Build An Efficient Inverter, Win A Million Dollars

Little Box Challenge

Google and the IEEE are giving away a million dollar prize to an individual or team, that can build the most efficient and compact DC to AC inverter. The goal is to design and build a 2kW inverter with a power density greater than 50W per cubic inch. To put that in perspective, conventional solar string inverters have power densities around 0.5-3W per cubic Inch, and microinverters around 5W per cubic Inch. So in other words, an order of magnitude more efficient than what we have now.

For the challenge, the inverter needs to convert 450VDC, with a 10 ohm series resistor simulating a solar array, to 240VAC @ 60Hz. Testing will consist of powering various resistive, inductive and capacitive loads ranging from 0-2kVA. The inverter is expected to regulate voltage within 5%, and frequency within 0.05%, while keeping the enclosure below 60 degrees C, and conforming to FCC Part 15 B (Unintentional radiators).

If you and/or your team can figure out the most efficient topology, switching frequency, novel use of high power wide bandgap (WBG) semiconductors, physically reduce the size of the input and output filters, and keep the whole thing running cool. Then get registered before the September 30, 2014 deadline. Inverters need to be functional and the results of this test procedure (PDF warning) sent in before July 22, 2015, then 18 finalists will be chosen to bring their inverters in person to a testing facility in the United States by October 21, 2015. The grand prize winner will be announced sometime in January, 2016

[Thanks for the tip Dmytro]

This Arduino power inverter would need a serious upgrade to enter. And speaking of entering challenges, it’s still not too late to enter our very own Hackaday Prize!

86 thoughts on “Build An Efficient Inverter, Win A Million Dollars

      1. Ah but at 95% efficiency the box will have to get rid of 100W which further adds to the required size just due to heatsinking. although it looks like forced air cooling is allowed. If you wanted to go fanless, looking at 98.5% efficient minimum (taking into account the available surface area and convection as well as radiation, almost equal effects in this case)

        1. Not much on worldly views eh? I’ve lived in the states for almost 30 years and still find English and it’s weird grammar rules puzzling – add in autocorrect and you never know – is the writer a dumbass, or just not all that good at English? I’m not sure there’s a direct correlation between good writing and good STEM skills.

          1. I am sorry you had to go daid303. I hope you had a good time while it lasted. I can only hope one day people like Bill will think before making these broad statements. Good luck in the next world! U will be missed.

    1. Grammar is something most engineers have no idea about. Nor do they really care to! Their grammar is Math which doesn’t lend a care to proper language rules. Can I get a shout out!?!

  1. It would be a lot easier if we switched to DC mains voltage. But that is another story…
    I find that they don’t really justify the size reduction: current one is still small compared to the whole house+solar panel thing. Now if cheaper is what they want, then yes that is always better.

    1. This. What we really need to promote on-site power generation is a DC voltage standard. It can exist in parellel with the AC standards while the latter is slowly phased out. Eventually, all AC from the grid will be converted to DC at the point of use. Google’s prize is for the opposite of what we need!

      1. depends on who “we” are. I think if you are in the middle of no where developing country the odd of getting lets say a brand new refrigerator than runs on a new DC standard is close to zero, getting an old standard AC refrigerator that was discarded because it wasn’t the right color is probably more likely

      2. Almost all large scale power generation is currently AC, by huge synchronous generators. If you use a large DC generator, efficiency drops off (brushes). Plus you need DC/DC converters (you don’t want 5000 V DC household wiring). Edison tried it (without DC/DC converters)… it didn’t pan out.

        1. It’s moronic to propose a different standard without doing any reasonable investigation into why we are using the current standard in the first place. Someone did a shitload of research long before you were born, in some cases there were several such studies. Stop starting from square one and leverage existing research to keep moving forward, not proposing some bullplop that’s likely going to cause more problems than solve.

        2. Yeah, AC was preferred at that time because it is easy to step up/down with a simple transformer. But with today’s electronics things are different, we may be able to accept more complex DC/DC conversion for a better total efficiency.
          There are already high power/voltage power lines that carry DC, but indeed probably nobody wants to do the change at end of the line.

          Still, come to think about it, most things I have at home(by number) could operate from DC without any change because they actually use SMPSs.

