How do I measure various parameters of a ferrite toroid

[ZeroResistance]

I'm keen on measuring various magnetic parameters of a ferrite toroid.

Namely
1. Max Flux Density / Saturation Flux Density.
2. BH curve
3. Coercivity
4. Remanance Flux Density
5. Permeability

Can these be measured with a simple DC source / Oscilloscope ?
Do we need hall sensors or another other sensors to do these kind of measurements.

In case I'm asking for too much, even max flux Density and BH Curve would do for now.
Thanks in Advance!

[iMo]

Can these be measured with a simple DC source / Oscilloscope ?

You would need an AC source, afaik..

[ZeroResistance]

Ok, thats good to know!
However any idea how do I go about measuring max flux density / saturation flux density?

[Colin55]

You need a CRO and a variable power supply and a variable oscillator that produces sine and square wave.
Then you need a ferrite toriod and some winding wire.
When you have all these sitting on your bench, let me know.

[ZeroResistance]

You need a CRO and a variable power supply and a variable oscillator that produces sine and square wave.
Then you need a ferrite toriod and some winding wire.
When you have all these sitting on your bench, let me know.

Yes now I have all three sitting on a bench, please let me know.
Scope PC OScilloscope Pico 2204
PS 0 to 30VDC, 5A
Function Generator Feeltech FY3200S / 24Mhz

[T3sl4co1l]

1. Set up a pulser circuit.  Easiest to do this in time domain.

2. Measure the core dimensions: OD, ID and height.
l_e = (OD + ID) / 2
Ae = (OD - ID) * H / 2
(Actually a little bit less because of geometry, and rounded corners and coating thickness if any.  Adjust if you like.)

3a. Estimate flux based on known properties.  Ferrites saturate at 0.2 to 0.45T.  Flux is:
Phi = Bmax * Ae

Flux is in units of Vs/t, so multiply by number of turns to get the in-circuit value, the flux on the total winding.

3b. Measure it with a pulser.

Make a basic flyback circuit: transistor pulling to GND, inductor from +V to drain, diode from drain to output cap, load resistor from cap to +V.  Drive transistor with a func gen, square wave, +10V/0V levels (verify with scope, don't trust the generator's dials!), low duty cycle (5-10%?), variable frequency.

Ensure the supply is well bypassed (a few 1000s uF local to the circuit).  Add a series current sense resistor with the inductor or transistor source.  Typical value say 0.1 ohm, and an IRFZ46N or the like for the transistor.

We use a square wave because, for a constant voltage, flux increases linearly with time.  So we read off time on the oscilloscope as flux.

Watching the current waveform, we expect it to rise linearly with time (current proportional to flux, i.e., constant inductance), until saturation occurs, at which point current shoots up rapidly.  This is why we use a variable frequency, and start on the high side and adjust downwards until saturation is observed.

One catch: note that a flyback circuit is unipolar.  It doesn't apply reverse flux to the inductor.  Ferrites tend to have some remenance, meaning the core remains somewhat magnetized between pulses.  This reduces your measurement somewhat.  If you were going to use them for half-wave applications anyway (flyback or 1 or 2 switch forward converter), that's actually more accurate, but a full-wave application (inverter, forward converter), you'll get somewhat more than twice the measured flux.  This is particularly exaggerated in magamp cores, which you will find have very little flux relative to their size.  Use this to identify them, and test them with another method (a half bridge inverter, maybe).

Now we can determine A_L, mu and Bmax.

L = V * dt / dI = A_L * N^2
A_L = mu * Ae / l_e
(Note that mu = mu_r * mu_0 by convention, where mu_0 ~= 1.257 uH/m, or nH/mm if you prefer.)

If you find mu is anomalously low, and Bmax seems to be awfully high (or it's not saturating at all because DC resistance is dominating -- in which case, use more turns and heavier wire, and try again?), it's probably a powdered iron, not ferrite.  Powder is used for inductors (energy storage), ferrite for transformers (power transfer).

4. Loss: harder to measure, you'll probably not get a reasonable figure with this setup (unless it's a shitty #26 powdered iron, in which case the turn-on step might be enough to read as parallel resistance, YMMV).  Instead, set up a resonant circuit with sine wave excitation, and use this method to find it:
https://www.seventransistorlabs.com/Calc/RLC.html#frq

Losses generally depend somewhat on level, so keep that in mind.  You may not be able to drive more than a few volts with a function generator, so a real at-power test has to be done with an inverter.

