inductance measurement

curves/ACFLUX.html?

The BH curve for a core does not retrace its steps exactly when you take field through a full swing. The core gets slightly magnetized in the process. A small AC waveform on top of the DC measures the inductance for taking the field up and down a small amount on top of the DC part. This will be a different value than you get from the slope while just increasing the current monotonically with time.

curves/ACFLUX.html?

I'm far from an expert on inductors (most of the ones I deal with are air core so permeability vs B field is not my area.) But when I need to know the inductance of some coil I make a resonant circuit with a know capacitance and then find the resonant frequency. I do this at a few different cap values because there tends to be some variation. (I don't care too much about accuracy. 10% - 20% is usually good enough for me.) I don't see why you couldn't do the same thing as a function of the DC current.

George H.

s/curves/ACFLUX.html?

=94

The peak (AC+DC) current may be 60-90A....

That's what I do. I'm not quite looking at the B-H curve, more like assuming B (if EMF is constant..) and viewing the H curve (amps).

This is due to the magnetic domains being "sticky" at low H. Iron has lower permeability at zero than at medium levels. It peaks in the 10-40k mu_r range, then drops back off on approach to saturation.

Driving an iron cored inductor with a square wave voltage, you'll see a triangular current that's twisted, steep at the beginning, flattening out, then rising again.

In principle, you should also see parallel resistance as an initial step before the slope. Parasitic capacitance has the same effect in the test circuit, so it may be hard to see. Those lossy yellow powdered iron cores (#32??) always get warm just from testing, but the resistance isn't quite low enough to determine in this manner. Maybe a solid iron bar would work.

Your measurement is "total". You can find instantaneous inductance from the current slope. If you use a bipolar signal (instead of pulsing the inductor from one side and waiting for it to relax before pulsing again), you can watch current in both directions. The difference (hysteresis loop) is loss.

I like that this method gives saturation in amp-turns directly. Why no datasheet specifies this is beyond me. It's a damned practical unit to use.

Tim

"hump":

It's often easier to put two identical inductors in series, connect to a power supply, squirt in some amps, and measure the small-signal inductance at midpoint. That avoids having to build a current source.

John

They turn brown as the paint burns.

John

When I was a kid I had a red two-incher explode. Since it was outside, loud and caused a 120ft length of wire antenna to smack onto the ground the neighborhood took notice. "Did you hear that? Sounded like an accident somewhere." ... "Ahm, nah, it was nothing."

curves/ACFLUX.html?

Hello Micheal,

When I understand you well, you did a transient measurement and determined inductance from the voltage/current relation during the transient. This ignores the effects of hysteresis on AC inductance.

The permeability versus DC magnetization is measured under low AC excitation so you travel around a small BH curve offset from the origin.

The typical curves for permeability versus AC magnetization are only valid under zero DC magnetization. For the physical background, please check a physics book.

I sometimes face similar problem. When I have to measure inductance versus DC current, I take two coils. Put them in series en feed them from a DC supply (with good capacitive decoupling. I inject a small AC signal at the point where the two coils are connected (via a decoupling capacitor). When I know the EMF and source impedance, I can calculate the inductance based on the voltage measurement (oscilloscope).

Note that from an AC viewpoint, the inductance form a parallel circuit (so actual inductance is twice as high).

I make the high current out of a 30 kHz 200W oscillator with high current schottky rectifiers. Maybe you can make the high current by rewinding the secondary on a mains transformer and also using two high current schottky rectifiers (and filtering).

When you use very large capacitors in your setup, you may be able to do the AC measurement on an (digital) storage oscilloscope (as long as the DC dI/dt is far below the AC dI/dt).

Best regards,

Wim PA3DJS

Those Micrometals things must be made from rusty cat-food cans. The newish "cool-mu" type mixes are a lot better but not insanely expensive like permalloy powder.

John

lower

I've used Micrometals sometimes. They were ok, and didn't smell :-)

The one I exploded was from Amidon but it was my fault. I should have bought the biggest size of 2-1/2" or stacked two of the 2" but that was so friggin' pricey.

..

hat

has lower

mu_r

out,

tep

test

n cores

Better in what?

The Micrometals core is available through distribution which is important considering our potential volume of less than 100/month. As to the measuring inductance, 3 strands of AWG20 have 10mOhm resistance, which is ~40W dissipation when running DC current I need. It is OK for the application (3% duty cycle) but seems to be too much if I need to make measurements. And I still do not want to deal with cludging-up 50-80A DC current source ... As to the method described in previous posting (thanks Wim).... The measurement may take a lot of time, satisfy my curiosity and give me same result I can find in datasheets. Not something I get paid for, BTW. Will it predict inductance in the circuit?

Often seen in toroids in some "Conserv-Energy branded" FEIT compact fluorescent bulb circuits after a little use.

Jon

A dead CF bulb conserves plenty of energy. :-)

lower

With serious AC excitation, the powdered-iron cores can smoke. I'd suggest you test one at actual operating levels.

John

I didn't say they were necessarily dead at the time. Some are still limping along with a nicely over-toasted toroid. But yeah, a lot of them are dead when I rip the circuits out of them. FEIT actually makes some better units under their own name as the branding. But the Conserv-Energy ones sold at Costco are a disaster waiting to happen. I won't buy those, anymore.

Jon

I think the ugly yellow ones are good for only 5 or 10% worth of current ripple, and not much volts/turn. Always made me wonder what you're supposed to use for a lot of volts and ripple (like up to 40%)... gapped ferrite? Is Kool-mu good enough for that?

Tim

I once invented the 3-phase switcher (which it's very likely somebody else invented first) and built a 120 amp CAMAC power supply, with three buck mosfet regulators running overlapped at 40 KHz and summing together. Two things I learned: the yellow Micrometals cores fry at decent ripple currents, and power schottky diode *can* have reverse recovery glitches... hundreds of amps worth.

I replaced the cores with CoolMu's and everything cooled off.

All those powdered-stuff toroids are essentially gapped by the filler between the metal flakes. The lower the Al, the bigger the effective gap. Micrometals even sells one with permeability = 1, no metal at all. Its saturation curve is close to ideal.

The CAMAC crates show up on ebay now and then.

John

Just in case someone now goes looking for the stuff, Spang (the mfg) calls it "Kool-Mu". Just like Kool-Aide and Kampground :-)

Hey, anyone know a place that makes super thin flex circuit? Like 2-3um per layer or 0.0003 (yes, that many zeroes ...)? Maybe I should open another thread for that.

Hello Michael,

m...

ike

that

n has lower

ee a

ng out,

step

he test

ron cores

When you combine the 2 inductor method with your transient DC discharge method and a DSO, it will not take that much time. Off course, when you have confidence in your calculations, you may skip the measurements...

As you know ferrite materials are non-linear materials, when your measuring conditions are close to the circuit conditions, your measurements will predict inductance. In case of large excitation (where currents contain harmonics given an sinusoidal driving voltage), the inductance depends on how you define it (peak current values, RMS, first harmonic only, etc).

Best regards,

Wim PA3DJS