such significant demands in power consumption lend to the prioritization of efficiency and reliability.

  33. Tomi Engdahl says:

    GaN’s Power Density Carries Unlimited Design Potential
    Dec. 2, 2021
    GaN FETs maximize power density, enabling a whole new world of higher efficiency with smaller, more reliable, higher-efficiency power-management solutions—even in the RF space.

    What you’ll learn:

    How GaN helps boost efficiency in data centers.
    What is the LLC resonant converter?
    Using the AC-DC-AC converter for UPSs with no transformer.

    Gallium-nitride (GaN) FETs have inherently superior performance versus traditional silicon FETs. This advantage in performance enables design engineers to push the envelope in power designs and reach new levels of power density and efficiency. Applications range from ac-dc power supplies to multi-kilowatt, three-phase converters.
    GaN Power Density in the Data Center

    GaN technology enables energy-efficient power supplies for data-center servers. GaN power transistors reduce the weight, size, and cost of data-center power designs while also reducing energy consumption. GaN’s high-speed switching enables market trends like ultra-thin power supplies, motor-drive integrated robotics, and the ability to achieve greater than 200% more power density in next-generation data centers.

    High-performance server and telecom applications need high-efficiency and high-power-density isolated dc-dc converters. As such, LLC resonant converters look like an optimum choice for this use case. This converter architecture has a zero-voltage-switching (ZVS) capability from zero to full load, plus a low turn-off current for primary-side switches. In the case of switching frequencies lower than resonant frequency, synchronous-rectifier (SR) devices are turned off with zero-current switching (ZCS).

  34. Tomi Engdahl says:

    DC Power Circuit Wiring Color Code Standards

    DC power installations, for example, solar power and computer data centers, use color coding which follows the AC standards. The IEC color standard for DC power cable color code is listed in the table below.
    Function Label Color
    Protective earth PE green-yellow
    2-Wire Unearthed DC Power System Positive L+ brown
    Negative L- grey
    2-Wire Earthed DC Power System Positive (of a negative earthed) circuit L+ brown
    Negative (of a negative earthed) circuit M blue
    Positive (of a positive earthed) circuit M blue
    Negative (of a positive earthed) circuit L- grey
    3-Wire Earthed DC Power System Positive L+ brown
    Mid-wire M blue
    Negative L- grey

  35. Tomi Engdahl says:

    SuperGrid Institute Responds to Energy and Climate Demands
    Power conversion solutions for the electricity transmission network of the future.

    Researchers and developers at SuperGrid Institute use Ansys electronics software solutions to perform studies on power converters, critical links in the chain between electric generators and consumers, for their clients.

    As an independent research and innovation institute based in France, SuperGrid Institute is dedicated to developing technologies and services for the supergrid — the electricity transmission network of the future — that uses direct current (DC) and alternating current (AC) at very high voltages. Power electronic systems are key drivers for the integration of large amounts of renewable energy injected into the transmission system, and therefore a main focus at SuperGrid Institute.

    One common use of DC power is electric vehicle recharging stations. Innovative power converters and research efforts are needed to develop cost effective solutions for integrating these stations into the grid. SuperGrid Institute proposes a medium-voltage direct current (MVDC) grid structure in response to these challenges and offers grid studies and related technological solutions developed with industrial partners.

  36. Tomi Engdahl says:

    How SVT cables edge SJT cords for high-density IT equipment in data centers
    Jan. 12, 2022
    The UNC Group releases its definitive guide to using SVT cables in data centers.

    The UNC Group, a specialist in IT and networking cable solutions, has released a special report on SVT cabling titled:“SVT Power Cords in the Data Center: The Case for Using Them,” describing the benefits of using SVT power cords in data centers instead of the more commonly used SJT power cords.

    The new white paper outlines the challenges of using power cords in the data center, and demonstrates why the more flexible SVT cable is the best choice for saving space and reducing costs.

    “In this report, we demonstrate the superior qualities of the SVT cable to address power cord requirements in data centers,” said Amanda Garcia, Director of Business Development at UNC Group. “Because it is more pliable and easier to work with, our Fortune 100 clients have been using SVT cables for over a decade, and we want all our clients to recognize how these power cords can improve their data centers.”

    SVT Power Cords in the Data Center
    A Case for Using Them

  37. Tomi Engdahl says:

    A few months ago I made a comment to a YouTube video and I predicted that in the future, homes would be wired with DC. I got all sorts of negative criticisms and threats to have me locked up in the nuthouse. Well, it’s very gratifying to be vindicated and those Debby Downers can kiss my rosy red rear!

