Details of a inductor core saturation measurement circuit.

[David_]

Hello.

I am designing a inductor saturation test device, which looks like this:


but to simplyfie here is the basic circuit:


I have some question marks that I have failed to find answers to my self and I can't even order parts before some of them are answered, anyway here they are:

1, I have previously in hast soldered together a device such as this before(using the same 12V@16,5A as this) only without the mcu and using my function generator to control the MOSFET gate but I consider that a failure since the current couldn't saturate a T106 size toroid made out of material 26(yellow/with iron powder core used in ATX supplies), and I want to use the 12V @ 16,5A supply this time around as well and as I understand it I can increase the peak current by increasing the bulk capacitor and maybe increasing the delay between pulses to the MOSFET gate.
Is that correct?

2, I have tried my best to grasp the content found online about capacitor equations but I don't get if it is possible to calculate how high a current I can draw from a fully charged capacitor of x µF(at 12V) which is then discharged during a activation of the MOSFET in the above circuit that lasts for n µSeconds?

3, if the voltage to be switched is 12V how high Drain to Source Voltage(V_ds) do I need and does the fact that the MOSFET is going to switch a heavily inductive load influence the V_ds needed?

4, my application does not utilize a continues current so that parameter from a MOSFET datasheet is unrelevant, but the pulse peak current is defined over a time duration that isn't even close to that of which I will be sing and it also is defined for 1% duty cycle.
How do I relate a peak current in my design to the rated current of a MOSFET?

I am very uncertain about how high a current I should aim for and how to design the circuit to be able to deliver that, I will probably not need the peak current often but I need to be able to saturate cores with high currents at rare occasions.
It would be a big inconvenience for me if I can't use the 12V @ 16,5A supply

About the current peak, is occurs to me that it should be limited by the inductor series resistance... But I guess that the power supply have some fancy current limiting mode, but a big cap should allow for higher pulsed current peak...

Regards

- - - Updated - - -

I should say since I do from time to time attend this forum frequently that I am not as lucid as I use to be(not that I was particularly lucid before), I am in the process of finding a ADD medication that works and my ability to think and grasp concepts is 100% dependent on my medication. And to not bore you with this subject I can simply say that I am now and for a the coming month on a med that isn't close to good enough. Which result in a direct effect on how confused and fussy my mind is so I apologise if I will write more confusing stuff than usual.

 

[erikl]

Just an answer to this question:

2, ... how high a current I can draw from a fully charged capacitor of x µF (at 12V) which is then discharged during a activation of the MOSFET in the above circuit that lasts for n µSeconds?

This depends on how much voltage drop ΔV at the capacitor C you can tolerate during the current I period Δt: I = C · ΔV/Δt

 

[c_mitra]

You are passing a square (or rectangular for that matter) pulse into an inductor. Because of the self-inductance, The pulse will be integrated into a triangular form. Remember that dV/dt is large but finite and di/dt will be constant. Near the end of the ramp, the core will approach saturation and the di/dt will almost collapse (i will become constant). You need to measure i as a function of time and when you see the break in the straight line, you know your core has saturated.

1. pulse height: this will decide the rate of current vs time graph;
2. pulse width: duration for which you will see the rise of the current;
3. self-inductance: will determine the initial slope.

The current through the inductor is exponential and we want to focus on the early part of the graph.

Capacitive discharge appears interesting but the voltage will not stay constant and it will have a tendency to ring.

It will be better to increase the voltage and see the waveform for several microseconds.

 

[srizbf]

what is the value of inductors range that you are proposing to test ?

what is the expected operation and saturation current range ?

 

[David_]

I see, I'll be thinking about what's been said.
In the mean time I am mostly going to measure inductors in the µH - mH range, for example I am currently trying to wind a 3-4mH inductor which I will want to test and see where the saturation occurs(that coil needs to sustain 10A without saturating). But as most coils I am using are for DC-DC converters and transformers in AC-DC converters so if possible say 1µH - 10mH or something like that would cover the hole range I think.

As for currents, I will search for data on my first prototype for this to see how high a current wasn't enough but just guessing I might say 50A()if that's realistic at all?)

Regards

 

[c_mitra]

I see, I'll be thinking about what's been said.
In the mean time I am mostly going to measure inductors in the µH - mH range, for example I am currently trying to wind a 3-4mH inductor which I will want to test and see where the saturation occurs(that coil needs to sustain 10A without saturating).

I see that you are trying to make coils that can support 10A on a regular basis. I expect that you wish to have some reasonable headroom and therefore the coil should be able to support 12-13A without saturation. Saturation must take place above 14-15A current. Your power supply can handle max 16A which is just sufficient. We need to have some R in series that will allow the LR time constant to be selected in the mS range so that the current rise can be clearly seen. Just like a multimeter, you should be able to switch in /out different values of L. when you have small values of L, you will need a higher value of R so that the time constant is in the mS range (unless you are using a scope). When you have higher values of R, you will need higher voltages to drive 10A current through the choke.