      3. Try checking out the wall warts for a number of the electronic devices/appliances you own. Chances are very good that you’ll find that, among them, they run on quite a variety of different DC voltages. So, with that in mind, are you still going to suggest that it would be easier to get both the entire electronics device/appliance industry and the plethora of local building codes to adopt a single DC voltage standard at the same time all consumers throw all their old devices/appliances out versus just developing a single widget that converts DC to AC? While you’re getting right on that, Google & Co. will be working to get someone to develop a converter we can use for the numerous decades we’ll all be waiting for you to get the job done…

        1. Do you really think it would be more efficient to convert locally-generated DC to AC, then convert it back to DC using wall warts, most of which aren’t switched? Far better to have a DC standard and start designing new appliances to run directly from DC, which many of them already use internally. Old appliances could still run on AC grid power, and new appliances could run on rectified AC as a backup for locally generated DC.

          This is similar to the move toward electric vehicles. Your argument is analogous to saying we shouldn’t move toward electric vehicles because of the problems getting the auto industry and local building codes to adopt a standard for charging stations. The challenges are real, but they’re not a reason to hold back progress.

          1. I never said we shouldn’t work towards that. I just said that, in the mean-time, the rest of us will still probably benefit from the existence of a device like the one Google & Co. are working towards. The world needs both because no amount of wishing will make the wholesale changing of embedded standards a fast process. What the world doesn’t need is to follow a simpleton thought process, like yours, that says we should ONLY work towards a “perfect” solution and totally ignore anything else.

      1. It exists… Several 400kV HVDC lines are in use. They typically use stacks of SCR’s as the switching elements. A power engineer once told me it’s currently about the same price as a transformer for 400kV, but going down pretty fast.

    2. Go chinese then, what output would you like it labeled as. It may only last 5 minutes though due to no errorc checking or saftey shutdowns. Got to love the surge of induction motors.

    3. Ehhh, we kinda figured this out a hundred years ago. Didn’t we? Correct me if I’m wrong but AC is totally efficient for traveling long distances without costly amplification right? A carrier, if you will. Also DC will suck the life out of you if you complete the circuit to ground. Converting AC to DC locally is so much more efficient! I’ve had a few beers but I’m still questioning this comment, I can’t believe you posted this

  2. I find it necessary to point out the obvious here. A functioning design for a 10x-100x more space efficient power inverter would be worth an awful lot more than a million bucks. If you’re going to go through the trouble to design and build such a revolutionary new system, may as well go the extra mile and start your own company to build and sell it. Sure you can get a load of cash quick if you give it to Google, but in the long term it seems like a foolish decision to me.

      1. Actually, the impression I was given was that they don’t even require you to publish or reveal to them how your converter works. You just have to provide them with one, or more, so that they can test it for 100 hours of “real world” use to confirm that it meets the contest criteria.

  3. Why the hell do we need 240Vac? honestly what is it with this fetish of making a DC power system convert power to work with legacy out of date AC devices?

    If you install a solar system you install DC appliances. because losing 5% of my power means that on the long winter cloudy days I lose 5% of all that battery storage which is precious precious storage capacity.

    Plus DC appliances are more efficient so you have lower losses and consumption. Converting DC to AC to run legacy crap is like gold plating and polishing a turd. It is a dumb thing to do.

    1. I hate to break it to you, but AC motors are more efficient and cheaper than DC motors. Ask Edison. Or Tesla. Same goes for the whole system: AC is more efficient and cheaper. That is precisely why it has won.

      1. I thought AC won because you can use transformers, and thereby send power along long lines at high voltage and low current. Since current causes heating loss.

        The real advantage of the system is this high voltage. Which was only possible in the olden days through using transformers, which back then only used AC. Now we have methods of DC-DC conversion and silicon only gets cheaper. And as people point out, modern really-high-voltage lines use DC, as do inter-country links.

        Using DC means not worrying about synchronising every power station up to the grid frequency, though that’s been cracked long ago anyway.

        I think what might really help, is a low-voltage DC bus for the home. Like USB sockets all running from one line. Except maybe with a higher voltage and power rating than USB. Since so many things in the home run on DC now, and plenty more things could. Perhaps a central efficient convertor in each home would save a few watts over a country-wide scale.

        1. Now That! Is a great! Idea!!! Building central home DC converters in new home construction is a brilliant idea! But leave AC transmission alone, it the best, safest most efficient way to get electrons down the line

    2. Good luck finding all the devices/appliances you want that all run at exactly the same DC voltage. Last I checked, it seemed that almost every device/appliance I owned ran at a different DC voltage level once you got beyond the wall wart. There is virtually no standard in place. I suggest you set aside a lot of money and space in your home for all the different DC/DC converters you’re going to need to buy and maintain…

    3. Im on DC in the van and I live in New Zealand (perhaps you have heard of me) we are 240v, being DC means installing BIG FAT EXPENSIVE COPPER WIRES, you seem to have missed the point of high voltage power transmission lines.

