permeability

This is a result of the BH loop having significant width.

If you are either using an ungapped structure (toroid) for low level filtering, or a gapped structure like a ferrite rod antenna that will see only small excitation, the lower, initial permeability at least has to be included in your design. If the core will be gapped and be operated at high flux levels, like most power transformers and energy storage inductors, then the maximum permeability is more useful, but you also have to keep track that you are not getting too near saturation.

A better reference might be:

I once ask a similar question regarding ferrite cores in radio antenna impedance transformers. The antenna signals are measured in microvolts and the permeability is measured with much larger signals. Also when you test the transformer the test equipment uses larger signals. So, I wondered what the transformer really looks like at the puny power levels. The group didn't think there was anything to the observation, and told me to find something else to do. But I still wonder! Mike

I was looking at making a switch mode power supply to convert 12 volts to 400 volts at 50 or 100 watts. If there's no DC in the primary, then driving it with a square wave drive at 100% duty cycle the flux density should be B = E / (N Ae 4 f) where B = flux density in teslas E = drive voltage N = turns Ae = core cross section in square meters

If that's right then I may be able to get started on the transformer design without reference to the permeability of the core material. To determine Ae I compared the dimensions of the surplus cores I bought against the cores in some Amidon literature and it closely matches the dimensions of their pot core number PC-3622-77, Ae = .0002 m^2. That tells me I can get away with as few as two turns on the primary if I drive the tranformer at several tens of kiloherz. The Amidon table lists the PC-3622-77 as capable of 90 watts at 20 kHz. It is made of #77 material, not N41 like the cores I have, but I'll take it as a ballpark figure.

I think that's right. The permeability will just affect the magnetizing current. It is flux swing that supports the winding's volt seconds.

In this sort of application, you may have to keep an eye on power lost per volume of core material.

At that frequency, you have to account for the skin effect and proximity effect on the losses of that heavy primary conductor. Twisting 7 or 19 strands of magnet wire to make that heavy conductor will lower those losses. You should use about half of the window area for the primary and also for the secondary.

Is yours a pot core?

It's a pot core with the same dimensions as the Amidon PC-3622-77 but made of N41 material instead of the #77. BGMicro had these cores a while back with plastic bobbins and windings on them, something like 50 cents apiece.

(snip)

Material N41 appears to be a "power" material (good saturation flux and fairly low losses) so you should be able to get a t least 90 watts through it. Do you need a constant power flow, or will the power have high peaks and longer periods of lower power? If so, you can use more turns of thinner copper and raise the copper losses at high power, but lower the core losses all the time.

I understand you're saying with more turns there's less flux swing and less core losses. Possibly this core's window will be big enough that I can get some extra turns in there without compromising with the wire thickness.

By the way, is there an established way to quantify core losses based on the properties of the transformer, frequency, voltage etc etc? About all I know so far is that as the frequency goes up so do the core losses, which may call for reducing the flux density. But it's all pretty vague.

Generally, the manufacturer supplies a graph of loss per cubic cm versus flux swing and frequency. I am having little luck finding this graph for this material, but it should be pretty similar to other power materials of the same permeability.

Fair-rite supplies such graphs in its catalog. Perhaps you can find one of their materials that is close to type N41.

But regardless, running your transformer at no load produces about the same core loss as running it at full load. Just the copper losses are missing. So you can check the temperature rise from core loss before you add in the copper losses.

Thanks, John.