Details of a inductor core saturation measurement circuit.

Okey so the prototype is coming but I am having a time of more ADD trouble than usual so I can't even thus far finish a post before I am somewhere else doing... something.

But during assembly of a prototype I came to think about a couple of things that I can't answer my self.

1, I had at some point thought that I could vary the frequency of the switching of the current to observe effects of frequency upon the saturation of the core, but does frequency even do anything to the saturation in a manner similar to how frequency have an impact on inductance value?

2, I don't know where I am going with this as I can't grasp my thoughts far enough to even see a glimmer of the conclusion, but frequency is easy to think about if it is a sinusoid signal. But I am apparently having some difficulties thinking of frequency when it comes to square-waves or actually pulses because I think that my application will be using unusually small duty cycles and as such there will be a considerable dead-time.
If we are talking about frequency dependant characteristics of some physical object and the excitation signal is square, is it in any way relevant to the situation if the duty cycle is 0,005(0,5%) or 0,5(50%)?

3, Isn't this endeavour in some sense futile assuming that the coils that I will test is used for SMPS situations where the core is experiencing considerable DC-bias voltage?

4a, is there any way possible to adapt my design to include an optional DC-bias voltage?
So that the core saturates under realistic circumstances.

4b, or does such an idea demand a completely different design, I ask because I don't understand how AC and DC interacts well enough to have a clue about where to begin(other than by possibly volatile experiments where the magic smoke of components are released).

I am sure that a DC-bias does influence the saturation but is it changing the saturation point or is it simply "using up" the available magnetic regions so that a smaller AC current will saturate the core while the saturation is occurring at the exact same point as if there where no DC-bias.

Regards

1, I had at some point thought that I could vary the frequency of the switching of the current to observe effects of frequency upon the saturation of the core, but does frequency even do anything to the saturation in a manner similar to how frequency have an impact on inductance value?

Click to expand...

Why frequency will have any effect on the inductance value? (not at least for reasonable frequencies- Hz-KHz-MHz)

2, ... ...But I am apparently having some difficulties thinking of frequency when it comes to square-waves or actually pulses because I think that my application will be using unusually small duty cycles and as such there will be a considerable dead-time.
If we are talking about frequency dependant characteristics of some physical object and the excitation signal is square, is it in any way relevant to the situation if the duty cycle is 0,005(0,5%) or 0,5(50%)?

Click to expand...

A square wave can be approximated as a sine wave. The Fourier decomposition of square wave shows that it has lots of odd harmonics: https://mathworld.wolfram.com/SquareWave.html

You can always assume that you are exciting it with a sine wave PLUS another sine wave with three times the frequency PLUS another....

This is a robust description in the mathematical sense.

3, Isn't this endeavour in some sense futile assuming that the coils that I will test is used for SMPS situations where the core is experiencing considerable DC-bias voltage?

Click to expand...

DC and AC are equivalent because they can be superimposed. At any instantaneous moment, everything is DC! You simply shift your operating point to another (higher or lower) and you need not repeat your experiment will all possible DC bias.
The underlying theory lies with the B-H curve. If you know the B-H curve, you know at what magnetic field you material saturates and what current is needed to make such a magnetic field.

4a, is there any way possible to adapt my design to include an optional DC-bias voltage?
So that the core saturates under realistic circumstances.

Click to expand...

It is the current that is responsible for the magnetic field and bias current simply shifts the operating point up or down (or is it left or right)

4b, or does such an idea demand a completely different design, I ask because I don't understand how AC and DC interacts well enough to have a clue about where to begin(other than by possibly volatile experiments where the magic smoke of components are released).

Click to expand...

If you are planning to design an instrument for commercial application, it is perhaps wise to include such an option. I am not sure because I am not an engineer.

I am having some difficulties with the prototype, I am building it on a strip board and using a rather large heatsink with s tiny tiny fan(last time I did this my heatsink got uncomfortably hot, this time the RDS(ON) of the MOSFET is around 3,9mΩ as opposed to around 77mΩ as it where last time) but even so I thought it be fit to use a larger heatsink.

I am utilizing a 19,6V supply so lets call it 20V and I have 3 10,000µF capacitors that are rather large, I am making room for all three because I want to observe who it might differ from having 10mF and 20 or 30mF. Math is not my cup of tea and I had found a site that was about a device such as this that also went through the calculations for the capacitors but I can't no matter how I trie find it again!!!... However my math class starts next week so I will hopefully be able to make more sense out of simple equations in the future.

