50 kW/liter power converter: dream or reality?
Why a power density of 50 kW/liter?
Let’s imagine a 100kW electric vehicle: the volume occupied by the power electronic converter will be superior to 20 liters with current power density (typically 1kW/l to 10kW/l depending of the application). However, with a 50 kW/l converter, it will only occupy 2 liters! You could reply that 20 liters is not a big deal on a passenger vehicle, and you are right. The volume is not the main purpose of reaching a 50 kW/liter power converter…
The problem is, as always, the cost.
On one side, the converter intrinsic cost is highly correlated with its mass. More power integration leads to cost reduction on raw materials. On the other side, electrical energy must be used as efficiently as possible as every kWh is paid. In conclusion, the aim is having the best efficiency with the best power density. However, these two constraints are opposed. Power converter integration are facing a huge issue. Indeed, the more integrated power converter is, the more the heat losses are concentrated in a given volume and the more it is difficult to cool the converter. Power semiconductor devices buried inside a PCB are more difficult to cool than power module attached to a big air cooled heatsink. The power density is an image of the converter capability to evacuate heat.
A higher power density means a high efficiency (low losses), a reduced cost in raw materials, and a payload increase in the transportation field with the volume/mass reduction. The power density is a key performance indicator in power converter design and “the grail” of 50 kW/l is a revolution: 10-time current standard!
How to reach the power electronic “Grail”?
- Wide band gap (WBG) power semiconductor devices: the intrinsic properties of WBG materials such as SiC (silicon carbide) or GaN (gallium nitride) are much better than Si (Silicon) devices. The performances are increased, especially in terms of losses, allowing for more efficient / compact converters development. However, the WBG semiconductor devices are still not as mature as the Si ones. The technology must be improved in term of performance, reliability and cost reduction. This technology also raises issues on the semiconductor environment as the fast switching speed demands a careful implementation of the switching loop and causes electromagnetic interference (EMI). The diamond is considered as the perfect WBG material but although progress has been made in recent years, many challenges have to be addressed before considering industrial use (a summary of the state of the art and perspectives is indicated at the end of the article).
Power Converter trade-off between efficiency, cost, mass, volume
- Integration: the compactness levels in <1000W power supplies are staggering. 3D interconnection, chip on DBC, embedded power dies in the PCB, etc. The technological deployed is impressive and full of original ideas. However, these technological advances are difficult to replicate to higher power level as the requirements in current rating, voltage isolation and losses dissipation are harsher. To what extent is it possible to reproduce the solutions used for higher power converters?
- Design automation: the power converter design must be optimal in the goal application and the room of improvement is significant. Power converters are complex systems where a single parameter change has impacts across the whole design, and the number of degrees of freedom is considerable (architecture series/parallel, power semiconductors, switching frequency, modulation strategy, filtering,…). The best power converter is a trade-off between key performances indicators of the application (cost, mass, loss, volume,…). PowerForge has been developed in this context. PowerForge addresses the switching cell - the core of a power converter - and its electrothermal environment (filters and cooling system). Based on a design operating point and key specifications (semiconductor devices, topologies, switching frequency…) PowerForge evaluates the performance and the cost of the power converter.
Below are a few challenges that you can manage into the software:
- Compare Si IGBT with SiC MOSFET by considering the filters cost, mass and volume for a given current ripple.
- Benchmark 2-level topology with multilevel topology (NPC, T-type, Flying Capacitor) to evaluate the semiconductor cost reduction (due to the non-linear dependency of cost versus voltage rating) and the impact on filters.
- Take the machine characteristics into account for the power converter design.
Benchmark power converter solutions: topologies, power semiconductor devices, modulation strategies.
Power automation, integration and wide bandgap devices are the keys.
Nicolas, Ph.D., CEO
PowerForge by Power Design Technologies
References
- Diamond power devices (Donato, Rouger, Pernot, Longobardi, Udrea): https://iopscience.iop.org/article/10.1088/1361-6463/ab4eab
Litre is volume, kg is mass. Your graphs show mass. 50kW @ 99% efficiency is 500W heat to get rid of...
Scientist (retired)
4 年50kW/liter is "only" 3 times more than the state of the art 5 years ago (ETZ, Kolar & friends). Looking at their designs, new dielectric and magnetic materials would do the trick. The materials R&D is either virtually non-existent or glacially slow, probably because there is no quick money to be made. As magnetics can be replaced by use of SCCs, my best bet is new ceramic materials, combined with technology to manufacture multi-layers with very small plate distances and using new edge soldering techniques.?