Electronic – build a high-frequency SMPS choke out of an arbitrary coil of wire

boosthigh frequencyhigh-currentinductorswitch-mode-power-supply

Suppose I want to build a boost converter that runs at 100 kHz, with an input voltage of 600VDC and an input current of 100 amps. (This is hypothetical, I know the semiconductors involved would be exotic.) I'd need a pretty custom choke to do that; laminated steel probably won't handle that kind of frequency, and there's not much off-the-shelf to be had in this power and frequency range. So I'll build my own! What could possibly go wrong?

One possible way of building a choke to handle this is to use an air core. No core losses if there's no core! No component sourcing issues, less knowledge of magnetics required. The obvious down-sides are that I'll need many more turns of wire, more volume in my chassis, there will be higher resistive drop in the path… so maybe it's not a cost-effective approach. But let's assume it is for a moment.

So now I'm building an air-core choke out of something like 6AWG welding cable. Of course, skin effect will kill me, so I really need some sort of equivalent flat wire or heavy foil. The inductance calculations are straightforward, just a matter of turns and geometry. But what's going to happen when I actually try to run that in my hypothetical boost converter? Am I prone to insulation breakdown? Do I need to vacuum-impregnate the whole thing to with varnish to avoid that? Is there some other reliability problem with what I'm describing?

Basically I'm asking, what black magic am I prone to run into when running high currents, high voltages, and high frequencies?

Best Answer

Yes you could use an air cored coil .BUT there are some issues.1 you will want to shield it because of EMC and it could muck up other parts of your circuit.Air cored coils are worse for radiation than ferrite coils.Now if you do a toroidal air cored coil which I havent done you will get a worse resistance to inductance ratio which at 100KHz will be crippling.2 Your real SMPS coil carries the DC load current and some AC current .The AC current heats the wire more than expected due to skin and proximity effects.In fact its possible to design a coil that has a AC resistance 10 to 100 times its DC resistance.So you must design for low AC resistance .Litz wire ,Single layer ,Brooks dimensions,Spacing turns by one wire diameter are some of the things that are accepted ways of minimising AC resistance .The importance of this AC resistance stuff depends on your convertor waveform.You may set up your convertor to keep the ratio of AC to DC currents reasonable.3 your coil will be bigger than a ferrite which may not be a problem when the size of your hypothetical product is determined by heatsinks anyway.By the way I did a 36 to 12V buck with an air cored coil that produced 5 amp .It was fine.