HELP! AADE L/C meter

Have you worked with glass / glassy core material? I would thik that stuff to be on the other end (except in cost..).

Yes, but -- ferrites (literally, chemically speaking: ferrate(III), compounds of iron) are the only type that are sintered.

When "powder cores" are talked about, they are exclusively referring to resin-bonded metal powders.

You can't sinter a metal powder, because you get, well -- a metal block. So your eddy currents come back, and it's a waste of good kiln time.

Which is actually quite relevant: because, as the resin breaks down (time

• temp), the material essentially begins to sinter, and eddy losses goes up, and temp goes up, and eddy losses go up, and...

Ferrites, however, do not age as far as I know.

There are quite many kinds of powders in use, all iron (or nickel) based, but with various other things included (molybdenum is one that features prominently, for some reason).

In contrast, I think most/all metallic permanent magnets are sintered. Alnico might've been; I think SmCo and NdFeB both are. Of course, eddy currents are less important to a permanent magnet!

Tim

The other end of....losses?

That stuff is actually quite great for certain purposes. You're almost certainly using some right now: the magnetic amplifier 3.3V regulator in ATX power supplies is the most common use, as well as high impedance common mode chokes.

Completely amorphous material is less common; apparently, losses are slightly lower, and properties slightly improved (hysteresis, Bmax? not sure), by annealing it just enough to make it nanocrystals floating in a mostly amorphous matrix. It's still just as brittle, so either way, it can only be shaped into stripwound toroids and bonded into cut cores (C and E shapes).

Because permeability is so high, while losses are lower than those of other hi-mu materials (like good old fashioned permalloy and supermalloy), the inductivity is astronomical, making great common mode chokes. A typical comparison of common mode chokes shows something like an octave more bandwidth (lower cutoff frequency, higher impedance) than the highest permeability ferrites (which top out at 10-20k mu). Interestingly, the impedance curve shows a diffusion characteristic over a certain range (presumably, a result of skin effect in the strip layers).

It's also great for very wideband pulse transformers -- the small signal properties extend very low, and with relatively high Bmax (0.8-1.2T usually), the signal level or flux capacity is also better than a ferrite core of the same size.

For power conversion, you'd consider this material for compact transformers in the 10-50kHz range: the frequency is too high for laminated steel (even 1 mil GOSS strip), and too low for ferrite (it would be massive). Trying to drive >0.8T at 50kHz is kind of a stretch, but may be acceptable with, shall we say, active cooling (water jacket, circulating oil?). Obviously, you need to consider the cost - size - performance tradeoff. Not common, but good to know for when you need it.

Amorphous/nanocrystalline materials are *not* good for energy storage, because, with such high mu, of course, it doesn't store energy... so you need an air gap, but air gap means fringing fields. Which means field lines penetrating all those finely crafted layers of metal, which means ridiculously bad core losses (not so bad electrically, but the localized heating cooks the material, so it's not good to use this way).

I believe there are powder cores using this material (either ground up, or sprayed in such a way that it forms nanoscopic particles), which do achieve quite reasonable characteristics at permeabilities useful for energy storage.

Tim

Metglas? Radical stuff. I know a guy who hydrogen anneals it and gets the permeability up to about 1e6.

It's not that expensive nowadays. I've played with it some (actually have a spool of the flat material too, off ebay) but we haven't used any in production. Metglas permeability drops with frequency, so ferrites are still better for fast stuff.

Wow; quite a treatise; a LOT of info I did not know. Tell Google Groups to paste that on their chalkboard.

Who is that guy? Julien Bergoz?

Jeroen Belleman

At last...approx due to difficulty in tweaking frequency dial of HP3312. T1 pri 90.97KC DCr=4.06 ohms, sec 91.25KC DCr=4.09 ohms, series aiding 28.88KC DCr=8.18 ohms (bifilar wound). T2 pri 85KC DCr=2.93 ohms, sec 86KC CDr=3.12 ohms, series aiding 89KC DCr=6.07 ohms ("split" winding; resonance rather broad).

So, as I recall the AADE will run at about 40kHz when measuring 88mH. So your not at resonance, so I would think that is not the reason the AADE is have difficulty. Although the on T1 series aiding you are close to measuring at self resonance. What happen on T2 series aiding seem wrong. Mikek

Tim - I thought that fringing could be controlled by enclosing the gap inside the coil. I used to specify some E-I laminated core inductors with a center leg gap to minimize the fringing. At least the inductance came out as designed. Was I fooling myself, perhaps?

John

Robert - If the inductance really is 88mH, then it would require somewhere in the area of 300pF of distributed capacitance to be resonant. I'm not sure, but I think that is a bit unreasonable.