Even with the very simplest equations I have a really hard time deducing what the formula means in practise, sure I grasp that I = U/R means that the higher the R the lower the I and the higher the U the higher the I but that is as far as i get... sort if.

My concern with my prototype is about the distances between the circuit components, it looks like there will be somewhere around or up to 10cm between the IR2110 gate driver output and the MOSFET Gate pin and there are a number of cm in between the large capacitor banks and the inductor source connection and then some cm from the inductor return connection to the MOSFET Source pin and then come more cm from the MOSFET Drain back to the power supply GND.

A earlier post raised concerns with this sort of things and also recommended that I'll use high frequency capacitors across any large bulk capacitors.
Will ordinary ceramic SMD capacitors suffice for this?

Do anyone have anything to say or note about this situation?

Is there any particular high frequency ceramic multi-layer capacitor type or is that hole genre considered high frequency?

I have been looking for high frequency(or SMPS) aluminium electrolytic's but they are hard to find from suitable suppliers, previously I have placed two orders from digikey, that is something I won't be doing again though. The first time it appears as they missed to charge custom and boarder charges and when you add that to the cost as it came to be the second time it is no longer an option to use digikey when mouser is around.

I live in Europe.

There is a much better way to go about all of this.
Build yourself a conventional diagonal half bridge flyback circuit.



Both mosfets turn on simultaneously and the current ramps up, mostly being drawn from the very large reservoir capacitor C3.
When the mosfets turn off, all the energy stored in inductor L3 causes diodes D5 and D4 to conduct and RETURNS almost ALL of the ramp up energy back into C3.

The main dc power supply keeping C3 up at a constant voltage, only has to supply a small current to replace the energy lost from conduction losses in various components, which should be relatively low.

The same energy circulates around the circuit with minimal loss, and minimal make up requirement, from the dc power supply.
You might be able to ramp up to 50 or 80 amps peak, and do it with only a 3 amp dc supply.

First requirement is a typical adjustable dc bench power supply, say 0-30v at 3 amps.

Next requirement is a fixed fifty percent duty cycle square wave generator, adjustable over a very wide frequency range to drive both mosfets simultaneously.

A Hall effect current sensor to monitor the inductor current then goes to an oscilloscope, which can be externally triggered from the square wave generator.

By adjusting both the dc voltage and "on" time (with equal "off" time) the current will be seen to ramp up, then ramp down, then sit at zero for a very brief time before again ramping up.

By increasing both dc voltage and on time together, the current peak can be made to rise up as high as you dare go with the mosfets and shottky diodes you are using.

The dc power supply need only be small to do this.
Its dead easy to read straight of the oscilloscope amps per microsecond with so many dc volts applied, and easily see the trace bend upwards as saturation is approached. That will tell you both dynamic inductance and saturation limit.

Very easy to drive many things into severe magnetostriction and get groans and loud shrieks ! All without requiring a massive power source or generating fierce heat.

Capacitor C3 needs to be a really good quality low ESR type with a suitably high ripple current rating, The diodes and mosfets suitably rated for high peak current, and an appropriate range Hall sensor.
The heat sinks do not need to be large, especially as you can see and measure everything you need to see in a very short time.

Well that sounds very reasonable indeed and rather refined in comparison to my way(not that I thought of it nor did I come up with the basic design), the only concern I have with what you wrote Tony is that I have found it quite hard to find good quality low ESR type electrolytic's. But then again I am not entirely sure about that I should look for but some application notes have painted a picture of it existing the "ordinary" type of alu.electrolyte's and also some superior low ESR type or high frequency type of alu.electrolytes.

While spending hours searching through digikey I have only once or twice seen those explicitly different types of electrolytic's, but have I maybe got this wrong and all I should look for is a alu.electrolytic who's specification includes a low ESR and high ripple current no matter if the product is marked as such or not?

In any case I think I am set for such a design apart from if the bulk caps I got doesn't suffice, I have not any particular experience with the sort of circuit you show but the picture sure look simple and ad the power supply feeding the circuit is already isolated then it may be almost as simple as the picture suggests.

