Choke winding question

Did you play with the gap size?

The three cores modelled are the same ferrite material and the same size (RM12), they have different gaps giving them different Al values...

Cheers

You can't just buy something from Coilcraft?

They don't have much in the way of high-inductance high-current inductors. Everybody wants small, high-frequency stuff these days. I want large high-voltage high-frequency high-power stuff. Well, at least it doesn't require too many turns.

Well, that wouldn't answer my question.

Anyway, high voltage, high temperature, constrained space mean not COTS.

Cheers

What frequency are you choking?

You've just run into the Hanna curves.

RL

But my question is simpler, I probably haven't explained it well. When I use a manufacturers PC software to see what's the most inductance I can get from an RM12 core with 2.5A DC going through it without it saturating, accepting that I will need multiple chokes in series to attain 5mH, it seems that the /total/ number of turns required is much the same. Is that physics or just luck?

According to the Ferroxcube software, a 400nH core with 57 turns giving

250nH, 89T, 2.5A, 2mH, 3 needed, 267T for 6mH or 22uH/T

All the same core material, 3 different gaps for 3 different Al figures.

It seems suspicious that these three configurations come up with the same total number of turns for the same inductance at the same saturation current. Is this something simple which is well known in the choke-winding fraternity, is it coincidence, or is it an inevitable consequence of inductor physics? It may well be an obvious thing which I simply didn't know.

Cheers

What is different about the cores that give them the different ratings? Is it size, materials, what? Are the winding dimensions the same? What is the same vs. what is different?

.... Phil

Gap. The Ferroxcube RM12 core range offers a small series of cores with the same external dimensions, but with a progressively large gap in the centre leg.

The effective magnetic path length for all of them is 56.6mm, but the largest gap - 1.57mm - is equivalent to more than a metre of path length in ferrite and dominates the inductance of any coil wound on the core.

Whence the 160nH per root turn for a core which offers 5.6uH when ungapped.

The EPCOS data sheet isn't much different

but their application notes used to be a whole lot better.

is 6688 pages and 33.15 MB. It's more comprehensive than accessible.

The late great Tony Williams circulated a bunch of the more useful material to his friends some ten years ago in smaller chunks.

Saturation occurs at the same N*I for a given core shape and size, independent of gap*.

Physically, this is because Ae is the same, and you can only pack so much flux density Bsat into the core. Bsat (== Vs/mm^2) * Ae (== mm^2) == Vs (flux, integral V dt applied to the core).

The average (or effective) permeability that's best depends on a number of optimization factors, but it's in the 20-60 range. The lower A_L cores in that selection will have this property.

As you note, lower mu_eff is higher energy (because the energy is stored in the air gap, which is larger), but harder to wind (takes more copper for the same losses). (If we had superconductors, a very low mu_eff would be manageable, and we could do air-core inductors without worry.)

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

Thanks for the info. As to the cores, Clive says he is using the RM12 cores in 250, 400 and 1000 nH. But I don't see a 1000 nH core listed.

You out of your mind ??

For the same winding, increasing the gap increases sat current.

It's the reason for adding a damn gap !!!

.... Phil

I'm not using them, I'm using their design software to work out what I need. If there's no Ferroxcube 1000nH, there'll be an Epcos.

Incidentally, running the same program on a 5300nH core gives 100uH, 4 turns, 2.5A turns, 50 cores needed and 25uH/T which is still pretty close to my observation, ie I'd need 200 turns total across 50 cores for

This may be obvious to someone 'skilled in the art', but it's new to me.

Cheers

That's 688 pages.

Oops. That's what I meant to type.

Buh, why did I type N*I? I even said flux is constant later... (Thanks Phil)

Tim