[SOLVED] INDUCTORS inductance change with current

[bowman1710]

Hi Guys,

Quick question, how can I work out the change of an inductors inductance with the amount of current flowing through it, is there an easy way to test this?

Cheers

 

[FvM]

Some professional LCR meters are providing a bias-T option to apply a DC current while measuring inductance.

A different approach is to apply a constant voltage to the inductor and derive L versus I from the observed dI/dt.

 

[schmitt trigger]

If you know the core's vendor and material, you can find curves which show the decrease of permeability with applied ampere-turns.

The other approach is to use LCR meters which can apply bias. Usually those meters are expensive. I remember using back in the 1990 some Wayne-Kerr units, which could be customized for different current ranges.

 

[Warpspeed]

is there an easy way to test this?

Possibly the simplest way to do it is to build a half bridge diagonal flyback circuit, using two mosfets, and two fast diodes and a very large electrolytic energy storage capacitor. The electrolytic should ideally be one with a suitably high ripple current rating.

By manually controlling the incoming dc supply voltage, and the switching frequency, you can then monitor the ramping up and ramping down of the current in the unknown inductor.

Each switching cycle the current flows first out of the capacitor, and then back in again during the flyback phase. The incoming dc supply only needs to supply circuit losses, which should be only a small fraction of the total circulating current.

So you can ramp up to some very high peak currents readily enough, and watch as the core goes into saturation.
By knowing the applied dc voltage, and the rate the current ramps (amps per microsecond), inductance is very straightforward to calculate.
And you can plot inductance versus current that way.

It is possible to monitor the onset of saturation up to some tens of amps with only a very few external incoming amps being needed to keep the whole the process going.

 

[zorro]

A quite simple method:

Using a frequency variable sinusoidal generator (with a series resistor such that R>>wL and a decoupling C), and an oscilloscope or an AC voltmeter, measure the resonant frequency of the inductor with a fixed capacitor (frequency that gives maximum voltage in the parallel LC).
With current flowing across the inductor (using a DC current source or a voltage source with a series resistor such that R>>wL) measure the variation in the resonant frequency.
With the two frequencies and C you can calculate L and its change.

Regards

Z

 

[andre_luis]

The biggest shift on inductance would occur more accentuated beyond the rated magnetic flow of the core, not only making it a reactive component but also dissipative, so that it would be also expected a 'resistive' component on its equivalent model on increase of current flow.

 

[bowman1710]

Thanks guys for all your help, its much appreciated. Didn't expect so much help over a question like this. "FVM, schmitt trigger" I will look at my LCR meter to see if it has the bias capability. "warspeed, zorro", I will have a look at both of these and see what outcome I get.

 

[Anna Conda]

Put two chokes in series and put DC thru them, put a large cap across the ends, run a sig gen via a 1k res to the mid point and the gnd to one of the ends, the chokes are in // for ac, choose a freq, say 10kHz, increase the DC and use a scope to measure the voltage across the 1k, and compare to the AC voltage across the chokes, when equal, 2.pi.f.L = 1k ohm, taking the DC up will eventually reduce the L and you can calc it depending on the volt ratio, easy and effective, don't forget you have 2 chokes in // for the 10kHz AC.

 

[andre_luis]

Although the effective value of inductance can shift as a function of the current across the inductor, I´m in doubt if we can assume that DC biasing current would affect on this AC analysis.

 

[Warpspeed]

How much dc are we talking about here ?
Big difference between 30mA and 30 amps.

 

[zorro]

Although the effective value of inductance can shift as a function of the current across the inductor, I´m in doubt if we can assume that DC biasing current would affect on this AC analysis.

Yes, it does!
AC behaviour depends on the slope of the H-B curve around the point determined by DC biasing.

How much dc are we talking about here ?
Big difference between 30mA and 30 amps.