 

[David_]

I have ordered parts to try a prototype of this circuit, I also ordered a toroidal core somewhat at random based on the physical size and the permeability. That I did mostly to learn but I have a problem.
I forgot to include the gate drive IC in the order and I would prefer not tu use a hole IR2110 to drive a single low-side MOSFET switch, do anyone have any suggestion on a circuit that would be suitable for switching a low-side switch on/off as the one in the schematic in the top post?

Regards

 

[FvM]

To tell the truth, a half IR2110 is much better than a ZXGD3003 (complementary emitter follower), when driven with 3.3V logic. I presume, your MOSFET needs 10 or 12V Vgs for full turn-on, but it gets less than 3V in your design. A discrete driver solution would need ca. three discrete transistors.

 

[David_]

Ok, then I must have miss-read the datasheet since I thought I had made sure that my logic level signal would result in a VCC(of the driver chip) level output. Thanks for saying so, after having thought about it why should I make thing harder for me by not using a cheap IR2110. The old ones are quite cheap now, at least on ebay.

Would a chip like IR2110 drop in price when they released IRS2110?
I have not been involved in these things long enough to have been able to observe how the price market works.
I guess what I wonder is if the prices is fixed by the cost of the process that made them or if it is just as with everything else that is for sale in this world, product fall in price when newer versions of that product comes out?

I'll just use the low-side portion of a IR2110 and be done with it.
I am making this 12V for now but as soon as I have found a higher voltage supply I will change, I found a project online from a guy who appeared to know exactly what he was doing and had a long page about such a device as this and he showed that a low current supply of no more than 1A could be used to charge capacitors than in turn deliver somewhere around 20A if I recall correctly, in short burst of course.
I'll link to the page in a later post when I find it again.

Regards

 

[chuckey]

I have a problem with this circuit, perhaps some one could clarify my thoughts. If the pulse width is > 5 X L/R. The current rises in an exponential manner, ending up at 12V/Rtot (16A?). So monitoring the current will give an exponentially rising voltage. Now if at some current the inductor saturates, say 10A, The first part of the current monitoring wave form will be an exponential due to L/Rtot up to 10A. After this current the top bit of the waveform will be a much faster exponential of Lsat/Rtot. As this transition will not be sharp, I am not sure how you will be able to decode the waveform into any thing meaningful. i.e. the inductance has fallen by 1%?, 10%?, 100%?.
Providing C X Rtot >> pulse width it should be OK. I would include some RF decoupling caps to offset the self inductance of the electrolytic capacitors. Keep all leads short and thick.
Frank

 

[David_]

I'm sorry I missed the last post here.

I don't quite follow what C X Rtot >> pulse width is supposed to mean?

I have another problem, I have opted to use IR2110's low-side portion which is an IC I thought I had gotten my head around, but I don't understand this at all now.

Please direct you attention to the picture below:

(The mcu part is as you se simplified so the picture is more readable, I haven't added the high freq caps parallel to the big bulk cap jet as suggested but I will)

I have a hard time expressing what is wrong with it but something feels wrong...
Would you mind look it over and say what you think?

I would really appreciate it, but also imaging that we changed the 19,5V supply to a 32V supply.
Would that not be causing problems demanding a second regulator to solve?

Regards

 

[c_mitra]

Have you considered feeding the coil with a symmetrical triangular voltage? You will have less headache with measurements.

 

[dick_freebird]

I wonder if all of this is trying too hard. You could switch high side
voltage with almost anything (a contactor / relay, a FET, a 25A
light switch) because what you care about is at the other end
of the cycle, where the volt-seconds have run you up past the
saturation flux density. Seems to me all you want is a decent
reading in time of the current (sense resistor and ADC) along with
the applied voltage, looking for where the winding current starts
to become nonlinear with time. That, you could get after the
fact by processing the data stream. But just what constitutes
onset of saturation, seems to be arbitrary to some extent. You
will have a hard time figuring out even what one particular
core vendor means by saturation flux density. Let alone finding
any broad agreement on methods / criteria.

You might even find that your simplest and cheapest way
might be a car battery and a {whatever}A automotive
circuit breaker. Certainly won't run out of current, get
a nice (if slow) repetitive waveform, ensure that the
winding is protected (unlike the light switch proposition)
and so on. Two 'scope probes and dump the mess to a
.csv file (if it's a modern 'scope)?

 

[c_mitra]

You might even find that your simplest and cheapest way
might be a car battery and a {whatever}A automotive
circuit breaker. Certainly won't run out of current, get
a nice (if slow) repetitive waveform, ensure that the
winding is protected (unlike the light switch proposition)
and so on. Two 'scope probes and dump the mess to a
.csv file (if it's a modern 'scope)?