  4. So effectively they want a 2kw solar inverter that’s smaller than a 4 inch cube? And keeps within the specficed tolerances?


    Good luck to anyone who’s giving it a shot!

    1. Exotic design probably requires exotic magnetics and fewer semiconductors, e.g. using magamps in the regulation as opposed to pwm and resistive current sensing. I suspect it is doable but requires some well funded researchers; well out of the reach of hobbyists.

      1. I am sure that it is possible to achieve the requirements. The volume requirement will be the real challenge as the efficiency requirement is not very stringent.

        If someone wants to have a crack at it, great! Here’s what I think will be most challenging:
        – Cooling
        – Mechanical integration
        – Size reduction of input filter (DC capacitor bank)
        – Size reduction of output filters -> I expect that high switching frequencies will be required.
        – Providing VAr output – this will take many single phase topologies out of the running.

        As cost is not considered at all in the judging creiteria, I expect to see the following:
        – High density magnetics (inductors) – maybe unsing planar and or integrated magnetics (e.g. 2 inductors on one bit of ferrite). Possibly even integrated into the PCB.
        – Wide band gap semiconductors, most likely Gallium Nitride (GaN), but possibly Silicon Carbide (SiC).
        – High switching frequencies in the range of 45 – 140 kHz. Therefore the magetics will be ferrite or powder core type.
        – Fairly simple toplogies such as the H bridge, or 3 level inverters such as the Neutral Point Clamped or T-Type.
        – I totally expect to see PWM control.
        – Customised cooling systems (e.g. custom machined heatsinks), possibly even custom built fans.

        Things I feel are impractical about the challenge:
        – No consideration of cost
        – Primary judgement criteria is density, not efficiency (this is debatable)
        – Some Intellectual Property (IP) worries about Google’s rights to ‘make derivate works.’ (May not be a problem).

        Contenders to look out for: ETH Zurich, Fraunhofer.

        1. They can patents to prevent others from using the idea without paying their dues. This does sound like a blue sky University Master/PHD level project. The cost part is plain old engineering/implementation.

          Not sure about the concern with size when you are talking about a 2kW solar array. :)

          1. A patent are just a piece of paper you’ll need piles of money to defend it if it’s anything worth. As for using a cpu cooler, you have 40 cubic inches to work with, a 60W cpu cooler would take most of that

  5. I don’t understand why the emphasis on small size – assuming they are targetting solar, surely efficiency and cost are the most important aspects, and reducing size is bound to compromise the other parameters.

  6. couldnt you dead bug some of the parts to bring down the cubic inch?

    if it is just a simple ratio of dividing the watts by cubic inches of the unit then dead bugging some of the parts could say shorten the length or height of the package.

      1. To win this challenge, you’ll need massive density. Dead-bugging _through hole_ parts could save space, but we will actually see surface mount madness (0603 and smaller) as people try to squeeze every last milliliter out. You can dead bug those but it won’t save any space.

        Finally, it will come down to an arms race between a) compacting the electronics by filling every nook and cranny and b) still getting cooling air through the thing. I am really looking forward to seeing what people come up with.

        Regarding efficiency: I sort of agree with you, fonz. But I put a different spin on it: I think that by the time you can fit your inverter in the box _without it overheating_, you will have met the efficiency target by default.

        1. You know how big you can make the box, so you can pretty much calculate how much heat you can get rid off. Once you have done that you’ll realize it isn’t making the components fit that is the issue, it is that the semiconductors needed to meet that efficiency is pretty much non existent

        2. Power components are the ones that is going to take up the majority of the volume in a converter like this. Your constraint is what packages of the parts come in. They might be small bricks that you have to screw onto huge bus bars. :)

          The only place that you might find some alternatives is the control circuits which shouldn’t be taking up that much of space. SMT stuff is easy.

          Besides, this isn’t a competition for a basement budget anyways. It is more a R&D or University Masters/PHD level project. So technicians or some outside manufacturing isn’t out of the question.

  7. Unfortunately, my design requires dwarf star alloy – or neutron star material.
    Great magnetic field density, fits the size requirements.

    Sure it’s heavy, but weight is not specified in the criteria…

    1. Hm… Maybe it’s an impossible task and they know it. Suppose the best result is a much smaller and better inverter, but still not good enough to win the money. So Google are getting the R&D done, possibly with some general principles they could use, without having to pay anyone the prize money.

      Competitions in general are a way of getting much more work for your money, lots of research and work, but you only have to pay the winner.

