DC – An idea whose time has come and gone?

To anyone who has a passing knowledge of the History of Electrical Engineering (yes there really is such a thing), the 1888 War of Currents figures as one of the most prominent events. It was Thomas Edison and DC vs. George Westinghouse and Nikolai Tesla championing AC. As you know, DC, or direct current — something you use with anything electronic, especially battery operated — presents a constant voltage level on the line. On the other hand, AC–or alternating current — something you use whenever you plug a cord into a wall outlet — presents a sine wave that alternates between plus and minus 169 volts 60 times a second, or +/-325 volts 50 times a second, depending on where you are in the world. [BTW, if you don't recognize those voltages, it's because, by convention we talk about the Root Mean Square (RMS) voltage which is the peak voltage divided by the square root of two, or 120 VAC and 230 VAC.] At the time it was a no-holds barred, knock down – drag out standards fight that would make WWE wrestling promoters blush over the promotional ethics used. Thomas Edison went around to state and local fairs zapping everything from dogs to elephants with AC to “prove” that AC was unsafe. He went so far as to invent the Electric Chair as a way of saying AC was so unsafe that it could be used to reliably execute criminals.

In the end, AC won out because of two inventions — the transformer and the synchronous motor. The transformer solved the transmission distance problem of the day of getting power more than a mile away from the generating station. It did it by using the first secret of energy efficiency – using as high a voltage as you can. Being before the invention of electronics, DC didn’t have a technical answer. The second invention, the synchronous motor, allowed you to use the alternating current directly at the load — a motor to convert the electrical power to mechanical power, and now at the point of use without belts and pulleys. With the clever invention of the Electric Utility and government-sanctioned Monopoly model, AC and the power grid went on to become the Top Engineering Achievement of the 20th Century, and never looked back.

122 years of tradition, unchanged by progress

And that’s pretty much the way we have left it ever since, or as I like to say: 122 years of tradition unchanged by progress. But consider this from the Center for Power Electronic Systems (CPES): by the end of 2010, power electronic systems are expected to control 80% of all electricity used.

If you look under the covers, you actually find that it’s a lot worse than that. While AC is used between systems and components because it’s been the standard for over 100 years, internally, lots of things convert to or from DC. In fact, anything with a logic chip in it or consumer electronics is using DC. And when you look at alternate energy production, Photovoltaic, wind turbines, water pico-turbines, they are almost always DC. Now if you are asking “so what?” I’ll tell you what: energy efficiency. Maybe it’s time to really question our assumptions; what if everything we “know” is wrong?

The two secrets of energy efficiency are using the highest voltage you can and doing the least conversions. In fact, we at Intel Labs followed a natural progression into our new Energy Systems Research Lab because we started out trying to optimize energy efficiency for Intel. We started with the processor. But today, the processor in a laptop running typical office applications constitutes only 5% of the total power used by the system. Thus, optimizing the last 1% of efficiency here is well beyond the point of diminishing returns. So we have expanded our view over the last five years to the system and then to the enterprise.

DC – An idea whose time has come and gone … and come again

In 2006 we participated with Lawrence Berkley Lab in a study of power for the datacenter. The published, peer- reviewed findings are that you save an astounding 28% over the current North American AC power distribution practices by using DC. This kind of energy efficiency improvement simply does not exist at the component level any more. In 2008 we teamed up with EYP Mission Critical, an HP Company, and Emerson Network Power to examine how DC power distribution for a datacenter stacks up against high-efficiency AC (called 400Vdc before the new standard proposed by EPRI) powerpoint slides and white paper. Looking at a 5.5MW datacenter expansion as an engineering study, we found that even high-efficiency AC is still at least 7% less efficient than 380VDC. Even the Green Grid, which has members who are heavily invested in the current AC distribution, agreed the efficiency savings are significant. In fact two years ago, the Intel Eco-Tech Office held the first IT “Great Debates” and the first of three was AC vs DC in the Datacenter.