I will adapt my design to a version of what you proposed and we'll see where that takes me, I have recreantly researched hall effect current sensors and I know of just the one I need. Maybe I even should throw in two hall sensors... one low current and one high current so that I can if I would find the need to measure saturation in the range of 1-5A just as well as 10-50A or higher. Those numbers where just grabbed out of thin air but you get my point.
I am not sure but it sound as a hall sensor tailored for up to 100A would not be very useful for measuring 1A or such a low current, and I have recently begun fiddling with low current DC-DC converters for use with battery powered measurement equipment and I always want to find out and document as many parameters as I can about the components I use in designs.

I find your design much more attractive, thanks.

I just had a look on e-bay and there is nothing there right now really suitable.
But check out the "Evox Rifa" low esr capacitors on e-bay from time to time.

These all come in a white plastic insulated can.

https://www.ebay.com/sch/i.html?_od...XRIFA+low+esr.TRS0&_nkw=RIFA+low+esr&_sacat=0

Just to give an example of what goddies you might find in the Evox Rifa range.

47,000uF 40v, continuous ripple current at 10Khz, 50.7 Amps, esr 8 milliohms 16nH inductance. Size 65mm dia, 105mm length, screw terminals.

The awesome specifications come at a price though.

These things are horribly expensive new, but come up on e-bay both new and second hand from time to time.
Its just a case of being lucky and grabbing something when it is there.

Not convinced you need fabulous ESR/ESL. If your ramp
starts at zero (volts, and thus current) any ESR drop is
also zero (at least, to start with), while ESL is a ratiometric
term during ramp that you can de-embed by simply logging
applied inductor voltage (rather than assuming a supply value).
You should at least not see much of a turn-on divot, and the
supply ramping down while current ramps up only makes a
tiny bit more math.

A good way, and probably no more expensive, is to use
many parallel smaller caps to drive down (by paralleling)
the parasitic series terms. But again, you could simply
in SPICE check all of this out, add your own rough ESR/ESL
elements and see what if any impact there is, and where
you ought to put your measurement points.

If you are willing to collect and post-process a data stream
then the electronics can be very simple and non-critical.

Its not so much the esr, as the ripple current rating.

Trying this with a cheap ordinary aluminium electrolytic will probably just open circuit the foil or boil the electrolyte.

We are not talking about a couple of amps rms, but many tens of amps rms.

If you have just fabricated some large wonder inductor and wish to set the air gap so it can safely run up to 30 amps without saturating, you may need to run it up to 40 amps during testing.

There is no substitute for 40 real peak amps.

The best thing about simulations is the capacitors never explode in your face.

Warpspeed said:

The awesome specifications come at a price though.

These things are horribly expensive new, but come up on e-bay both new and second hand from time to time.
Its just a case of being lucky and grabbing something when it is there.

Click to expand...

I've never heard of them before but I have been on the look out for such caps, I have seen short videos of them being in use for just such applications as this. On youtube there is a guy whom have built a design where he charges such a large white cap and then discharges it with a triac or something similar(through a coil).
I will for sure be keeping an eye on ebay for them, thanks.

dick_freebird said:

only makes a
tiny bit more math.

Click to expand...

I'm very interested in the topic of enhancing all measurement scenarios with more math, I'm particularly bad at math but tomorrow is the introduction assembly for a math class I am going to attend so that will hopefully change.

As I want to include a XMEGA microcontroller in the design there are for sure opportunities to do some processing of the collected data, are you suggesting that I not only measure the current through the inductor but also the voltage across it?

Due to my low level of knowledge about mathematical processes I do run for cover as soon as such an endeavour is perceived but at the same time I really want to learn more about this concept because I do recognize the power of being able to utilize mathematics in such a way.

But I do have problems with the math on paper during a project like this, this is not an ordinary case of a person being priorly un-intressted in maths but it is only in the past 2-3 years that I have been capable of any real intellectual development(there's a medical story behind that, thank the gods(or the scientific revolution) for amphetamines. Seriously, of course as a manufactured medication and nothing else. This is way of topic but the fact that I have been attending this forum is a direct effect of me being under treatment with these medications and I have finally found the final solution which is named Attentin, it's the gift of being able to learn and develop an understanding for electronics which is wonderful).

