Google are always shaking things up. And they are shaking the digital power electronics and low power loss converter space with this competition.
For those of you who don’t know, the Little Box Challenge is a competition where you can win $1 Million USD for creating a 2kVA inverter with a power density of 50W per cubic inch. For those not familiar with this unit I calculate this to be 3.4 kW per liter. Other specifications are around efficiency, AC voltage and current THD and noise, and the DC side current ripple. The inverter is a single phase device.
This is an interesting challenge. It is also put together and presented in an interesting way. The marketing website has pictures of hybrid cars, laptops and washing machines. The inverters in these are three phase inverters. The fan in your laptop is driven by a three phase inverter as is the motor in your hard drive. Plugin hybrid car AC to DC converters for battery charging are single phase but they are probably not inverters and in a strange juxtaposition, there is a picture of a hairdryer. Then I realised I had misunderstood. The inverter is to power the appliances and charge the electric vehicle.
When you read on it becomes clear that the the little box challenge technical problem is single phase and more to do with solar inverters or probably standby inverters for your household single phase loads. Sorry to those at Google for misunderstanding the picture.
The minimum efficiency is do-able now with silicon and/or SiC. I have an in-production 2.88 kW isolated digitally controlled AC to DC converter that I had downstairs on my desk now. It is more efficient by some distance than the minimum efficiency required in the contest. It has SiC diodes in the PFC stage. Aside from that it is all silicon. And it is from six years ago. Just measuring it now and it is 10 inches by 1 and 5/8 inches by 4 and 3/4 inches for a total volume of 77.1875 cubic inches. This gives 37 W /cubic inch so it is in the ballpark. All of you who do this stuff know that to get to 50W per cubic inch will take some effort. Having said that looking in my power supply there is some air space.
The Littlebox requires a a DC side ripple current that is quite low. I haven’t worked out the 120Hz filtering volume needed to do this with capacitors but am assuming the volume will be significant. By adding the DC side ripple requirement the challenge becomes really quite different and is more about energy storage replacement or the equivalent of valley filling. The solution to that might be some sort of integrated reasonable switching frequency active filter on the DC side that only operates at certain conditions so the 20% current ripple can be met. Perhaps the active filter could gyrate the storage capacitor. Or maybe some sort of pulse number increase like the old HVDC ripple current re-injection schemes. I might dig a few of those papers out to see what use they’d be.
The AC side specification doesn’t seem too harsh. A good inverter feedforward lineariser for PWM precision, dead-time and minimum pulse width with a good repetitive or resonant notch filter based current controller will meet this THD and noise requirement. It’d be good to know the measuring instruments noise measurement approach. The magnetising current of the transformer is supplied by the inverter so this will mean that getting both the current THD and the voltage THD below a certain limit may not be actually possible. This will depend on the transformer magnetising characterisitcs.
It is possible that you will be able to meet the power density with the minimum efficiency as the cooling might be OK. That said the description of the testing doesn’t cover exhaust air temperatures as would a UL standard. There may be room for that kind of really high temperature exhaust air that UL are always so troubled by. It might be that the contest rules need to be tightened on this.
For those of you considering this from outside the US make sure you get some US fuses as the UL and IEC definition of the current is significantly different. Maybe a part number specification for the fuse for the testing would be useful. US fuses will be hard to buy outside the US in any case.
It’d also be good to get some idea of what the transformer in the test is so any EMC filter design can be optimised for that transformer.
The solution to the DC ripple current was solved by Tesla (Nikola – not the car company) back when he was doing his thing. It is three phase AC systems where balanced three phase currents give constant power flow and so constant DC side current. If only we had three phase appliances.
Unfortunately most of the people who legitimately have a good shot at meeting this challenge will not be allowed to as they are already doing this type of stuff for a job and so will not be able to meet the disclosure requirements in the terms and conditions.
Other issues might be getting your gear into the US.
For power density trends, analysis and a good commentary on the current state of power converter density take a look at http://www.hpe.ee.ethz.ch/uploads/tx_ethpublications/IEEJ_PowerDensity_Paper_FinalRevised_03.pdf.
The paper presents data that shows that the switching frequencies that maximise power density are surprisingly low.
Maybe Professor Kolar and his team at ETH will have a go at the little box challenge.