Something I've often wondered about. . . .
If it wasn't for the constraints of size and acoustic noise, what would be the most efficient frequency to use for a DC-DC converter?
Frequencies seem to be ever increasing, but with it comes core loss, skin effect and switching losses, and the need to use ferrite which can only withstand a tenth of the flux density of silicon-iron.
Low frequency problems are mainly copper losses.
Aviation uses 400Hz, but is that because it is still the best for efficiency , or just because it was the best frequency to use when the standard was written?
(For want of an example, I'm thinking of a DC converter from 250V to 50V at 100A output)

 

I would look at 50khz on the low end and 200khz on the high end.

 

400Hz was chosen for aviation as a huge improvement in efficiency over 60Hz primarily in weight. But that was the olden days and technology has continued pushing the frequency higher for the benefits it gives in efficiency, materials and size.

I don't know enough to say if there's an upper limit to gains with frequency. Seems not.

I would look at 50khz on the low end and 200khz on the high end.

It seems frequencies are now well above that, eg
https://www.ti.com/lit/an/slvaed3a/slvaed3a.pdf?ts=1727743757893

 

I think the high frequencies in use today are mostly for size, but also for ease if filtering to smooth DC.

 

I have done 1mhz and 2mhz but not at 100A. There are some real nice GAN MOSFETs that run very fast.
If you don't know where to start, then start at 50khz.

 

i think BobTPH hit it on the head.... small circuit size, fast response and low ripple

when circuits are not just small footprint DC-DC converters...when using longer wires, frequencies remained at rather lower end of spectrum such as 8-16kHz (industrial servos, variable speed drives, various inverters for welding and plasma cutters etc)

 

(....I'm thinking of a DC converter from 250V to 50V at 100A output)

I misread the scale of your spec! I think switching losses will dictate a modest frequency at that power level.

 

400Hz is what could be best produced from an engine driven alternator in an aircraft. It also allows the use of normal production machines. The higher frequencies work well with electronics but are not easily possible with mechanical generation systems.
For a 100 amp inverter the lower end of the spectrum will probably allow easier windings with that larger diameter wire. Unless you use LITZ wire.

 

I think 400Hz was also a compromise between the reduction in size for motors and generators compared to the increase in eddy current and magnetic hysteresis loss in the magnetic material, as operating frequency is increased.

 

I was thinking that if I reduce the frequency, then I reduce switching losses, skin effect, proximity effect, and I can probably use a thin-lamination iron core. That would mean a Bmax of 1.7T instead of 0.2T for Ferrite, so the core wouldn't necessarily have to be a lot bigger.

 

Referencing posts #8,9,and Ten, it will be a trade-off. That is a big part of what engineering is all about. always the conflict: Size-Cost-Weight- and Power. SCWaP for short.
So it will require an evaluation to see what frequency will be the highest without excessive energy spent on magnetizing and demagnetizing the core material.
Of course the availability of the optimum material, as well as the cost, may also become important.
I have a 48 volt power supply built of several 20 amp modules that provide redundancy and allow repair without a shut-down, by replacing a single module. Rather obsolete but still useful for HAM radio application, powering an amplifier. Acquired as surplus it made sense, but not if it had been new.