Thus, the main benefits from 380VDC distribution for the datacenter are:

- 7% Energy Savings vs. High-Efficiency 415VAC; 28% vs Current Typical 208VAC

- 15% Less Capital Cost

- 15% fewer PSU components

- 33% Datacenter Space Savings

- 200% Reliability Improvement, which goes to 1000% if you directly connect the battery bus

- Elimination of harmonics and inherently immune to other AC power quality issues

- Natural affinity to alternate energy generation (Photovoltaic, and wind are ~400Vdc internally, and you actually lose energy & efficiency when you are forced to convert to AC)

So with all these advantages, who wouldn’t jump all over this? Turns out lots of folks are wading into the DC waters. Japan is out in front with their Green IT initiative that started in 2009. In December, they proposed a standard for what is now being called 380Vdc. The Electric Power Research Institute (EPRI) has DC Power Partners group that is working through a lot of the details of a 380Vdc Standard. They have just elected to fold under the EMerge Alliance which has been championing low-voltage DC for the office. The DC Power Partners has proposed a voltage/time envelope to match the well known CBEMA spec for AC-powered IT equipment at 110VAC. Connectors have been defined for power supplies and are available in production quantities, as well as connectors for rack interconnect using the existing IEC connector and standards. A +/- 190 VDC distribution wiring scheme is being proposed to address human safety and is actually safer than Vac distribution (lower voltage and lower human hazard). All of these technologies were reviewed at the latest TC38 working group meeting in Tokyo, Japan. The neatest thing is that other industries are starting to align around the 380Vdc standard – Photovoltaic panels and Hybrid Vehicles and Florescent lighting are already using these kinds of voltages and are snapping to this standard.

In New Mexico, Intel Labs just opened a new lab, the New Mexico Energy Systems Research Center (NM ESRC) for collaborative research, with the participants in the New Mexico Green Grid. One of the first projects in the NM ESRC is to connect up a DC microgrid between the existing 10kW Photovoltaic Array, DC-powered servers (working with Intel IT) and then expand the microgrid to include an interface to energy storage (DC), high-efficiency office lighting and power (also DC … think laptop plugs in without a power cube), and then grid-tie it to the AC power on campus.

To borrow the WWI doughboy’s battle cry: Thomas Edison, we are here!

22 Responses to DC – An idea whose time has come and gone?

  1. Tom Friend says:

    I can remember touring the Celilo Converter Station in 1978. At that time it was an all valve (vacuum tube) design. It’s not far from the Dalles, and certainly worth looking up as a long term project to promote the lower losses of DC power transmission.

  2. David Geary says:

    Guy,
    Very well said! Thank you!
    Although industry seems to have focused in on the energy efficiency gains, with a 380v DC power system. I think it is the many other advantages that will ultimately lead to large scale adoption. Greater reliability, less cost, less space – all coupled with better system control and protection, which I am confident will result, all point to a compelling story. A DC power infrastructure allows us to take a big step back in order to look at the BIG PICTURE. We should look at electrical engineering design in a whole new way. Simply applying AC power system design principles to DC power system design will severely limit benefits. We have the opportunity to change the world in our grasp which makes this a very exciting time.

  3. Chris H says:

    Other significant historical advantages of AC over DC were taught when I was young. AC mechanical switchgear needed a smaller spark gap than DC as it quenches at the zero-crossing point. AC transformers also allowed for safe isolation in user equipment. DC had an electrolysis effect between the conductors in humid conditions.
    At a given voltage AC is less likely to be lethal than DC if someone grasps the conductors – as it does not cause the muscles to lock in position. Lower than 50vdc is regarded as “safe” – although 32vdc has been known to kill in humid conditions. A high voltage shock is survivable if the shock current is high enough to stop the heart beating. A lower current can cause fibrillation of the heart that is lethal unless it is quickly stopped by a medic applying a shock from a defibrillator.

  4. galLee says:

    The 380Vdc goes into a box that distributes 24Vdc to all the cubicals (usually ceiling mounted). Some laptops can plug directly into that, or there are smart-connectors that do the DC-DC required to charge the laptop. We are working with the EMerge Alliance on putting together a demonstration site.

  5. galLee says:

    Ah, the old chestnut, safety. This is where Edison started trying to show AC was dangerous in 1888, and it’s no surprise to see it’s still in the playbook. But there really is no operational difference. Turns out at these voltage levels both AC and DC are lethal; standards and practices exist are are in daily use to provide for appropriate protection in both cases. Don’t forget that the AC voltage number is expressed as an RMS voltage and the peak voltage is actually 41% higher (square root of two). Furthermore, the data actually shows that DC is less likely than AC to cause heart fibrilation. (Turns out that with +/- 190Vdc distribution, IEC personal protection is actually safer than high efficiency AC.)
    With respect to the other issues (electrolysis, zero-crossing and spark gap, etc.) these are red herrings, as their solution has already been engineered by the companies providing equipment rated for DC applications. The relative ease or difficulty of achieving that is about as interesting as the mideval arguments over how many angels can dance on the head of a pin … and completely irrelevant to building 380Vdc systems, as those have already been taken care of in the design and certification of those components. In fact DC is used in telephone, ship and train applications every day, worldwide.

