That may sound like a strange question.

Jim Williams at Linear Technology published some very nice current tracings for the inductor in a switch-mode power supply (see: AN-35), which show the beginning of a rapid increase in current at the end of the on-cycle, as the core nears saturation. I made an inductive probe to study that property. Of course, one sees the rapid increase in current when the "switch" to the inductor turns on, but as the change in current tapers off, the probe signal falls almost to zero, as expected.

I want to see whether one can detect the increase in current with saturation, when that is superimposed on the decrease mentioned above.

Tektronix has a solution to that (See: various Tek patents), which uses a combination of a Hall sensor and coil. I am experimenting with doing it electronically (mathematically).

Attached are some tracings of inductor current in a mock switching regulator. The top tracing is current to the inductor measured across a sense resistor. The bottom tracing is my probe including a differentiation component.

Now, I want to test it with a known, saturated condition. For practical reasons, I am limited to about 2A at 6V using my regular bench supply. However, I am willing to cobble up something else, if I thought it had a reasonable chance of success. One thought was to put a second winding on the inductor and use that winding to add a straight DC component to the magnetic field. Any suggestions for size and the magnitude of currents that will be necessary or alternative ways to get a known saturated condition?

Thanks. John

 

Not off the top of my head as to magnitudes, but the area is pretty well researched. One popular power control methodology in the 1950's and 60's was the magnetic amplifier. It used a control transformer called a saturable reactor. The flux coupling primary to secondary could be controlled very accurately by a small DC current, first from a thryatron tube, and later from an SCR.

Her is a link to the Wikipedia article - http://en.wikipedia.org/wiki/Magnetic_amplifier.

 

Those current curves look mighty resistive. Your resistor may be controlling the current more than the inductor. The resistor needs to be very small.

You should see a linear increase in current if you are primarily inductive. When it becomes more resistive, the curve slopes like what you have. In saturation they bend the other way, like in AN-35,Figure B6

 

Take a look at Ron Dekker's "Flyback for Dummies" page.
http://www.dos4ever.com/flyback/flyback.html

It's really not for dummies - he just has a sense of humor
Have a look at his "Inductor test bench", roughly halfway down the page, and the scope plot just below showing a toroid in saturation. Note the sharp increase in current at saturation.

The sharp increase in current implies a sharp change in voltage. You might be able to detect that across a small resistance in the ground path using a simple RC differentiator circuit and an opamp or comparator. But once you've entered saturation, you're already in an inefficient area of operation. Perhaps have the circuit detect saturation, and self-adjust to keep maximum current through the toroidal windings just prior to saturation.

 

Those current curves look mighty resistive. Your resistor may be controlling the current more than the inductor. The resistor needs to be very small.

The resistor is relatively huge (2 ohm) and was put there to control the current. But, it was designed as a mock circuit (just a current-limiting resistor, inductor, switched mosfet, decoupling capacitors) that would stay within the limits of my bench supply and stay reasonably cool as I messed with the probe. My circuit is like the "inductor test bench" in the Dekker article referenced by SgtWookie, except for the resistor. I needed some current limiting as the voltages are 0-6 and the inductor is 100 uH. My good bench supply is limited to 2A or so.

You should see a linear increase in current if you are primarily inductive. When it becomes more resistive, the curve slopes like what you have. In saturation they bend the other way, like in AN-35,Figure B6

I almost posted B6 as the example originally. Here is is for discussion. I suspect we are saying the same thing, but using different words. In Williams' examples, the current reaches a constant state or increases linearly over a few uS. In saturation, the rate of change of the current increase is much greater, the trace slope increases sharply, and is still positive (i.e., it spikes). Williams uses a Hall-stabilized probe. I have also attached an image from his paper showing the difference between the Hall-stabilized probe and a simple current transformer, as I am using. When I don't mess with the transformer output, I do get tracings virtually identical to his (See: lower tracing of DSCN0691, attached; used a AD620 IN-AMP, so I lowered the frequency for this illustration).

My goal is two-fold:

1) Find out what saturation looks like with the simple transformer probe. That is, is the current spike enough to be obvious on the declining slope of the curve. Unfortunately, Williams doesn't show saturation using only the transformer probe.

2) Assuming the answer to #1 is that a small amount of saturation may be hard to detect under the current limited conditions on my bench, can I manipulate the data to effectively transform the non-saturated curve to a straight line so that saturation would be more easily detected as an upward curvature.