Also, if it's a powder core, you may be able to identify it by color.  If it can be positively identified, the datasheet will give useful info.  They don't usually give Q unfortunately, but it can be derived from the material curves (loss at frequency).  Here are some examples: https://www.seventransistorlabs.com/Images/Powder_Core_Q.png

Tim

[voltsandjolts]

T3sl4co1l, thank you sir, you always take the time to provide in-depth and insightful replies. I invariably learn something from all of your posts.

Powder is used for inductors (energy storage), ferrite for transformers (power transfer).

I didn't know that. I guess that is a 'rule of thumb' as I have seen powder toroid transformers but what is the reasoning behind it?

[David Hess]

We use a square wave because, for a constant voltage, flux increases linearly with time.  So we read off time on the oscilloscope as flux.

I have done it this way before but had to compensate in the math for winding resistance.  It worked great for measuring inductance versus current and as a bonus, it returned ESR at the same time.  I was only interested in saturation flux.

Designing a high current constant voltage pulse generator output stage was fun when I did not know exactly what I was doing.  I got down to about 1.5 microseconds but if I did it today, I could do at least 10 times better but a B-H curve tracer would be more interesting.

Powder is used for inductors (energy storage), ferrite for transformers (power transfer).

I didn't know that. I guess that is a 'rule of thumb' as I have seen powder toroid transformers but what is the reasoning behind it?

Powdered iron has a soft saturation characteristic making it more suitable for energy storage or any application where a DC current is present.

[T3sl4co1l]

A definition: inductors are intended to store energy, transformers aren't.

Thus, a flyback transformer would be more descriptively called a coupled inductor.

I don't know how common this definition is, but I think it is helpful.  If you adopt it, do make sure to specify what you mean, in any case.

A non-ideal transformer encompasses both, having magnetizing and leakage inductances specified.  Theoretically speaking, it's just different values of the same general component.

For cores, the root of the matter is this:
e = B^2 / (2*mu)
The lower the permeability is, the higher the energy density (e, J/m^3 or whatever).  A very low mu isn't practical (~mu_0) because of the nonzero resistivity of copper -- if we could get rid of that, we could make even more compact inductors.  (Indeed, we can, and do; superconducting magnets aren't quite as practical for AC or energy storage purposes, but they are indeed capable of storing an impressive amount of energy!)  So, as it happens, we find best results with a mu_r around 10 to 50.

(And yes, if we were stuck with, say, zinc wire, we'd find best results around a mu_r of, say, 50 to 250.  The physical size would be MUCH larger -- the wire has to be so many times thicker to get the same overall resistance -- but you'd end up with a similar inductance and Q factor.)

Whereas transformers aren't intended for energy storage, just for isolation and transformation.  The magnetizing current should be very small (reduces errors in signal applications, and losses in power transformers), meaning the mu_r should be very high (typically 500-5000 for power transformers, up to 15k for ferrite pulse transformers and CMCs, and special materials (nanocrystalline and permalloys) over 100k).

Ferrite of course isn't useless for inductors, you just have to gap it so the average permeability drops into the sweet spot.  Typical ferrite inductors have a good Q, but they can be more expensive.

Tim

[beowulfenator]

Can these be measured with a simple DC source / Oscilloscope ?

I've come across this simple device:
http://elm-chan.org/works/lchk/report.html

You can wind a few turns and measure saturation current.

[ZeroResistance]

Can these be measured with a simple DC source / Oscilloscope ?

I've come across this simple device:
http://elm-chan.org/works/lchk/report.html

You can wind a few turns and measure saturation current.

This is pretty amazing, its a similar setup to what T3sl4co1l mentioned.
However how do we convert saturation current to Saturation density? I

[chickenHeadKnob]

A definition: inductors are intended to store energy, transformers aren't.

Thus, a flyback transformer would be more descriptively called a coupled inductor.

I don't know how common this definition is, but I think it is helpful.  If you adopt it, do make sure to specify what you mean, in any case.

A non-ideal transformer encompasses both, having magnetizing and leakage inductances specified.  Theoretically speaking, it's just different values of the same general component.

F.. snipage...

Ferrite of course isn't useless for inductors, you just have to gap it so the average permeability drops into the sweet spot.  Typical ferrite inductors have a good Q, but they can be more expensive.

Tim

If a clueless noob like myself wanted to experiment with a sine output royer oscillator operating at say 40-50 kHz what core to choose? I would think energy storage is required for the oscillator yet I also want it to function as a step down transformer.

[ZeroResistance]

1. Set up a pulser circuit.  Easiest to do this in time domain.

2. Measure the core dimensions: OD, ID and height.
l_e = (OD + ID) / 2
Ae = (OD - ID) * H / 2
(Actually a little bit less because of geometry, and rounded corners and coating thickness if any.  Adjust if you like.)

