How can you reduce the size and weight of power electronics for aerospace?

I would qualify the statement "By increasing the switching frequency, you can reduce the size and weight of these components, as well as the losses and heat dissipation." by adding: However, this is correct up to a point. As the switching frequency further increases inductors and transformers may in fact become larger and their losses will increase. This is because the losses of the ferrite materials used to construct these high frequency components, is increasing with frequency.

Higher switching frequency is no panacea. There is a lot of missconceptions regarding magnetics on this topic. Many people assume that they will get smaller components, as the total losses will go down and a smaller component would be able to handle the heating generation. The reality is harsher: while a higher frequency will reduce the magnetic energy stored, the total losses and size could increased. This is due to ferrite materials losses increasing with frequency, skin depth in conductors is reduced, so thinner wires must be used (e.g. Litz) which increases the number of proximity effec interactions, increasing total losses. Additionally, creepage distance and clearance increase at higher freq. so more space is needed for insulation.

We haven't quite got that point. We now have switchers at 10's if MHz and also at 100's of MHz. My IEEE paper with Georgia Tech shows what happens to magnetics size. Enpirion had shown us this years ago with their 4MHz monolithic switchers that included inductors. I agree there us a point if diminishing returns a, or even increased loss, but we are far from that limit.

Regarding the magnetics. Most designers have been doing the design by hand for decades. However, when you try to reduce the size and increase the frequency, the parasitic effects are attenuated in the classic designs and have a big impact. These effects like proximity losses or additional core losses are hard to predict manually. So, the best teams are using advanced modeling and FEM approaches to reach the right values.

Switching frequency is a balancing game. As you increase the frequency you switch the power dissipation (losses) from from the magnetics to the the switching device. assuming you've chosen the lowest loss devices for both, the best efficiency is achieved when you've effectively balanced the power dissipation between the 2, evenly distributing the loss.

GaN is very useful in many applications, but not high enough voltage for most EV. Still wide band gap is advantageous, but using SiC, not GaN

GaN MOSFETs have already found some niche applications, e.g. cellular chargers. However, they have not been applied in the, much promised, high voltage high power products such as EV inverters. Is there a hidden reason for that?

The semiconductor industry has done a tremendous acceleration in the adoption of new technologies. If the specs permit the use of these technologies, the size can be dramatically reduced.

I believe the reason GaN FETs have not taken off in the higher voltage applications is due the SiC FETs squeezing them out in that application. GaN typically has slightly lower losses (lower rdson) but are just a bit less stable than SiC. SiC is still way more efficient than typical Si FETs, not typically as good as GaN but good enough and also more stable. So, in high voltage applications where a little more loss can be tolerated SiC is typically a better choice. In lower voltage applications where minimizing losses is more critical and out weighs the stability issue GaN FETs makes more sense. It's a matter of practical application.

Consider the metallurgical structure of GaN and how easy it can be to processed as a resilient medium for ionic energy dissipation for extra orbital transport. The effects on the orders of braking radiation (Bremsstrahlung) as well as up to Ionic winds and soliton waves which have higher pulse pressure as traveling further in the atmosphere, it becomes important to consider applications for providing sufficient EMI shielding.

This is true, but the modules are very high power density. See our 0.8V/2000Amp core power board at SC23 next month or at DesignCon. We designed, simulated and fabricated boards using modules from two different companies (ADI and MPS). The have very different topologies, which made it interesting. We'll present a paper about these 2000Amp Designs at DesignCon.

In as much that size and wight are important, cost is in the real world is of major concern. So when suggesting a design approach one should reveal how competitive is it in terms of cost.

Using this modular approach is effective for distributing the power dissipation (heat) around the board and is effective when size and weight are less of a concern. However adding additional supplies will always add space and or weight to the design.

