Shock-induced AL changes

I've seen inductance v. frequency of a bead (presumably with internal cracks) that looked like random number generator output, and verified the 'cracks' hypothesis by finger pressure (which moved the cliffs and slopes). The bead looked normal. The only worse plot of component character in memory was an elderly Zener diode that curve-traced as a sloped breakdown (rather than abrupt) with circa 1V broadening of the trace. If given low breakdown current, it was either amplifying noise or generating multiple unstable current filaments. Always parallel a Zener without a capacitor!

On a sunny day (Tue, 14 Nov 2017 22:36:58 +0100) it happened Piotr Wyderski wrote in :

Compression of the magnetic material in the rather weak binder due to impact force?

Ferrite cores are susceptible to force

A normal observation is that you test your inductor/transformer and it performs as intended

Then for what ever reason you decide to pot the core, but that will change the properties since when the potting cures the stress on the cores changes the AL value. And, the time constants can be long. AFAIR many minutes

Handle cores with care, any mechanical influence may change the properties

Cheers

Klaus

I once tested a ferrite core for A_L and saturation, using a pulsed test. On adjusting the PRF, *tick* -- and now way higher current draw? I touch the core, it falls apart -- the core had shattered into regular sized hunks, because I hit an acoustic resonant frequency (and the air gap drew extra magnetizing current).

Magnetostriction is very real, and significant for some materials.

Some datasheets warn not to hit an acoustic resonant frequency, and most warn not to saturate, shock or heat-treat some materials (NiZn I think is particularly susceptible?), lest the B-H curve change.

And of course, the mechanical property of magnetic steel is well known. Supposedly, a hard enough whack, to a piece of metal aligned N-S, will magnetize due to Earth's magnetic field, and subsequently serve as a compass needle. Although I've never seen that presented as any more than an urban legend, no data to back it up. I've not observed it myself, at least with modern alloys. Physically speaking, I could imagine it being something like, crystal slippage causing magnetic domain shifting, serving much the same purpose as mechanical dithering or AC bias, momentarily overcoming the material's hysteresis. The consequential work hardening also creates hard traps that can pin domains. But I'm not aware that anyone's actually tested this.

For ferrite, it would of course be rather difficult to have slippage planes, without the mass falling apart from fracture -- but perhaps a smaller scale effect applies, whether microfractures, or something more basic.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/ 

"Klaus Kragelund"  wrote in message  
news:999dbfa0-2566-422b-acb7-f0697edf2e16@googlegroups.com... 
> Ferrite cores are susceptible to force 
> 
> A normal observation is that you test your inductor/transformer and it  
> performs as intended 
> 
> Then for what ever reason you decide to pot the core, but that will change  
> the properties since when the potting cures the stress on the cores  
> changes the AL value. And, the time constants can be long. AFAIR many  
> minutes 
> 
> Handle cores with care, any mechanical influence may change the properties 
> 
> Cheers 
> 
> Klaus

Since I worked on making MnZn ferrites a *long* time ago, we well knew that physical shock, as well as demagnetisation ( sinusoidal decaying mag field ), and thermal shock, particularly passing down through the Curie temp ( about 180 C for MnZn ) would cause an effect called disaccomodation. This is nothing to do with physical damage like micro-cracks. Its a relaxation of the magnetic domain structure within the material. There are in fact two different time decay functions, one with a TC of a few hours ( very variable, depending on processing ) and another much longer ( but smaller ) one with TC around several years.

The permeability ( and hence the Al, and other related numbers ) would go up significantly, 10% was not uncommon in the first minute, and then decay logarithmically with time over the next 24 hours or so, after that it got hard to measure due to things like temp drift. Increasing the ambient temp makes the decay much faster.

So it was always standard practice to wait at least 24 hours after any form of shock before making a precision measurement on the permeability.

For high precision inductance, eg in LC filters, you should always use gapped cores, since the shift is in the intrinsic permeability of the material. "Diluting" the material with an airgap dramatically reduces all the drift effects. Eg if the intrinsic permeability is 2000, and the gapped permeability is 200, then all the drifts due to disaccomodation and temperature, etc, go down by a factor of 10.

--
Regards, 

Adrian Jansen

Cool, thanks. So, same thing as ceramic chip caps aging? I wonder if MLCCs are shock sensitive too.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

AL isn't a critical feature for cores used in mag amps. Why are you binning with more than 1/2 order of magnitude for this characteristic?

RL

In a high A_L (~1600nH/t^2) material? I don't think there is any "explicit" binder as in the powered iron cores, just uniform ceramic. The shock is negligible, too, 2cm? I think the casue lies deeper, as Adrian suggests.

Best regards, Piotr

On a sunny day (Fri, 17 Nov 2017 20:13:48 +0100) it happened Piotr Wyderski wrote in :

He mentions 2 effects. The second one lasts longer and decreases logarithmic. Also it deceases faster at higher temperature. I would no exclude any binder effect / structure changes in the compound. But I am not sure of course. It is way below Curie point... What is then left, density of magnetic material, hitting it compresses it. Put it in a vice and see what happens (careful not to break it)? You could try to de-magnetize it if you think that is what is causing changes?