  6. Liang Downey says:

    Very good history reviews Guy! When people bring up the AC vs. DC debate, they tend to think that the world stopped evolving since 120 years ago. Westinghouse or Tesla, even Edison himself, failed to predict the coming of Internet and PC revolution. We are now living in a digital world – everything a device – in a ubiquitous or pervasive computing environment; we call it being SMART, Smart Grid, Smart Building, Smart City, Smart Planet…. Back then, the “Loads” were limited to lighting.
    The “Loads” will be 100% DC in a not-so-distant future, counting the Electric Vehicle!
    On the supply side, the 120 year old centralized one way power generation scheme will be evolved into more and more distributed energy networks, some innovators call it “Enernet”. Taking the approach of how “bits” are moved around in the cyberspace, “watts” will be distributed in a similar ways. The language of that “watts” is DC inherently.
    This is an exciting age, driving a 4th industrial revolution!

  7. I strongly suggest that the DC voltage standard should be +/- 340 VDC and not 380 VDC. This allows use of PWM inverters to run three phase 230 and 460 volt motors. We, here at the UB, demonstrated superiority of DC over AC for rural and agricultural loads for an irrigation system at a farm in Batavia, New york in December 2001. We had also started, with partial DOE funding, a “Central Office of the Future” project for Verizon’s Center in Rome, NY with Fuel Cells as a power source to demonstrate superiority of DC in 2002. Unfortunately, the project was abandoned, after all project enigneering had been completed, by the Verizon management when the “Mover & Shaker” behind it retired and myopic new management took over.

  8. Guy AlLee says:

    Thanks so much Ed for the link to the studio360 website stories on Tesla and Edison; they are a well presented and entertaining diversion in the history of electrical engineering.

  9. Guy AlLee says:

    @ Dr. Safiuddin on +/-340Vdc — We very deliberately picked 380Vdc because you want to get to as high a voltage as you can afford for efficiency. At the same time this standard is targeting Low Voltage applications only (

  10. Dear Guy:
    Low voltage applications, according to IEEE/ANSI Standard 141-1993 [Table 3-1], cover ratings up to 600 VAC. That is, peak voltage of 848.5 volts. Voltage ratings of electrical equipment are dictated by insulation ratings with respect to ground. My proposal of + 340 VDC falls within this low voltage classification. All PWM drives for 480 VAC motors have a front end AC-DC rectifier interface operating around 680 VDC. These do fall in to the low voltage class. Also, to the best of my recollection, the old telephone circuits operated at 160-170 VDC and not 190 VDC. A 340 VDC power system feeder can also be set up as + 170 VDC, which would allow operation of 120 and 240 VAC motors through PWM inverters without a need for AC-DC rectifier interface for each drive.

  11. Guy AlLee says:

    @ Dr. Safiuddin on +/-340VDC — We absolutely agree on the 120 and 240VAC motors. No reason a PWM inverter can’t be designed to work on 380VDC now that it is the standard, and, thus, eliminate the conversion at the motor. (Or more likely just be built into the VFD to accept 380VDC directly, with lower cost, higher reliability and fewer components.) As the 680VDC (+/-340VDC) proposal is higher than 600 it is arguable whether it would still be considered “Low Voltage” depending on which international standard gets applied. And more than that, since 680VDC(+/-340VDC) is higher than 400VDC, it fails to meet the cost requirement of staying with volume priced components. Finally, there is a third concern with safety, in that the IEC 60479 RCD operate time requirements cannot be met (cost effectively). I can certainly see a different standard developed to cover this — one that takes the requirements into account and bears the higher associated costs that the bulk of the applications in PV, datacenters, lighting, EVs, and VFDs don’t need to bear.

  12. Dusty Becker says:

    Actually in the old telephony world the #4 ESS switches were set up to operate on 140VDC. In more recent times we have used +/-190V distribution on buried tip and ring circuits to extend DC power to ONU’s for Fiber to the Curb applications, although in those instances the power was limited to SELV levels in a manner that avoided hazards to craft standing in muddy ditches. The point being, DC above 48Vdc has been around in the distribution world for some time.