To reduce the size and weight of aerospace power electronics, the power system architecture plays a crucial role. Traditional centralized architectures involve a single large power converter supplying multiple loads, simplifying control but increasing size and weight. Alternatively, modular or integrated architectures with multiple smaller power converters distributed throughout offer flexibility and scalability, reducing the size and weight of power converters and wiring. However, this approach necessitates more complex control and communication schemes.

So far nobody's mentioned the "easiest" way. That is, to reduce the load. The load represents the mission and the power electronics has no say about that (nor the realism of requirements flowed down). For all you know as "The Power Guy" there could be several layers of "up-margining" against imagined worst cases on the board, box and system level power requirements. At the other end of the chain you have the problem of rejecting waste heat radiatively to vacuum. That is at best, emissivity, area and temp rise. You want to raise heat sink temperature to get radiation up. But that is not what the electronics want. There could be a play for a heat pipe and heat pump tandem cooling (presuming a spaceworthy pump) to improve that thermal chain.

To minimize the size and weight of aerospace power electronics, optimizing the control of the power system is crucial. Traditional control methods involve fixed or adaptive algorithms based on predefined models, offering reliability but often being conservative. Alternatively, employing advanced methods like model predictive control, machine learning, and genetic algorithms enables dynamic adjustments based on real-time measurements. These advanced techniques enhance the system's performance and efficiency, contributing to a reduction in the size and weight of power electronics.

The reality of cooling Power Electronics is that is difficult to predict. There are modern tools, like Ansys Icepak or Trafolo (which supports multiphysics simulation) that helps with the design, but their adoption is slow, as Engineers clasically preferred to build a prototype and test it in the laboratory, even if that supposed many iterations and months of work. Simulations can be used instead to reduce the time, and with the proper programming skill it can ieven be automated. Regarding Magnetics, you can even find online tools, and like it happens in many fields today, the best ones are Free and Open-Source!

Heat dissipation has been key in the power supply design. Since the first high-power density converters without fans, companies worked very hard in the mechanics of integrating all the components and keeping them working. The only difference between the two competitors might be the cooling magnetic

To reduce the size and weight of aerospace power electronics, the packaging and cooling of power devices play a vital role. Traditional methods involve bulky and heavy materials like metals and ceramics, providing stability but increasing size and weight. Innovative techniques such as 3D printing, additive manufacturing, and microfluidics offer a more compact and lightweight alternative. These methods enhance thermal management and reliability, allowing for customized and efficient power electronics in aerospace applications.

These are essential for considering the defined topology a power converter will be expected to be in terms of utility. However, in application, it becomes worth considering how the power utilized will be expected to behave with the addition or subtraction of the most fundamental elements that make a design feasible. These are all excellent factors to list. Of these however, air-gap and window utilization factor influence the efficiency (at least in terms of the magnetic components), but allowing for integrated cooling solutions will need to be adjusted accordingly. To facilitate other aspects of final design, profile height is important for improving thermal and aerodynamic feasibility…

The Aluminium heatsinks and the metal casing are the heaviest.I would use a plastic cover for the case with minimal metal for grounding.The big FETS and big printed circuit boards can be replaced by smaller surface mount equipment which is much lighter.

The other section that should be here but isn't discussed is converter topology selection; soft-switched converter designs make it possible to increase switching frequencies (hopefully up to or close to the point of diminishing returns with regard to magnetic size). If a power switch is turned on or off with voltage and current across it then that switch briefly dissipates the full power across it, which generates a lot of heat. Wide bandgap semiconductors help because they can be switched much faster, but using resonant or quasi-resonant topologies (LLC, DAB, or many others based on modified hard-switched converters) allows the voltage or current through the switch to be naturally driven to zero before switching and thus reduces losses.

Power electronics optimization is a mix of experience, knowledge of the theory, and component choice. examples are... Use of soft switching and resonant converters, Magnetics core choices optimized for hot spot temperature Windings design, eg wire types, layer interleave, winding sheet converter topology choice of switch frequency AI will not replace the decades of accumulated experience. Jon