3a. Estimate flux based on known properties.  Ferrites saturate at 0.2 to 0.45T.  Flux is:
Phi = Bmax * Ae

Flux is in units of Vs/t, so multiply by number of turns to get the in-circuit value, the flux on the total winding.

3b. Measure it with a pulser.

Make a basic flyback circuit: transistor pulling to GND, inductor from +V to drain, diode from drain to output cap, load resistor from cap to +V.  Drive transistor with a func gen, square wave, +10V/0V levels (verify with scope, don't trust the generator's dials!), low duty cycle (5-10%?), variable frequency.

Ensure the supply is well bypassed (a few 1000s uF local to the circuit).  Add a series current sense resistor with the inductor or transistor source.  Typical value say 0.1 ohm, and an IRFZ46N or the like for the transistor.

We use a square wave because, for a constant voltage, flux increases linearly with time.  So we read off time on the oscilloscope as flux.

Watching the current waveform, we expect it to rise linearly with time (current proportional to flux, i.e., constant inductance), until saturation occurs, at which point current shoots up rapidly.  This is why we use a variable frequency, and start on the high side and adjust downwards until saturation is observed.

One catch: note that a flyback circuit is unipolar.  It doesn't apply reverse flux to the inductor.  Ferrites tend to have some remenance, meaning the core remains somewhat magnetized between pulses.  This reduces your measurement somewhat.  If you were going to use them for half-wave applications anyway (flyback or 1 or 2 switch forward converter), that's actually more accurate, but a full-wave application (inverter, forward converter), you'll get somewhat more than twice the measured flux.  This is particularly exaggerated in magamp cores, which you will find have very little flux relative to their size.  Use this to identify them, and test them with another method (a half bridge inverter, maybe).

Now we can determine A_L, mu and Bmax.

L = V * dt / dI = A_L * N^2
A_L = mu * Ae / l_e
(Note that mu = mu_r * mu_0 by convention, where mu_0 ~= 1.257 uH/m, or nH/mm if you prefer.)

If you find mu is anomalously low, and Bmax seems to be awfully high (or it's not saturating at all because DC resistance is dominating -- in which case, use more turns and heavier wire, and try again?), it's probably a powdered iron, not ferrite.  Powder is used for inductors (energy storage), ferrite for transformers (power transfer).

4. Loss: harder to measure, you'll probably not get a reasonable figure with this setup (unless it's a shitty #26 powdered iron, in which case the turn-on step might be enough to read as parallel resistance, YMMV).  Instead, set up a resonant circuit with sine wave excitation, and use this method to find it:
https://www.seventransistorlabs.com/Calc/RLC.html#frq

Losses generally depend somewhat on level, so keep that in mind.  You may not be able to drive more than a few volts with a function generator, so a real at-power test has to be done with an inverter.

Also, if it's a powder core, you may be able to identify it by color.  If it can be positively identified, the datasheet will give useful info.  They don't usually give Q unfortunately, but it can be derived from the material curves (loss at frequency).  Here are some examples: https://www.seventransistorlabs.com/Images/Powder_Core_Q.png

Tim

This is tremendously profound! Many thanks for your generosity and your valuable time!

So as as summary of your process
First determine L, then AL, then the absolute permeability of the core.
All this is brilliant! 

Then using all these, determine the saturating flux density of the core?
Er! But how to calculate the max flux density in the core? 

This is an excellent process that you have suggested and does the job with simple equipment. However can this done more accurately with a sensor like a hall sensor.

[BravoV]

Can these be measured with a simple DC source / Oscilloscope ?

I've come across this simple device:
http://elm-chan.org/works/lchk/report.html

You can wind a few turns and measure saturation current.

This is pretty amazing, its a similar setup to what T3sl4co1l mentioned.
However how do we convert saturation current to Saturation density? I

In this forum, we have similar inductor saturation tester circuit with energy recycling, designed by respected member here Jay_Diddy_B -> Inductor Saturation Tester



And successfully built by other member here too

[T3sl4co1l]

If a clueless noob like myself wanted to experiment with a sine output royer oscillator operating at say 40-50 kHz what core to choose? I would think energy storage is required for the oscillator yet I also want it to function as a step down transformer.

Yes, you need magnetizing inductance to set the resonant frequency, in that case you would have a "coupled inductor", probably one with tight coupling.

Tim

[T3sl4co1l]

Then using all these, determine the saturating flux density of the core?
Er! But how to calculate the max flux density in the core? 

Putting it all together: Bsat = t_sat * Vcc / Ae.