    I considered that the major incentive is that silicon is cheaper than steel, and about the same high efficiency.

    House Runs 100% on DC Power — Purdue University Project

    Did you know there’s a silent war going on inside your home? Alternating current (AC) electricity comes in from the grid, but many of your appliances and lighting run on direct current (DC). Every time you plug in a TV, computer or cell phone charger, power must be individually converted from AC to DC — a costly and inefficient process. Purdue University researchers have proposed a solution to the problem by retrofitting an entire house to run on its own efficient DC-powered nano-grid.

    “Large-scale distribution of DC power through a house in the 21st century is really uncharted territory,” said Jonathan Ore, a 2020 Purdue Ph.D. graduate who served as the lead researcher on the project. “You can’t just go to the hardware store and buy DC circuit breakers or other critical distribution systems. We had to create this infrastructure from scratch.”

    Purdue researchers, in collaboration with Rectify Solar, developed a patented distribution system that enables the house to integrate both DC power — from solar panels, wind turbines or battery storage — and AC power from local electrical utilities. The system is also modular

    “The creation of the 380-volt DC load center was definitely a challenging and rewarding experience,” said Phil Teague, co-founder and CEO of Rectify LLC. “We used biomimicry and the neural connections of the brain as our inspiration, and added smart technologies and control mechanisms. Transitioning to DC can simplify homes, buildings and the grid as a whole. This project helped me realize that DC is not only the future, it always was.”

    Why DC power?
    AC has been the dominant infrastructure in the world’s electrical grids since the late 1800s, when the “war of the currents” saw Thomas Edison’s dream of a DC-based electrical infrastructure lose out to George Westinghouse’s AC system. But while the “war” may seem to be over, two recent developments have prompted researchers to re-investigate DC’s benefits. The first is the increasing availability of renewable energy sources — solar panels and wind turbines — as well as energy storage in large home-based battery packs. These devices are all naturally DC, so to have a DC-based home infrastructure enables this energy to be delivered with almost no waste or inefficiency.

    “A DC-house can potentially sustain itself for short periods of time by generating its own renewable energy and detaching from the grid through the help of on-site stored energy.”

    DC Nanogrid House

  38. Tomi Engdahl says:

    Data Centers Feel the Power Density Pinch
    Aug. 5, 2021
    The data-center industry has struggled to achieve the best balance between space and power, but sizable steps toward that goal have been taken over the last few years. This article discusses ways to optimize power density for data-center operations.|7211D2691390C9R&oly_enc_id=7211D2691390C9R

    What you’ll learn:

    Ways to achieve efficient power-density solutions.
    How hardware acceleration comes into play in data-center operations.
    The physical aspect of dealing with floor space.
    How GaN options can significantly improve efficiency.

    Data centers keep the internet running 24/7. Approximately 44 zettabytes of data in our data centers come from 8 to 30 billion connected devices worldwide. Forecast experts expect that to increase by 10% over the next year. Thus, power-supply design economics changes are in process.

    Modern central processing units (CPUs) in the data center are power-hungry, making it difficult to juggle the balance between power and space. In turn, power density in data centers has risen dramatically. In the first decade of the new millennium, data-center power density was in the region of 4 to 5 kW/rack. Today, we’ve reached 8 to 10 kW/rack heading toward 20 kW/rack (Fig. 1). This also means more heat and added cooling costs.

    The amount of power consumed by each data-center rack continues to climb.9,10 The 48-V dc rack, instituted in 2016 by Google, was a big step toward improving power density in the data center. And dc-dc converters made significant data-center efficiency and power-density improvements in terms of hardware accelerators as well as high-speed processors. GaN power switches may be the most significant savior for the data center versus the incumbent silicon MOSFET solutions. Designers are beginning to see the value of GaN for dc-dc converters that can deliver efficient power to data centers. Focusing on low-voltage power is essential in information processing.

    The data center’s future looks brighter with the design efforts mentioned in this article. More new and innovative power design architectures will be developed as power engineers strive for even better power-density numbers moving forward.

  39. Tomi Engdahl says:

    Based on a standard quarter-brick format, the Advanced Energy Artesyn NDQ900 non-isolated DC/DC converter provides 48-V power conversion in compute and telecom applications….

    DC/DC converter enables transition to 48-V infrastructure

    Based on a standard quarter-brick format, the Advanced Energy Artesyn NDQ900 non-isolated DC/DC converter provides 48-V power conversion in compute and telecom applications. The device delivers up to 900 W of power with efficiencies as high as 97%, along with a single regulated 12.25-V output. In addition, a PMBus interface allows flexible digital control and monitoring.