Not really. It's not matter of current: 30 mA flowing through 1000 turns has the same effect that 30 A through 1 turn.
What is important is the magnetic field intensity H (measured in A/m or A*turn/m) and the material (i.e. its B vs. H properties).

Z

 

[Warpspeed]

Not really. It's not matter of current: 30 mA flowing through 1000 turns has the same effect that 30 A through 1 turn.
What is important is the magnetic field intensity H (measured in A/m or A*turn/m) and the material (i.e. its B vs. H properties).

That is all perfectly true, but the test methods will be very different.

With 30mA and lots of turns, you may be able to use a signal generator, a resonating capacitor, some clip leads, and feed it from a high impedance current source, with a reasonable chance of being able to measure something useful.

If it really does need 30 Amps dc, or more, then driving it straight from a bench type signal generator, may not do very much. Unless you have a pretty big power amplifier to modulate that 30 Amps.

Theory is one thing, but extremes of current and power, and very low impedances can sometimes make things much more difficult to measure in practice.

For example, I have here some fairly high powered inductors, used to filter the PWM of a sinewave inverter. These normally work at 15 amps rms 50 Hz and go into soft saturation above around twenty five amps. Just not possible to measure any of that unless you can actually run it up to those kinds of currents and watch it fall over.

 

[Anna Conda]

The method in post #8 uses the chokes to isolate the DC supply from the measurement, this way the small or large signal inductance can be seen at any DC point on the choke, we have used this method for many years for all manner of chokes up to 50A and beyond, easy and quick to get a graph of real L vs thru current this way....
Also note that some iron powder cores show a slight rise of L as the current is increased a little, followed by the usual inherent drop off...

 

[FvM]

Using a pair of identical inductors is a clever method to save an expensive high current bias T. I also believe that a standard signal generator which can e.g. deliver 10 V into a 50 ohms (= 200 mA output current) will give sufficient signal swing to measure small signal inductance in this setup. So you won't need a dedicated power amplifier.

A current source with respective output capability is of course required.

The core hysteresis loop has to considered, you'll get slightly different inductance values in a series with rising or falling DC current.

Alternative to the DC bias/small signal AC measurement, the large signal pulse method mentioned in post #2 and #3 can be used. It gives effective inductance and losses for a specific switching application.

 

[Anna Conda]

Using the two chokes, effectively gives a current source from a standard good quality power supply (most of which you can run under current limit anyway)....

 

[Warpspeed]

Its certainly an ingenious method.
But what if you are trying to measure a single "unknown" component, which is more often the case.

 

[Anna Conda]

Well then, any serious power electronics tech/engineer would use the resonance method or have a decent inductance meter to hand...

 

[Warpspeed]

Measuring straight inductance is easy enough, the problem of doing it with considerable dc bias flowing, is the question posed by the original poster.

Strictly speaking, inductors should only ever see ac, as in tuning or filtering applications.
If there is dc present, it then (by definition) becomes a choke, and a quite different testing method is called for.

 

[Anna Conda]

In that case a large inductor (> 10x the unknown) could be used as the second inductor, either as part of a resonance test to provide DC current with a very high impedance to the AC test signal, or as per the test in post #8

 

[Warpspeed]

Its really doing it the hard way, and its prone to errors.

A simple ac inductor is best measured with ac, either using a bridge type of circuit, or the resonance method.

Where there is ANY dc bias, its far better to measure the rate of dc current rise when a dc step voltage is applied.
The current then rises at a rate proportional to the value of inductance.
As the magnetic path begins to saturate, the rate of current rise increases.
From that you can easily plot inductance change with various levels of applied dc current.
Usually the dc resistance of the choke will be low enough to ignore, but if it is not, that can be measured with an ohm meter and the resulting internal reduction of applied dc voltage applied when calculating the inductance.

Its a very reliable method, and is a lot more realistic for testing chokes used in dc circuits anyway.

The method I suggested in post #4 even recycles all the stored inductive "joules" back to the original dc power source for high speed repetitive testing.