Hi all,

There seems to be a noticable trend for operating converters at higher frequencies to reduce the size of the complete design. This has certain disadvantages of higher EMI, complex control etc.

does anyone know of much work that has been carried out into improving the power denisty of passives? (Capacitors and inductors?)

In the olden days (15 years ago) I roughly plotted the effects of frequency of operation upon efficiency for a straight PWM SMPS. I concluded that going much higher than 1 MHz was not productive. But with the advent of resonant switching, the dominant high frequency switching loss was reduced (not eliminated- now anti-parallel diode loss).

Parasitic effects, once an annoyance at 100 kHz, are now depended upon to create the resonant switching of the new 1+ MHZ converters. This makes each component and PCB layout critical to its operation. It is certainly job security for the engineer and the company's vendors. After all, the ultimate power supply, as viewed from management, supplies infinite current, in zero space and at zero cost.

In short, find a very good magnetics designer, and a very good PCB designer who actually knows power layout, and marry them (professionally). Only POLs (point of load) regulators have a narrow-enough operating range to actually buy off-the-shelf inductors. Everything else is custom.
Look at the blogs, LLC converter design is like balancing on a ball. With the younger engineers entering the field and without the understanding of parasitic elements within the models, things are going to be tough for a while. That is not to mention the EMI created by these under-defined components. Do not expect an off-the-shelf answer to your requirements.

I wish you the best of luck, Luke. Please learn from the older engineers.

Hi Marty,

Many thanks for the input. The have hightlighted some key issues that need to be addressed.

I perhaps should add some more clarity to my request. I was pondering whether there was any learning from the SMPS sector with regard to power dense inductors that could be carried over into the world of motor control. DC ripple currents are becoming more of an issue and are begginng to require some novel filter designs to bring down acceptable levels. Obviously this would be a custom design but it was more a theorectical question as what is the cutting edge for SMPS inductors.

The flat wire design is available from many sources now. Its' an important technology that the MW designers have been using for decades. Finally arrived on the plate for low power designs, a good thing!

The beauty of the flat helix design is that it has low dc resistance, and the proximity loss of a single layer design.

Here is an example from our workshop notes:

5 turn inductor design, one sample wound with foil, one sample wound with the flat helix.

Each winding is 4 skin depths thick.

AC resistance of the flat wound is 20 times DC resistance.

AC resistance of the foil wound is 76 times DC resistance!

So the flat wound is 4 times better on AC, equal on DC.

Costs a little more, but the manufacturers have this process easily under control now. There were many examples of this at the PCIM show in Nuremberg.

All very interesting...But what if that inductor had to be say 200 uH and carry 500A? The only viable option I found was to use a few inductors in series to bring the individual core size down. Even then, it wasn't practical for my application (interleaved buck). Foil windings or flat wire is a must.

Jay..
Foil winding at 500A could get very warm depending on the sheet insulation and layers needed for voltage gradient... Thermal conductivity must be good... Typically spacers are placed to create air flow in the winding build...This will increase window build "F" ...
Core typically will have multiple gaps to keep fringing per gap below 10%...
Most optimum designs a Noncrystalline C core with dual bobbin winding in series... If you have more than a single layer Flat wires still need to be optimized for Proximity effect and transposition will be needed in the windings...