Common myths about high frequency transformer and inductor design are explained on this page.

Magnetics parts are frequently misunderstood, and almost always made out to be much more difficult than they really are. Good magnetics design does NOT need a lot of complex analysis. The problem is that there are so many variables to deal with, and so many small details to know. Like many other aspects of power electronics, these details are seldom written down in an accessible form. The list below is, unfortunately, just a start.

Magnetics design is also greatly complicated by the fact that there is so much misinformation out there, in data books, papers, and elsewhere.

 

Myth #1: Fill the window with copper for an optimum magnetics design

Rarely will it be best to fill up the entire window area available with copper. In many inductor and transformer designs, it is not uncommon to find that adding an extra layer of windings, using a heavier gauge wire, will actually increase total losses in the part due to the proximity effects in the wire. Don't be concerned about leaving space in the window, even if it's only 25% full.

This myth is a hangover from line-frequency magnetics design days.

 

Myth #2: Core loss = copper loss for an optimum transformer design

It's not unusual to have core loss and copper loss differ by an order of magnitude in some designs. Don't worry about it - this doesn't mean that your design is bad. There are a lot more parameters which define optimum design; in a high-frequency switching transformer, and making the core and copper loss equal is usually the least of the concerns.

This myth is a result of looking at designs which are fairly homogenous from a thermal standpoint - such as in 50/60 Hz designs. It certainly doesn't apply to cases with a few thin layers of exposed wire such as we often have in high frequency designs.

 

Myth #3: Leakage inductance should be about 1% of magnetizing inductance

Never be satisfied with this as a design criteria. Calculate the expected leakage, measure it in the finished transformer, then do everything you can to make it as low as possible without making other critical parameters of the transformer much worse (e.g. capacitance). In some cases, leakage will be less than 0.1% of the magnetizing inductance. In some cases, with a gapped core, it may be close to 10%.

And never, never, use this criteria as a specification to the magnetics vendor, it will indicate to him that you don't understand or really care about leakage inductance. You should specify leakage as an absolute number, plus or minus a percentage, typically about 20%.

 

Myth #4: Leakage inductance is a function of the core permeability

Some people think that adding a core to a transformer winding will
couple the windings closer together, reducing the leakage between them. Others have heard that the core will couple with the fields between the windings, dramatically increasing the leakage.

The fact is that the leakage inductance of two concentric windings
designed for a modern switching power supply transformer is almost independent of whether the core is present or not. This is a surprising fact - adding a material with a permeability of several thousand close to the windings has very little effect on the leakage inductance, changing it by no more than perhaps 10%. Try the experiment if you don't believe it - measure the leakage inductance of two windings with and without the core present.

 

Myth #5: An optimum transformer winding has a current density of 500 - 750 circular mils per amp

If you're not from the US, don't even bother calculating what a circular mil is. Current density is irrelevant. What matters is how much dissipation is in the wire, and whether there is sufficient cooling to keep its temperature in the acceptable range. Given the extreme types of cooling which may be used in power supplies,
ranging from liquid immersion at one extreme, to operation in the vacuum of space at another, the appropriate current density in the wire will vary widely.

After designing practical transformers at Ridley Engineering for more than 20 years, frankly we have no idea what the current density of any them was. The only relevant question is: how hot is it?

This myth results from an artificial constraint placed on transformer design to try and simplify and reduce the number of variables available.

Myth #6: Primary loss = secondary loss in an optimum transformer

The losses might be similar in some cases, but don't worry if one is much lower than the other. Again, the only thing that matters is how hot the windings get with your type of cooling. Other parameters which are important to you will determine how balanced the losses in the windings are.

Myth #7: If the wire diameter is less than the skin depth, there are no significant high-frequency losses

A very misleading and dangerous assumption. If you have a lot of layers in a winding, you can have very large proximity losses even with very small wire gauges, well below the skin depth in diameter.

Myth #8: The open-circuit resonance of a forward transformer should be much higher than the switching frequency

It's a surprising fact that the open circuit resonance is irrelevant,
it doesn't even matter if it is less than the switching frequency!
How do you know this is true? The ideal core would have an infinite inductance, and with a small but finite winding capacitance, the resonant frequency would then be zero!

What IS important is the short-circuit resonance. This should be at least two orders of magnitude higher than the switching frequency, if possible.

Related page: Transformer impedance measurements

© copyright Ridley Engineering, Inc. 2007