And I have trouble making sense of the very simplest of equations, such as Q = C * V.
Though I have begun to research the Q, I know that the unit is jouls but I can't as of yet make out how to translate a unit of mJouls into actual current at a certain voltage over a certain time(I am just guessing that is how it would be done)

On the subject on math in the context of this project, I have gotten so far in my exploration of the XMEGA microcontroller that I can now successfully transmit data over a serial(USB) connection via a COM port so I do have the option of processing numbers on a PC if that would be more suitable. I can't really know what kind of processing is suitable for a uC and what is better reserved for a PC. But I do have a Matlab license and I do like the idea to make use of Matlab in these kinds of situation and I particularly like the idea of sending a data stream(s) and present plots of some parameter in relation to some other parameter. That is perhaps not so useful in this case as the parameters to plot against would be something that can't be adjusted with software(like number of turns).

But I may get ahead of my self here, I should start making a prototype of the suggested circuit and see how it works with the caps I have, my LCR meter is not suitable to perform ESR measurements in this case.
I measured my recreantly inquired 10,000uF caps but the meter can only handle these caps at 100/120Hz and at that frequency the resolution of ESR is pretty low and all I can find out was that the ESR is somewhere under 0,0Ohm which leaves lots of room for possible ESR. I do have multiple 2200uF 63V caps as well, I haven't measured them but I found them yesterday(I thought I had lost them) so I should use my LCR meter and see what it can make out of them.

As I make out this situation after reading posts here is that I may construct a prototype with the caps I have but that would be a relatively low peak current version, even though I don't currently have any large inductor that is going to be used at such a high current as 30 or 50A I will sooner or later be facing such a situation and for me it would feel kind of pointless to put so much time into constructing a test rig such as this and not enable it to reach those high peak currents.

It isn't that far into the future that I will need to test a forward or full bridge transformer core that is going to be able to deliver perhaps as much as 300/500W continuously, it will depend greatly on how I decide to manage the secondary side but come to think of it that situation could require rather large currents. And it may then also require a output inductor for those high currents so I should inquire a high peak ripple current capacitor ASAP.

Build your prototype with whatever parts you have there and get it working, and see some results.
It can all be built with relatively low power parts to begin with, no problem there, just realise that you can easily "wind up the wick" to the point where some part of the circuit fails.

A good starting point will be the voltage and current ratings of the shottky diodes and the mosfets. You should be able to find some specifications for the capacitor as well, and the main one to look for is the ripple current rating.

Most circuits using electrolytic capacitors are filtering fairly pure dc, and do not have high currents surging in and out of the capacitor. This circuit is really stressing the electrolytic by having extremely high current surging in and out repetitively, something most electrolytics never see, and were never designed to do.

The low esr capacitors are constructed in a different way internally to ordinary general purpose electrolytics, which have much more fragile construction.
The low esr is just a by product of the better construction and is not really relevant.

What is relevant is that the robustness of construction also produces far less internal heating, and can safely carry a much higher pulsing current without failure.

If your capacitor is unhappy it will usually get dangerously hot before it fails, and that is a pretty good guide during experimentation.

No need for paranoia here, but if you want to build a good reliable inductor tester with the capacity to measure at high peak current, it should be built with suitably robust parts. I have blown my own unit up several times, and for me, its always the mosfets that go bang.

Every time, this has happened it has been my own stupid fault, changing the frequency range on the function generator by pushing a range change button without first reducing the voltage on the dc power supply to zero.

The math is very straightforward.
The current ramps up when a dc voltage is applied across an inductor.
How quickly it ramps is proportional to voltage, and inversely proportional to inductance.

So if you know your voltage, and can see the slope of the ramp on your oscilloscope, its easy to calculate the inductance.

If your inductor normally works at with 300 volts switched across it, you can test it at only 15 volts by increasing the on time twenty times.
It will still saturate at the exact same current but twenty times slower.

Its still an accurate test, and you can sometimes hear the core whine from magnetostriction, because most of your testing will be done well down in the audio range, even though the part will eventually run at much higher voltage and frequency.

Reading Warpspeed's post #24, I agree about making some kind of oscillating circuit. Send current into the inductor during one half of the cycle, then see what current level you get during the second half.

If you saturated the inductor, then current should drop immediately when you start the second half.

What actually happens is you drive this with an exact 50% duty cycle, ideally from a flip flop, but a commercial function generator will work fine.