    The NDQ900 non-isolated bus converter has a wide input range of 40 V to 60 V.

  40. Tomi Engdahl says:

    How a fully-stackable eFuse can help meet ever-increasing power needs of servers

    As demand for data increases, so does demand for servers and data centers, and thus higher demand for power. Industry trends suggest that power per rack, which was 4 kW in 2020, will be as high as 20 kW in 2025.

    Given limited physical real estate available for data centers and servers, the delivery of more power in less area is known as a high power density requirement in server power architectures. Increasing the efficiency of server power supplies can also keep cooling costs down.

    Servers are usually scalable and are hot-swappable in order to meet different processing requirements and maintain high system availability. To achieve seamless hot-swap functionality, server motherboards and power distribution boards employ hot-swap controllers or eFuses. Components such as eFuses in server power supplies need to provide higher current to meet increased server power requirements. Protection devices such as hot swaps and eFuses also need to handle high peak current to match the higher peak-processing capabilities of modern microprocessors in servers.

    Traditionally, high-power server designs include hot-swap controllers with multiple metal-oxide semiconductor field-effect transistors (MOSFETs). But server power and power-density requirements are increasing exponentially. To satisfy these needs and simplify these designs, consider the TPS25985 (80 A peak) and TPS25990 (60 A peak with the PMBus interface) eFuses in server power architectures. The TPS25985 and TPS25990 can support 60 ADC and 50 ADC, respectively, and have an adjustable current limit of up to 60 A and 50 A, also respectively. It is possible to stack multiple unlimited TPS25985 and TPS25990 eFuses to achieve higher current.

    Achieving high power density

    Power density is a must-have requirement for modern server power supply units (PSUs). The latest generation of server PSUs are in the range of a 3-kW (250 A at 12 V) power rating. When selecting an eFuse, it is important to have the highest current in the smallest size. The TPS25985 packs 80 A of peak current in a 4.5-mm-by-5-mm package. Figure 3 shows some of the TI’s eFuses.

    By integrating a MOSFET, a current monitor, a comparator, active current sharing and a temperature monitor, the TPS25985 and TPS25990 eFuses significantly reduce the total printed circuit board or printed wiring board area.

    Current-share and current-monitor accuracy

    Hot-swap controllers cannot control the gates of multiple paralleled MOSFETs very precisely; therefore, current sharing by paralleled MOSFETs is not accurate. Precision amplifiers can help achieve high current-share accuracy and current-monitor accuracy, but adding them increases the total solution size. It is challenging to gauge the die temperature of the MOSFET, and therefore impossible to guarantee its thermal protection in transient and steady-state conditions.

    The TPS25985 and TPS25990 eFuses have integrated active current sharing and direct access to MOSFET die parameters (voltage, current, temperature), which allows accurate control of all eFuse gates connected in parallel and accurate die temperature monitoring of integrated FETs. Compared to an eFuse without active current sharing, the TPS25985 and TPS25990 enable design engineers to optimize the number of eFuses and the performance of the system.

    Remote monitoring and control

    The TPS25990 adds PMBus interface capability to the system. The TPS25990 enables single-command power cycling with an adjustable turnon delay, which allows system design engineers to sequence and reset the system remotely. The TPS25990 also offers black-box capability, where seven events are recorded with relative timestamps. The TPS25990 incorporates high-speed analog-to-digital converters that enable users to plot one signal of their choice, mimicking a digital oscilloscope. The GUI for the TPS25990, along with its other features, helps design engineers not only reduce their total development time but also quickly identify and resolve field issues, which are generally very difficult to reproduce and troubleshoot.

    Thermal considerations

    Server power systems operate at a wide ambient temperate range (–40ºC to 85ºC). Hot-swap controllers or eFuses experience even higher ambient temperatures. Therefore, power design engineers become concerned about the thermal performance of these devices when high currents are packed in small packages. The TPS25985 and TPS25990 eFuses alleviate this concern, with the ability to operate at a 125ºC junction temperature. The TPS25985 and TPS25990 offer an RDS(on) of 0.59 mΩ and 0.79 mΩ, respectively; RDS(on) spread across process, voltage and temperature variations is limited. Thus, these eFuses experience very low self-heating and a wide operating temperature range without sacrificing the derating. Figure 6 shows the case temperature of the TPS25985.


    Design engineers can reduce development time and design cost by using the TPS25985 and TPS25990 eFuses in server power architectures. The eFuses’ low RDS(on) reduces power losses in the system, helping data centers achieve their efficiency goals.


Leave a Comment

Your email address will not be published. Required fields are marked *