Inductor core saturation: DC vs peak current value

Question

I am trying to figure out something. I am currently working in SMPS design and as I get deeper different questions arise.

Magnetics is not one of my strengths and this is why I find pretty difficult to answer the following question: What is the current value that will make a coil's core saturate?

All datasheets that I saw give us Isat as a DC bias current that drops L by 10,20,30%, but this is DC bias. Does it mean that my inductor can withstand larger peak current values than Isat? If so, how do I calculate it?

All application notes that I read use WC value of IL+deltaIL and compare it with Isat (which is a D.C. bias,) so it is confusing.

Answer 1

Saturation isn't a purely DC characteristic; what matters in core saturation is the instantaneous current (technically, the instantaneous H-field).

Whether your application can handle the inductor going briefly into saturation each cycle is another question entirely; I prefer to avoid saturation as much as possible, but that's really not feasible in many applications.

Note that what the inductor can withstand is entirely different from what makes it saturate. Saturation is not a destructive process, and as the current falls they will leave saturation. Saturation and hysteresis do, however, work together to create one of the major types of core losses, so constantly driving it into and out of saturation will cause the core to heat up. This should be accommodated for in your thermal design.

There is also no one current value above which the core is saturated. Specifying a saturation current is the same kind of convenient fiction as specifying a diode's forward voltage; any amount of current will cause a decrease in inductance, but the change in inductance is minuscule at sufficiently low currents, just like the current through a diode is minuscule at sufficiently low forward bias. Most inductors I've seen specify saturation current as the current when the inductor reaches either 90%, 80%, or 50% of its nominal value, referred to as 10%, 20%, and 50% saturation respectively. Make sure you check which one your inductor is specified at!

Here's an example of what I mean:

(source: a datasheet from Magnetics)

This is the saturation characteristics for a magnetic core that's intended to be wound by the designer, not wound at the factory, so it's specified in ampere·turns instead of amperes and \$A_L\$ instead of \$L\$, since they don't know how many turns of wire you're going to put on it. You can see that below about 50 A·t the inductance is roughly constant, but then it starts to fall off as current increases. This is an air-gapped core, which makes the fall-off more gradual than a solid core at the cost of having a lower \$A_L\$, but the same non-instant fall-off is there in any core.

Answer 2

does it mean that my inductor can withstand larger peak current values than Isat? If so, how do I calculate it?

There are two things your inductor can do wrong when a high current flows

a) Saturate
b) Overheat

Saturation results in a drop of inductance. It's reversible, and doesn't damage the inductor. It's not a problem to the inductor, or any other components around it, if we are applying a controlled current.

However, we're often not applying a current to the inductor, but a voltage. A voltage will cause the current to ramp up at a rate of V/L amps per second. Often a design will control the current by controlling the voltage, and the time for which it is applied. If the instantaneous current gets to Isat, L drops, and I will carry on rising, but now it will rise much, much faster. This is often a problem for the rest of the circuit.

Overheating can result if the rms current causes too much heat to be lost in the resistance of the inductor. Short peaks are generally not a problem.

Answer 3

Magnetization dynamics in ferromagnets are fundamentally not faster than the magnetic resonance frequencies in these materials, which typically is in the range of 1..10 GHz. Therefore, current peaks shorter than nanoseconds will not matter for sure.

Even beyond this timescale, the magnetic domains in granular core materials "creep" from one favorable site to the next (usually grain boundaries) getting stuck ("pinned") in between. This process slows real magnetization dynamics even further for such materials. As a result, I would estimate that any pulse shorter than some tens of nanoseconds will not saturate the core.