This is a simple and effective solution for simple inductors but there is no way to limit the current and if you connect directly, the inductor will appear as a short after a short time and melt the coil. We must have an electronic switch that can be turned off soon after the current reaches *dangerous* values. For many core materials (iron core, for example) the B-H curve is not a straight line and for hard ferrites, the saturation point, as you mention correctly, is somewhat subjective. The current waveform can be processed in software and lots of information can be obtained (we can get the full hysteresis curve) but that needs a lot of data points.

 

[David_]

Since I've been learning about cored coils I have taken Saturation to be a point determined by each application and nothing else. And in "each application" there are room for personal preference.
In my design, is it possible to observe enough information about the coil under test so that I could have a "Saturation depth" parameter that would influence the result based on if saturation depth is set to 10% or 50% or some other 1-100%?
% would ideally be related to drop in inductance, in worst case maybe I could have option in the user interface to enter the inductance of the coil in order to then be able to set a saturation depth...

As for the implementation, I have seen others do it this way and I understand too little of the requirements to know of any other technique, I know for sure that this works so I'll go with that and hopefully after I have completed this I will have gained enough knowledge that I then can tell how it could have been done in a simpler way.

c_mitra, could you elaborate more about using a triangular voltage?

Regards

- - - Updated - - -

Oh yeah I forgot.
I am about to build the prototype using a function generator and scope instead of the mcu, and I can now choose between 12V, 19,5V or I could generate 30Vdc(28Vac) or double that.

I have already put all needed parts on my desk so after this choose I'll solder it together.

What do you think I should go with?
Is higher better as in "the higher you can get the better it will be for you"?

 

[c_mitra]

c_mitra, could you elaborate more about using a triangular voltage?

When you apply a DC step pulse (0-Vmax), current in the inductor increases in a linear fashion. That means di/dt is const and the back emf will be constant and only slightly less than the applied step voltage. In an ideal world, the inductor has zero resistance and the current will increase to infinity. In real world, the inductor has finite resistance and that will limit the current. ALSO, the inductance will fall off at high current (unless it is an air-core inductor) and that will reduce the back emf produced and current will be more than linear...

When you apply a DC triangular voltage, the dv/dt is constant and the current in the inductor will increase in a non-linear fashion (i^2 as a function of t) and you will be able to get higher current than a step voltage (for the same peak voltage). When the core saturates, the back emf will be reduced and there will be a break in the same region (current will now increase faster). That means using the same Vmax you will be able to see the saturation because the current will be higher...

 

[FvM]

When you apply a DC triangular voltage, the dv/dt is constant and the current in the inductor will increase in a non-linear fashion (i^2 as a function of t) and you will be able to get higher current than a step voltage (for the same peak voltage).

I don't what your presumptions are, but this is surely not generally true. There are two possible scenarios:

- power supply with continuous current capability. Maximum current is limited by indictor series resistance, and of course power supply maximum current.

- power supply with limited energy capability, e.g. the capacitor boosted circuit assumed in this thread. You get highest inductor current if the power supply is simply switched to the inductor = no energy lost in the current source transistor. Ramping the inductor voltage will reduce the maximum current in this case.

 

[c_mitra]

- power supply with continuous current capability. Maximum current is limited by indictor series resistance, and of course power supply maximum current.

Most inductors have small series resistances and series resistances like 0.1R or 0.01R are not uncommon. Particularly if you are looking for a high frequency application. Automotive batteries, widely used as source of power in many inverters, have still lower internal resistance. #13 suggests an automotive battery as a source of power. For most applications, we ignore these two factors. Maximum current can become 100s of A and the inductor may melt even.

- power supply with limited energy capability, e.g. the capacitor boosted circuit assumed in this thread. You get highest inductor current if the power supply is simply switched to the inductor = no energy lost in the current source transistor. Ramping the inductor voltage will reduce the maximum current in this case.

I believe the exact opposite. If the power supply is simply switched (square wave) on (or off), dV/dt is very large (rising edge) and current is practically zero. Once dV/dt becomes zero (flat-top), current increases linearly to infinity. (this will certainly be true for an air core inductor that has nothing to get saturated).

 

[FvM]

I believe the exact opposite. If the power supply is simply switched (square wave) on (or off), dV/dt is very large (rising edge) and current is practically zero. Once dV/dt becomes zero (flat-top), current increases linearly to infinity. (this will certainly be true for an air core inductor that has nothing to get saturated).

Can be simplified to dI/dt = V/L, which is obviously true. I was stumbling however upon your previous claim that you would achieve higher inductor current with triangular current waveform. That's particular not right in the context of this thread about capacitor boosted power supply.

 

[David_]

Would a battery be a possibly better power supply for this kind of device?

I hadn't considered that a battery can deliver very high current shortly/low current over long time, so a battery/battery charger circuit could maybe give me access to larger currents so that I can saturate larger cores/inductors.