Because the intended purpose are low-power applications and the choice of suitable toroidal cores is very limited, let alone the binocular ones. My current best core is the 6x3x3 toroid made of F938 (A_L=5100). I'm experimenting with the old-school dual flux magamp designs (not the single-core Ramey construction now used in the PC power supplies) and then, to simulate a three-legged core, they have to be wound on two stacked cores (each core carries its own AC winding and a single control winding is wound on two of them at the same time). The A_L asymetry then shows up as saturation asymetry.

My long-term goal is to re-create the parametron, but I'm far from it.

Best regards, Piotr

Nobody is, hence the question. :-)

I'm able to lower the inductance by a factor of 2 just by squeezing the core using my fingers. Fully reversible as far as I can see. This can even have practical applications.

Will check that next week. :-)

Another interesting effect observed recently: When I drill a hole in the

6x3x3mm ferrite toroid (the transfluxor way) and add two windings, one wound in a typical way and the other through the small hole, the inductance of the bigger winding depends very linearly on the magnetizing current in the smaller winding, but then it reaches a saturation point of L(I=0)/2 and suddenly stops decreasing, no matter what control current is applied. Saturation of saturation, interesting. Checked that on 3 more cores made of P40 (one of them twice as big), very repeatable.

Best regards, Piotr

Can you show a picture of the core with the two windings?

What kind of drill did you use to drill the ferrite?

Thanks.

Sure, here you are (6x3x3mm toroids):

(with the cutters) (not a very high quality zoom)

It's more milling than drilling.

They're shown on the first photo, the smallest Dremel-compatible diamond cutters I had in my toolbox, don't know their exact specification. After drilling the AL value drops to about 2/3 of the initial value, again very repeatable. Drilling takes about a minute, but most of the time is spent on the initial hole positioning.

The professionals use ultrasound drilling machines, but they are way above my hobby budget:

Best regards, Piotr

Thanks. Excellent pictures. They show the two windings clearly.

However, it's not clear which winding is used to measure the inductance, and which winding is used to apply the control current. I'm assuming the winding through the hole is the main inductor, and the winding on the left with the black leads is the control winding where you apply the current.

It would appear the winding that goes through the small hole intercepts most of the flux going through the toroid. There is a tiny sliver of ferrite on the inside of the toroid but I don't think that has much effect on the results.

Could you simply wind the inductor on the right on the toroid in normal fashion without driling a hole, then measure the inductance as a function of the current through the control winding on the left? What do you use to measure the inductance?

The black one, made of thick kynar (it simulates the TIW wire, which is too expensive to be wasted for the sake of such experiments).

The small one, 9-15 turns of 0.1mm diameter wire.

Exactly the opposite. The idea is that there are two magnetic flux streams. The main one, induced by the black winding, goes around the entire core like in a normal transformer. The control winding induces flux which curls mostly around the small hole and changes permeability of the tiny toroidal slice around the hole. Since the magnetic path length of this slice is very small and so is the area, very little current is needed to saturate the slice. This results in a kind of induced airgap in the main toroid and huge inductance changes of its winding. The magic is that in the ideal case both windings don't see each other due to flux compensation.

Exactly. And it can saturate a small part of the material. In the less randomly selected materials more interesting things can happen too, e.g. flux switching. You might enjoy this:

Yes, for this particular core you need about 1.5 ampere-turn to get 10:1 inductance change.

Lutron LCR9184 set to 100kHz.

Best regards, Piotr

Here's the manual so you don't have to waste time on the Capcha:

Very interesting. Thanks.

Thanks for the excellent reference. I'm starting to think about various applications for this technology. It could make an excellent low frequency VCO where varactors are not available and conventional multivibrators are too noisy. A simple Colpitts with a current controlled variable inductor might be a simple and inexpensive solution.

I'm wondering if the hole could be drilled orthogonal to the current hole. That is, to drill into the side of the toroid on a diameter, instead of a vertical hole parallel to the axis. This should reduce the amount of material removed and increase the main coil inductance.

Further, the controlling winding could consist of n turns on one side, then another n turns on the other side of the ferrite, but wound in the opposite direction. That is, looking from the top, the bottom coil would be wound clockwise, and the top coil counterclockwise. The goal is to produce series-aiding flux that is at right angles to the flux from the main coil to improve the inductance ratio between zero and saturation, and to reduce the coupling between the two coils.

I hope my description is clear enough - I don't have any means of drawing a picture to illustrate it.

Thanks

I guess this is wrong. Both coils would have to be wound in the same direction to give a North-South/North-South alignment. Then it is just one big magnet.

However, the idea is to produce a flux that is at right angles to the flux produced by the main coil.

Now, if you saturate the ferrite at right angles to the main flux, will that still reduce the inductance in the main coil?

An alternative would be to wind the main winding around two toroids placed side-by-side, with the DC saturation winding around just one.

I kinda get what you're after, but a pencil sketch and photo off your phone would help - presumably you have a pencil and a phone?

Clifford Heath.