where is pin 1?

CM chokes are specified for inductance and for max current, which seems to be purely thermal.

People never seem to specify actual core saturation currents; looks like they assume that the net flux is zero. They typically saturate at a few per cent of datasheet current.

One could test the current rating without saturating a 2-wire CM choke, but I can't think of a way to test a 3-winding choke without saturating the core.

Presumably it is the Curie temperature of the core that matters, and you can measure that in an oven - inductance will drop with rising temperature and fall away very rapidly as you get close to the Curie temperature.

All you need to work out is how much heat you can get away with dissipating in the windings before you push the core above it's Curie temperature.

You will be able to dissipate more heat in all three windings before you overheat the core than you could in just two, but there won't be much in it.

It's pretty much a single solid thermally conductive lump.

Sticking a thermistor on the core through the plastic former will give you it's temperature explicitly.

That 3 way choke doesn't look symmetrical to me in the PCB pin layout.....

It's 3-way, 120 degree, symmetric. But there's no way to install it wrong... it always works.

Ferrite cores are thermally limited to < curie temp, with the higher permeability types having the lowest limits (<110C).

If it's an amorphous core matl, that isn't an issue, but Permeability drops with current rather rapidly in ungapped structures of any form.

You can run DC current through two windings in a loop with inductive decoupling while measuring L on the 3rd to get an idea of whats going on during lower frequency imbalance to your common mode element..

This choke cannot force net zero flux - it merely assumes that there is no 4th path. If your inverter 'wiggles' are large and low- frequency, RF will burst through on the peak imbalance.

RL

CM chokes can use very lossy magnetic material and still do their job well; it could be a good transformer if you use it as a current transformer, but not as an AC voltage source. Most 'planar transformer' applications DO NOT work well with lossy magnetics. RF would cook 'em.

Neither could I at 2:19am local time. As soon as I'd gone to bed I realised that three sine waves of identical frequency and amplitude but phased 120 degrees apart would do the job.

The trignometric identity you need to prove it is sin(a+b) = sin(a).cos(b) + sin(b). cos(a)

So I gave John Larkin his answer but he won't notice it because he has got me kill-filed. Sad, when he didn't know enough to work it out for himself

It must be highly provoking, all that silence.

RL

On the contrary, if John Larkin provided more of it, I'd be appreciably happier. On this particular occasion he had come up with a question worth answering, which is rare, so I've missed a rare chance to contribute something constructive.

The copper-loss based current rating, the thermal rating, can be done with the windings in series, and ignore saturation. And the series connction can reveal saturation behavior too, and I guess that's all I need to know. I'll chip a bit of the core and see if it's ferrite or powdered iron; it's sure heavy.

Sloman is just an insecure insult generator, and he'd rather insult than either learn or be right. There's no point on wasting time with idea-free toads like that.

What you said - at one point - was

"One could test the current rating without saturating a 2-wire CM choke, but I can't think of a way to test a 3-winding choke without saturating the core."

Neither could I when I first looked at your post, but after I'd slept on it I realised that feeding three phases of the same sine wave, spaced 120 degrees apart would do the same job on a 3-winding choke that two phases spaced 180 degrees apart do on a 2-winding choke.

There's no insult in that particular observation - or at least none that I can detect. I think I learnt something, and I'm tolerably sure that the observation is correct.

That does seem to be an idea that you didn't manage to come up with, so I may not be quite as idea-free as you like to think. I've got a couple of patents - not a many as the seriously creative people I've known - so I'd guess that I'm not entirely idea-free.

Getting help when you need it isn't an ego-boosting exercise, but it can save quite a bit of time.

If series windings are loaded out of phase, the effect of any flux imbalance (due to leakage) AND thermals can be monitored on the third winding.

RL

Judging from the Al values, It is probably ferrite - amorphous parts would be roughly 10x for same turns count, but there seems to be a lot of plastic present in the core volume. Weight suggests amorphous matl.

Iron dust is not used where high permeability is the criterion.

RL

Given the three separate clumps of windings, the leakage inductance, one winding measured with the other two shorted, is an impressively low 15 uH. Seems like common-mode chokes use very hi-mu cores.

In my application, I'd actually appreciate more leakage L as part of my EMI filters. The waveforms are ghastly.

CM chokes, working with safety-limited capacitor values, are largely ineffective outside a 150Kz-1mHz range.

The leakage, working on DM currents can use larger capacitors, if these don't contribute to pick-up loop area. This is in roughly the same frequency range - but unwise component values can produce peaking in the suppression band as CM and DM components react with each other and their series/prallel combinations.

Have you got a 3 phase LISN? No point in guessing if you can see effects iteratively.

RL

I'm not using this on a conventional AC power line. My box is a power source, a 3-phase permanent-magnet alternator simulator, and it drives a FADEC shorting/shunt regulator. Simulation shows a 6-amp RMS ground loop, and the 3-leg CM choke reduces that to 100 mA.

The choke is a horrible beast to put on a PC board.

Aviation regs are abundant, so you should look up actual requirements, standards and test gear.

RL

The alternator, regulator and loads will be a hell of a lot more difficult to simulate accurately than the interconnecting filter circuitry.

RL