  8. I question their heavy emphasis on size. They claim that the current “picnic cooler” sized inverters are holding back solar adoption. Seriously? Who *doesn’t* have room for this, when it comes to something important like powering your home?

    I’d say cost is what most people look at. And something 10X smaller, might be 10X more expensive; yet cost is a complete non-factor in this competition. Ruggedness, reliability, and repairability might also be sacrificed, but these too are non-factors.

    Even efficiency is a secondary consideration to size, so long the minimum requirement of >95% is met. Let’s say there’s two identical inverters, meeting all minimum requirements, and with efficiency of 97%. Shave some heatsink area off one, and compensate with a smaller area high-speed (and noisy) fan, which consumes power and reduces efficiency to 95.1%; that is the one that will win. This makes sense, how?

    1. Yea, I have to agree with you on this. The priorities for this contest seem to be more than a little skewed if the goal is to increase the adoption of Solar. There are very few situations where I can see physical volume trumping production cost in importance (airplanes, spacecraft, and a slim possibility that it’d be worth it when implementing solar in a place with extreme real-estate costs such as Manhattan). For everything else, the thing could be orders of magnitude larger without issue.

      It almost seems like the project manager for this contest had his/her heart in the right place but doesn’t know enough about the actual problem at hand to spec out the requirements correctly. The other possibility is that the stated goal of advancing solar isn’t, completely, honest and they have some alternative need for such a system that we’re not aware of yet. The really sad part is that the solar industry really DOES need radical advancement on this front as we are fast approaching a point now where the solar panels themselves will become so cheap that they no-longer contribute the significant portion of the system cost when compared to power handling equipment and skilled installation labor.

      1. Cost of exotic silicon etc should not be a driving priority because component cost is largely tied to the quantities produced and the effort put into designing low cost manufacturing processes for these components. So if a new inverter design becomes popular, it’s – what were previously considered – exotic components will reduce in cost.
        The solar goal is clearly not the only positive outcome of this comp. Inverters are useful in a wide range of technology, for example anything that contains a motor and google do plan on releasing a self driving car very soon. Considering this, you can understand why Power density is the driving factor. btw efficiency will be directly judged in the case of a tie.

  9. My funding was just approved by my institution and corporate sponsor to enter the challenge. :) The team I lead is convinced we can *at least* meet the requirements to enter. Aiming initially for 95.05% in three months time. We are very lucky to have this guy who is getting his PhD now but worked for years “designing” silicon for a chinese clone foundry, has all kind of tricks up his sleeve. We actually believe time is the main problem, as time to and from fab tends to add up. Cooling doesn’t sound too difficult and fluorinert actually solves more than one problem. I tried to recruit some guys from another institution who are some of the best ceramic people in the world but they are actually entering on their own as soon as they solve some IP licensing issues they have before getting funded. Their proposal is quite interesting using ceramics and ceramics interfaces (as in boundaries) in electronics, cooling, *and* containment, and they are actually proposing a whole new topology. Another obstacle my team is facing is our caps are within theoretical boundaries (and within manufacturers roadmaps) but do not exist yet (nor the form factor). A friend at *con tells me they can be made though.

    To be honest another problem we face is the corporate lawyers, as they foresee some IP conflicts of some sort, but even if we don’t formally enter the challenge we’ll be publishing a solution somewhere (as we are already funded) at the same time.

    Good luck to everyone!

  10. Also FCC compliance.

    And the volume is of the enclosing rectangular prism, so if your sphere is under 40″^3, it may not fit within the cube. Cylinders or domes need not apply.

    Sometimes I have to ask what is the point?

    Of this or the hackaday space challenge (that has an upper limit on the ticket to space price – if it even exists before the expiry).

    Of course someone wins these things, but I don’t see why most people bother.

    Whomever is the judge will pick something “cute” or “neat” not useful or innovative, or whatever, or the rules will be bent.

    If they were serious…

    First why the extra, extra, long timeframe? There are only so many ways to build such a box and were I to try, it would be done by October. Publishers Clearing House doesn’t have this long lead times. This is like bidding for a federal contract. Hey, maybe Solyndra or A123 can enter.

    Anything that takes over 3 months from start to judging is looking for something other than innovation and engineering and creativity.

    Second, why not win, place, and show for the three top entries that cross the finish line? There may be innovations, #3 might be more practical to produce (maybe #1 uses superconductors). Or smallest, most efficient, or most practical?

    Third, I’ve mentioned the FCC, but there is also electrical code – it needs the right colored wires, etc. Is this a bureaucratic code compliance contest or an engineering contest? How about making something and worrying about the trivial later.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.