The current ramps up to some peak value, if there were zero losses, all the stored energy would ramp down to zero, in the exact same time it took to ramp up. All the ramp up energy is fully recovered.

That never happens, because of resistive losses, and diode drops.
You lose some energy during the whole cycle, but you do not lose much.
The inductance cannot change, its identical both ways.

But you will find the ramp up time is very slightly longer than the ramp down time. It ramps down right to zero, then sits at zero for a very short time before ramping back up again.

I also find it interesting the the make up energy from the dc power supply is a direct result of circuit losses only, so if you are ramping up to 40 amps, and only 2 amps is required from the dc supply, that is a pretty good indication of overall conduction losses.

- - - Updated - - -

Maybe I even should throw in two hall sensors... one low current and one high current so that I can if I would find the need to measure saturation in the range of 1-5A just as well as 10-50A or higher. Those numbers where just grabbed out of thin air but you get my point.
I am not sure but it sound as a hall sensor tailored for up to 100A would not be very useful for measuring 1A

Click to expand...

If you start off with say a 50A hall sensor, that will be 50 Amps full scale with one wire going through the hole.
But you can wrap five turns of thinner wire around that Hall sensor, and that would give you full scale at 10 Amps.
Fifty turns (a bit impractical) but it would give you a very sensitive 0-1 Amp measurement range with a 50A Hall sensor.

Having two Hall sensors is not a bad idea, but you can very easily increase the sensitivity of a big one by a very useful amount.

Are you talking about those hall effect sensors that looks kind of like a current transformer, connects to the PCB with through-hole pins and has a hole in the middle where the wire would go through n times?

I had tought about hall effect sensors in SOIC-8 packages, though here are PCB patterns that loops under the sensor up to 5 times. Then exiting the loop with a via.

What I use is an LA55P:
https://pdf1.alldatasheet.com/datasheet-pdf/view/115009/LEM/LA55P.html

It requires +/- 12 v supplies and produces a 50mA output for 50A input.

https://www.ebay.com/itm/LEM-CT-SEN...951067?hash=item3ab2fbd29b:g:4UsAAOSwsB9WDj-w

Ok, I wanted to show you this and ask if you see anything wrong with this?
I am actually not sure where to measure the voltage of this but in the following picture both supplies connected to the circuits are isolated from PE-ground.


I where thinking to drive this with a function generator BUT I have to look at the internal structure of IR2110 first because both scope and generator is as is to be expected connected to ground internally(I have yet to get my hands on a isolation transformer for mains application, I have a couple of possibly suitable Iron-sheet toroidal cores that use to be two mains 50Hz transformers. But I have not been able to work out any turn number that I can actually wind realistically).

I have actually never ever used a flip-flop, about time perhaps...
Do a flip-flop present any particularly attractive features in this kind of circuit or is any 50% duty cycle square wave?
I guess that an accurate 50% duty cycle would be important if the difference in time to ramp up/down is what is interesting and you want to look at the dead-time duration.

- - - Updated - - -

I have not even shown the current sensor, but I am unclear about if the voltage waveform was something you talked about from a "during the build" situation or if you mean that I should measure bit I/V in all measurements.
In any case I will take your advice and rig up a low current version.

O/K a few points to think about.

Hin and Lin will be tied together, because both mosfets are to be driven simultaneously.
I would suggest you connect a logic level opto isolator between your function generator and the IR2110. That solves a lot of potential interfacing problems, and makes the input practically blow up proof.

The next thing is supplying power to the IR2110. There will be a minimum voltage, because the IR2110 has an undervoltage shut down on the output side.
I think its about ten or eleven volts and varies a bit for different IR2110 chips.

There will also be a maximum, and although the IR2110 will stand up to 25v, the mosfet gates probably will not survive that.

This power should come from a fixed voltage supply, maybe +12v or +15v not from C bulk, because you need to be able to adjust C bulk all the way down to zero, and all the way up to whatever voltage you decide upon, and have full mosfet drive all the way.

The dc voltage you monitor is across C bulk, maybe with a multimeter.

For minimum switching loss you want really fast switching, and R2 and R4 can be reduced to perhaps ten ohms to speed things up.

Other than that, you just pass one of the interconnecting wires to CUT through your Hall sensor, either one or a few turns, and you are set to go.

No real need to monitor the actual inductor voltage.