Sorry to add more tests to what has been suggested, but I usually measure resonant frequency, record it, then add some capacitance in parallel and check the resonant frequency again. Maybe try 10pF to start. You can then increase the capacitance if you don't see a change in resonant frequency. Don't make a major change in resonant frequency, just enough to facilitate your measurements. Then add another capacitor of equal value and measure the resonant frequency again. By using the change in resonant frequency along with the delta C, you should be able to calculate the inductance and stray capacitance of the coil.

It does appear that your inductor of high Q, so watch for sharp dips.

Cheers, John

You can then calculate the

That actually makes things worse, but I'm guessing you were using it at a low enough frequency that losses weren't a big deal?

In an infinitely tall stack of laminations, fringing doesn't matter, because the fringing only occurs in the plane of the laminations. Which is the direction it flows in, anyway, so no biggie.

It's when fields are curving in such a way that there's a Z component, penetrating through the lams (or strips, or whatever layered structure it is). Which occurs at the ends of a stack of lams, or on the inner facing sides of a stripwound (C or E) core.

The additional reason an internal gap is worse is, now you have intense and divergent fields penetrating the turns nearest the gap, so the eddy current (proximity effect, whatever) goes way up locally. The overall effect isn't dramatic, but you can cook wires doing that at higher frequencies! And, for the same reason, practically any shape of laminated or stripwound core, with non-distributed gap, will experience significant localized heating around those gap areas. Which won't be much at line frequencies, but you'll have a hard time in the 10s of kHz and up.

There's actually an optimal distribution of windings for minimized proximity and fringing effects, that can be calculated for a given geometry. That turned up some months ago here, I think?

Internal gaps are mainly preferred because it reduces the external field. Good for EMC and RF work. Internally gapped pot cores are pretty amazing at self-shielding and adjustability (insert Sloman's oscillator here ;-) ).

Tim

As I recall, I never went above 300Hz for the 500kVA inverters. And, yes, I had to cool the sheet copper winding with winding ducts and fans. Nomex insulation was used, It actually was a nightmare for me because I was thrown into it and learning how to do it back then. I had no opportunity to experiment. I just had to learn fast and do it for the item being shipped when the resident guru left.

I even had one customer that needed, I think, 600Hz and 75kva. Not an enjoyable design at that time. (Early 80's to early 90's)

That's good info, but I didn't have much choice since I was using sheet copper. I know that you didn't know how the coils were wound. Sorry.

:)

Years ago we used a gapped potcore with 4-1/2 turns at 660kHz. I think the material was 3C81 or maybe 3F3 with an A sub L of 160, if memory serves. The extra 1/2 turn caused a fringing problem that overheated the wire near the gap. 4 turns worked fine and 5 turns worked fine. Can you explain why the 1/2 turn caused more fringing and overheated the wire? Mikek

Depends exactly where the wire lands? Idunno.

You do have an imbalance, where half the core (the part of the side wall enclosed by the "half" turn, looping around through the rest of the circuit) is being utilized differently. Which should increase the field on the nearby side somewhat, but no more than proportionally (i.e., 10-20%).

Most of the excess flux distributes itself back into the center peg. Which means it sees the same air gap area and length. So there's really no such thing as a "4.5 turn" winding on a center gapped core.

The correct way to do half turns is to loop an inverse-shorting turn around the outer legs (or in this case, the sidewalls), enforcing flux sharing between the two sides.

I expect your "4.5" turn inductance was closer to the 5 turn figure than the

It'll also have more external field (the outer half of the "0.5" turn wraps around through the circuit), but that also wouldn't account for anomalous heating.

Other mundane explanations: you were testing at different levels or frequencies, and just don't remember exactly what all was done; the wire shifted position between winds; used different wire or core sets; etc.

Tim

I am suspecting that the cores are in partial saturation - maybe T2 more.

Hang a 1 uF film cap across the coils and measure the resonant frequency, then compute L.

The method we used to wind was to put a layer of 3M foam tape around the bobbin and then a layer of thin cardboard or electric motor insulation, (depends on time frame) Then wind 4 turns, both wires come out on the same side of the potcore, one from the top the other from the bottom. With 4-1/2 turns, the wires would exit on opposite sides, one from the top the other from the bottom. Just the center where the gap was got hot.

I don't recall, the only way we ever measured was to find resonance with a capacitor.

This inductor was used to to tune out the capacitance of a piezo disc the we ran at about 660kHz. The delivered power was supposed to be 250 watts. I don't know that I ever new the voltage across the inductor.

Ya, four turns was pretty easy to get evenly spaced. But I get the point. Mikek

I don't remember years ago very well anymore, but I had a similar problem in 1981.

It was either a ferrite EE core or a ferrite pot core. I could not get it to work with 1/2 additional or 1/2 less turns. I can't remember why it was important at the time, but I didn't know how to research the problem.

It takes an integral number of turns to get full balance and thus shielding effect. If one has 300 turns, a half-turn makes little difference. If one has four turns, it makes a big difference.

Joe Gwinn