A Hall effect sensor measures H (magnetic field intensity), in an air gap -- not actually more accurate, when you want to measure the flux inside a high permeability magnetic path!

Tim

[coppercone2]

they need equaations on this forum so bootleg

[ZeroResistance]

Then using all these, determine the saturating flux density of the core?
Er! But how to calculate the max flux density in the core? 

Putting it all together: Bsat = t_sat * Vcc / Ae.

A Hall effect sensor measures H (magnetic field intensity), in an air gap -- not actually more accurate, when you want to measure the flux inside a high permeability magnetic path!

Tim

Ok, so the (t_sat * Vcc) part is actually Volt-seconds right and that equals flux ??   
I agree that B = flux / Area
Secondly if we have 10 turns wound on the core? do we also multiply the Volt-seconds part with the Number of turns?

[GadgetBoy]

Just a nubbin thought. Would a gaussmeter be useful for this?

Sent from my ONEPLUS A3000 using Tapatalk

[ZeroResistance]

Just a nubbin thought. Would a gaussmeter be useful for this?

Sent from my ONEPLUS A3000 using Tapatalk

Noble thought  However as T3sl4co1l said earlier how do we insert a gaussmeter in series with a high permeability magnetic path without disrputing the very thing we are measuring itself. AFAIK gaussmeter can measure fields which are leaking out of a core / wire etc. But it cannot sense what is happenning inside a magnetic core.
Unless the gaussmeter uses some method of transformer action, I don't see how it can see inside a closed magnetic core like a toroid, It might be useful when the core geometry is open like a rod core?

[T3sl4co1l]

Bingo.

Note that, in case of a rod core, there is fringing flux everywhere, so we could use a sensor to map out the field around it, but there is no one singular number to say what's happening.

(Rod core inductors saturate somewhat more softly than closed style cores do.  The middle reaches Bsat first, which causes mu_eff to drop, so the inductance also begins to drop.  As current continues to rise, the saturated volume grows, mu_eff drops further.  Eventually the ends also become saturated, and the inductance drops to its air-cored value.)

The canonical way to measure flux, is with Faraday's law.  It's simply a voltage on a winding, then, and we can integrate to find flux (a passive RC integrator, or active op-amp integrator, can be used).


This is used in compensated current transformers, for instance:

In a regular CT, the sensed current is transformed (stepped down by ratio) and dropped across a "burden resistor" and sensed as a voltage.

In a compensated CT, there are two secondary windings: one driven and one sensed.  An error amplifier reads the sense winding, and servos the driven winding to force the sense voltage to zero.

Zero sensed voltage, means zero core flux, means freedom from core saturation and bandwidth down to DC!

The driven winding has DCR and LL voltage drops, but as long as the error amp's output has enough voltage to overcome those drops, it can successfully force the induced voltage to zero.  No current is drawn from the sense winding, therefore no voltage is dropped across its DCR or LL, and the sensed voltage is exactly the induced voltage (EMF).

Finally, current is sensed in the driven winding (usually with a shunt resistor), and you have a complete current sensor.

Because EMF is servoed to zero, the flux is very nearly zero (within limits of the error amp's accuracy: input offset, and voltage gain), so such a sensor is theoretically DC-capable, though because we're sensing through a transformer, we do need to do the good ol' calculus "Plus A Constant" chant.  In any case, a real one can significantly extend the bandwidth of a CT, from say ~kHz (for a regular CT) to fractional Hz.


All this, to say through example: yes, we can measure flux, or at least something very closely related to it (Faraday's law, EMF = -dPhi/dt), and we can apply calculus with some passive or active circuitry (integrator or differentiator) to get that last step, plus or minus a constant, or within some limits of DC stability.

And also yes, there are two fluxes, core flux (for a single turn), which is B * Ae, and circuit flux, which is B * Ae * N.  It may be helpful to carry turns as a unit, so you remember flux density is flux per area per turn, and also magnetic field intensity is amp-turns per meter, H = N * I * l_e.  (Which means, since B = mu*H, mu must have units of flux per amp per meter per turn squared.  Which is good, because A_L = mu * Ae / l_e gives us the necessary per-turns-squared in the formula!  Dimensional analysis, it's good stuff. )

Tim

[coppercone2]

compared to the other selectable elements, what is accurate enough for measurement of these parameters?

can you get anything useful trying to 'voltnut' this shit?

for small signals/filters you would end up being interested in the distortion resulting from manifestations, for chokes your gonna end up having heavy margins because of modern active semiconductors, for electromagnets your gonna end up testing it

is there useful things in power electronics or maybe for protection you can get out of precise measurements of these parameters that I am missing ? (with equipment more advanced then the kind described in this thread).