Magnetics question.

I think some common mode chokes have intentionally increased leakage inductance?

It is a 1:1 transformer, albeit not optimised for such a job. Its specs might not suit you of course.

NT

High permeability could be a power issue.

Might work, depending on the leakage inductance. Sounds like it already works.

I really like the dual-winding inductors, like the Coiltronics DRQ series. It's a good 1:1 transformer. Lots of people make drop-in equivalents. Coupling is good, 0.99+. Surface mount.

Magnetics are a nuisance, and custom magnetics are a giant nuisance.

Maybe.

There are kind of four sorts of transformers:

- Made to store energy. Closely coupled inductors. Example: flyback.

- Not made to store energy. Example: mains transformers, forward converters.

- Made to filter energy. May have high leakage. Example: CMCs.

- Made to resonate energy. Inductors with low coupling (k < 0.95, say). Example: LLC resonant circuits (with custom designed transformer).

Bifilar wound CMCs (e.g., Bourns SRF0905, most SMT "data" chokes) will have high inductance and low leakage, making good transformers. But they'll have low isolation (75-250V?).

Split bobbin parts (e.g.,

The leakage is such that, if you try using it for a forward converter, you only get about 1W at reasonable voltage drop.

You can't draw more current because you're limited by leakage. You can't drop frequency or raise voltage because you're saturation limited. You can't raise frequency because you're leakage limited.

You can resonate with it, but the load varies with, well, the load -- and now you need much more control (i.e., a controller and an opto feedback) to make it work.

BTW, that CMC I tested, I hipotted to some 15kV before it flashed over. Impressive for such a small part.

Tim

CMCs are more akin to transformers than coupled inductors, so you'd use them in a forward converter, not flyback. No worries!

Tim

Why do they?

Cheers

To get some differential-mode filtering, too.

This is a line driver transformer operating at 15kHz to 25kHz at 60V p-p. The CMC is just one I had around, it's 6.8mH on a 30mm toroid and could do with being less, probably nearer to 1mH. I'll try some more tomorrow.

It would save a lot of arsing around, not just winding but specifying, drawing etc. Item cost doesn't really matter.

Is the leakage due to the windings being separate?

Cheers

Sounds like a lot of flux, which will need to be pretty big (and yeah, 30mm, a pretty sizable core).

Flux is never rated on these things, but if you can guess at the core (based on drawings or pictures), you can get a rough idea of turns, A_L and flux. At least the core type is known: they're almost always high mu (>= 10k) ferrite.

Yes. Essentially, it is the inductance of the space between windings, independent of core*.

- If the winding is bifilar, then it's the stray inductance of the wire pair as a transmission line.

- If the windings are in layers, then the same applies (you can wind a two-wire transmission line edgewise).

- If the turns aren't so coincident (different pitch, multiple layers per section, bank or pi windings, etc.), then there's more distance between windings, and still more leakage. (All the extra distance makes for a rather high impedance transmission line, if you can call it that anymore.)

- If the windings are on opposite sides of a toroid, you can imagine it without the core: two bent solenoids side-by-side are not at all well coupled, so the leakage is quite high, essentially the self-inductance of each (coreless) winding alone.

(*Obviously, this takes a lot of qualification, and is easily defeated, say with funny core paths. But it works for conventional designs where k is near

Tim

It'll be a leaky low pass filter. other than that, sure.

Pray tell, why?

Leakage is that part of the flux that threads one coil without threading the other.

The quick and dirty way of measuring it is to short one coil and measure the inductance of the other. Shorting the first coil doesn't get rid of its series resistance, which complicates the measurement if you want to get a precise result.

If both coils are wrapped around a high permeability toroid or it's topological inverse, a pot cot, the coupling tends to be high - around 99%.

Tim Williams claims that split bobbins give you relatively poor coupling, and it is worse than you get with bifilar windings, but it can still be pretty good.

Thanks, I just did this on a few different CMCs around 1mH - 6.8mH and the 'secondary' shorted reading is around 1% or less, so I guess leakage isn't worth worrying my pretty little head about.

Cheers

I recognize PADS. They made a free version 25 years ago, running under DOS, and with a limit of about 100 components. I found that if you got anywhere near the limit it would crash spectacularly with random lines everywhere. I wondered if that happened with the full version but only with more components.

It had a great user interface, designed by someone who understands that the number of mouse clicks is supposed to be minimized.

Line-input common-mode chokes have two coils wound far apart, on opposite sides of a core, to obtain a high insulation-voltage rating. If they're used as a transformer instead, we're forced to live with high leakage inductance, but we get low capacitance.

Coupled inductors, OTOH, are bifilar wound and consequently have a very low leakage inductance, and high winding-to-winding capacitance. They can also have quite high magnetizing inductance, for energy storage. John likes them, me too.

Both beasts have a good 1:1 transformer ratio, but with dramatically different specs.

Whoa, 22M resistors, each with 467 volts, in your orange LED driver, yikes! What kind are they? Hey, are you ignoring the official voltage rating, because you've observed that high-value resistors don't change their resistance much at HV, aren't dissipating much power, and in general just don't seem to mind?

I am sure JL saw the problem instantly, and has specified high voltage types e.g.

(1206 are 500V working)

Some CM chokes have two windings on a figure-8 core. They must have huge leakage inductance, coupling 0.5 or less.