SgtWookie, That reference by Dekker is one of the best descriptions I have seen. My previous resources were from Magnetics and Arnold, particularly the design material from Donald Pauly at Arnold published in PCIM, 1996 (sorry, I didn't record the link; it's on the Arnold site).

All of your comments have been extremely helpful. When I re-ignited my interest in electronics, I started with Horowitz and Hill, built a constant current source and looked at the V/t curve for a capacitor. That was very enlightening, as one usually sees RC curves and I wanted to prove to myself just how straight a line I could get. From a practical standpoint, it is easier to study a capacitor with a few hundred K resistor and a 12V supply than it is to study a 100uH inductor with very low DC resistance, as current and heat become limiting pretty quickly.

Unfortunately, H&H spends 7 pages on capacitors and 3/4 of a page on inductors. It would not be an understatement to say I understand capacitors 10X and much as inductors, that is, in about the same proportion as discussed in H&H.

Again, thanks for the help. Any additional suggestions or comments are welcome.

I still haven't gotten what I am sure is a saturated core to study. Any ideas for doing that under relatively "stable" conditions? That is, I don't want to short across a 40A, 5V supply with a few coils of 16 AWG wire.

Best regards, John

 

It seems to me you just need to adjust (increase) the load, or current through the inductor, until it saturates. It is important with inductors and transformers to over-rate them for the worst-case load current vs. temperature rise spec.

If an inductor/xfmr goes into stauration, then it is effectively no longer there. It isn't providing its intended benefits, and circuit performance becomes questionable and possibly destructive.

 

I got it. Here is the datasheet for what I want to do. Thanks again for the great information and leads.

John

http://www.coilcraft.com/pdfs/pwrmag.pdf

 

John,

I assume you mean the test circuit. Just a note that those images are not exactly right. The text warns you. On each half of the peak, you will see the voltage ramping, then it will curve up to the saturation peak. Then, it will curve downward as the resistance becomes the limiting factor.
Which happens first depends on the values of the resistance and inductance and don't forget the differential amplifier's output resistance.

 

That makes sense. BTW, the mag-amp (G6422-A) is ordered, and I should get it by next Tuesday to study. John

 

The calibrated test coil and some other goodies arrived this week, so I thought I would post an update on my inductive current sensor for an oscilloscope. I revised my test bench to more closely duplicate the Dekker reference provided by SgtWookie, as that eliminated the need for the signal generator. The signal generator was great for testing various frequencies, but had limited ability to get low duty cycles. The Schmitt inverter allowed much better control of that aspect at the cost of easily changing the frequency.

The oscillator was based on the 74AC14 and used a single inverter section. The other inverter sections were paralleled to drive a IRLZ24 mosfet. To keep to the 50mA limit, I used a 51 ohm gate resistor (didn't happen to have a 100 ohm). A 0.05-ohm Ohmite CS3 non-inductive sense resistor was used for standardized/reference current monitoring. A ferrite bead with 27 turns of 30 AWG wire for wrapping was used for an inductive pick-up. The coil was center tapped and the ferrite bead had a slot cut in it so it could be slipped over an in-circuit wire (i.e., it was C shaped). The inductor signal was amplified using an
LT1102 operating at a gain of 10.

The test inductor was a Coilcraft MagAmp G6422-A with a specified volt-time of 133 v-uS.

I plan to post a project about this and mentioned it to Dave in the past. I need to do calibrations, clean up the board, and some other things before I am satisfied with it as a project.

Today, I am just very happy to have duplicated the published results and to see my little inductive pick-up working.

Here are some screen shots:

DSCN697: Current curve with resistor and excitation pulse (top trace). Frequency was 1.75 KHz, duty cycle 4.5% (25 uS pulse width). The current curve shows the test coil is well into saturation.

DSCN696: Same as above, but much less into saturation.

DSCN693: Comparison of inductive pick-up (top trace) with current sense resistor (bottom trace). Coil is well into saturation for this illustration. The inductive trace duplicated the current sense trace well over a wide range.

As you can see, accurate measurement of the volt-time was made impossible by the relatively high noise level. My estimated volt-time was 94 v-uS at a current of about 0.3 A based on the resistor drop. At least it's in the ball park.

Again, thank you all for helping me get this far. John

Edit: The DSCN numbers didn't show. L-R the numbers are 697, 696, and 693

 

Looks quite promising, John

Now just make a simple RC differentiator circuit that'll sense the fast rise time and trip a comparator to turn off the current just prior to saturation.