  13. yemi oyekan says:

    tesla vs edison
    volta vs galvani
    the former medival argument replicates in the field of medicine ,where the argument is if humans produce electricity or the static electricity is from the environment,the frog leg experiment argument relied on static electricity ,dc or ac wasn’t invented by then, but the frog muscle still twitched,a simple principle reproduced in the electric chair,the question should be would AC or DC be used safely,ethically or medicinal,i.e
    delgado reports intracranial electrodes used in the treatment of psychiatric conditions here not muscles but the emotions are the question,would feedback monitoring from wireless devices coupled to processors,provide the answer to the question,AC or DC which is safer,ECT and the electric chair any difference

  14. Guy AlLee says:

    @yemi oyekan — the medical investigation you advocate seems a little off topic, especially since the voltage levels we are talking about here are considerably higher than anything you would use in human studies, independent of if they are AC or DC. BTW, the data already exists for AC and DC with respect to protection devices to mitigate the hazard, i.e., RCDs for personal protection (IEC 1000/05 and 1002/05). From it you can conclude that 380VDC using a +/- 190VDC distribution is safer even than the prevailing 240VAC and the ability of personal protection devices to operate in sufficient time. (Remember, the peak voltage of a nominal AC voltage is 41% higher; 240VAC is really +/- 339V.) And when you look at 415VAC (+/- 587V), there is no acceptable RCD solution. We are satisfied that 380VDC can be safe and better than AC for this as well as all the other reasons mentioned above (efficiency, simplicity, reliability, footprint, lack of harmonics, fewer components and using less of the earth’s resources). I look forward to the broad adoption of 380VDC and the benefits that will bring to datacenters’ carbon footprint as well.

  15. Lyle Gentry says:

    This is probably not the right forum, but I’m looking for information on moving a private residence “off grid” and in the process running an entire household on a DC system. I’m not an engineer but it sounds like alot of the posters here are. My knowledge of electrical systems is extremely limited so please forgive me in advance for my ignorance.
    I’m operating under the basic premise that it would be safer/more efficient to run a household off a DC system when using a photovoltaic/battery bank power source thereby eleminating the need to use an inverter to convert from DC to AC. Is this correct? Is it practical to run everything (appliances, electronics, overhead lighting, HVAC, etc…) on a DC system? Are there wiring or parts issues? Is DC-powered equipment readily available? Or would I be better off just inverting the DC and going with AC? I’ve read something about a 48-volt DC system being used. Would this be adequate? It sounds alot less potent than the 380 volt systems discussed here.
    Any info/assistance would be greatly appreciated.

  16. Hello Guy,
    Applied Power Systems is a US mfr of power conversion products (AC-AC, AC-DC, DC-DC, DC-AC). I have been following some of the DC projects around the country, and the Intel Lab project seems to be one of the more thought out, practical projects. I would like to know how APS may become a part of this project or any others that you may be involved with.

  17. Guy AlLee says:

    @Peter Dowling – the easiest way for a manufacturer to get involved is to join the EMerge Alliance (www.emergealliance.org). They are defining and promoting the new 380VDC standard and taking it to ETSI where it is already a draft standard.

  18. Jeff Barton says:

    Has there been any research into distributing AC and DC on the same wires in a building? Mixing a DC souurce (ie solar panels) with AC bulk power and decoupling them as needed at the destination? Does this seem practical?

  19. galLee says:

    It comes down to technology and cost. Obviously the technology to multiplex AC and DC on the same wires exists. (This is essential what Power over Ethernet (PoE) is doing, although in that case the DC is power and the AC is signal.)
    Cost, on the other hand is where I think this approach will struggle. It creates interesting issues with existing components that assume pure AC or DC, which would have to be resolved. And it adds cost at the outlet, either in the form of a smart socket or a smart cord that can separate the AC and DC to give you the one you want. Finally it could put a burden on all existing AC appliances in your house and mine that have always assumed 0 volts DC — would they be OK as is or have we just introduced new failure modes?
    Finally, this is something that Intel Labs and the 70+ EMerge Alliance companies are looking at. In fact, NDN and the New Policy Institute are advocating just that — supplement AC with DC distribution in buildings — in their Electricity 2.0 whitepaper. They are proposing adding DC to the current wiring practices, although it would be separate outlets. I think there’s work ahead to first standardize the voltage and socket interface which is the key for a market to develop. But it’s not hopeless, either, as we see our research in 380VDC has already resulted in an ETSI standard for data centers, that is already having an impact far beyond in commercial buildings, alternative energy, energy storage and electric vehicle charging.