Ferrite-copper diode and non-linear effects in ferrite dielectrics

[F6FLT]

Hi All,

Last weekend I tested high-permeability ferrites as dielectrics, and I just had two feverish days because of all the weird things that were happening.

I connected 2 electrodes to these ferrites, and observed what happens when both AC and DC current are injected. A lot of things happened.

As this test followed my idea of an electromagnetic version of the Archimedes screw (see other thread), I was looking for signal at a double frequency of the ac input, which would be a sign of a double half-wave rectification due to an effect of Lorentz force on electrons bathing in the magnetic field created by their own direct current.
And I did have one, which depended a lot on DC polarization through ferrite!

In the setup, I placed an LC notch circuit to attenuate the fundamental frequency, and used an SDR to visualize the spectrum (it is clearer than the scope). See the measurement schematic and copies of spectra.
While we can barely see the harmonic 2, polarizing the ferrite in DC raised its level up to 40dB more! The polarization voltage must not exceed the AC voltage, otherwise the gain will no longer increase and even decrease.

I have done dozens of tests before I understand. This is not Lorentz's force at work but non-linear junction effect between the copper contact and the ferrite.

It should be noted that good contact is very difficult to obtain. I had first glued copper strips on the ferrites, with a lot of pressure. Despite this, the DC current fluctuates enormously, the resistance is not stable, but the frequency doubling effect is extremely clear.
I finished by tightening my contacts with screws, and there the resistance is stable, but the effect is not more pronounced.
The resistance must be of the order of KΩ to a few KΩ.

My first test was done with a large ferrite core, the next ones with smaller ones (see photo, for the last one with screws, the transformer coils were kept). The frequency doubling has been obtained for all.
If you are trying to repeat the test, choose ferrites whose resistance can be detected directly with an ohmmeter. They seem less frequent than good insulating ferrites. I only found three kinds in my collection.

When I realized that it was a junction effect, I had fun replacing the DC polarization with a 1 KHz signal, so I got an amplitude modulation. I also tried a frequency mixing of same principle as in superheterodyne radio receivers.
I finally replaced a ferrite by two inverted diodes connected in series, each in parallel with a resistor, and this setup gave similar results confirming that we have a diode effect with a copper/ferrite contacts. If the result is similar, on the other hand the effect is considerably less pronounced than with diodes.

I don't know if the principle can be used, it's not what I was looking for, but I think it's always good to know and perhaps to investigate further!

[Itsu]

Hi F6FDL,

thanks for the info, its a novel idea to use a Software Defined Radio (SDR) app for showing signals across a spectrum
I see you use HDSDR  (http://www.hdsdr.de) which i also use in combination with a Receiver dongle.

You just input the signal from your circuit into your PC microphone plug i guess.

Nice experiment with the junction effect between the copper contact and the ferrite,  it shows that if one knows
what one is doing you can stumble upon some effects and recognise it as a specific known effect.


Itsu 

[muDped]

Very, very interesting indeed!  So it is acting to some extent
like the Varactor Doubler in a sense.  Quite efficiently too.

Brings to mind a Sea Story from my time with the Navy's
Tempest Program in the mid '60s.  I have to run an errand
now but I'll continue when I return.

It was discovered that radio transmissions from Navy ships
developed Spurious Emanations in the form of Harmonics
and Intermodulation Products.  It was further discovered that
numerous corroded metallic junctions were the cause,
particularly at the Antenna Tuner/Couplers which were located
near the base of the Whip Antenna.  The very important
Ground connections had become corroded/rusted in
many instances and were acting just like a non-linear
diode.  The Whip Antenna (35') was used for radio
transmissions in the 2 MHz to 30 MHz range at power
levels up to 500 Watts.

[F6FLT]

Hi F6FDL,

thanks for the info, its a novel idea to use a Software Defined Radio (SDR) app for showing signals across a spectrum
I see you use HDSDR  (http://www.hdsdr.de) which i also use in combination with a Receiver dongle.

You just input the signal from your circuit into your PC microphone plug i guess.

Nice experiment with the junction effect between the copper contact and the ferrite,  it shows that if one knows
what one is doing you can stumble upon some effects and recognise it as a specific known effect.
...

Hi Itsu,

HDSDR is the most technically complete SDR software I know, but not the best GUI (I tried 4 others, and I developped mine but it takes so much time...).

I use the RSP1 from SDRPlay. It's a 12bits SDR. I also used the dongle as your but the SDRPlay is better (up to 8Mhz of band between 10 Khz and 2Ghz, no hole, better dynamics and sensitivity). There is an ExTIO dll to work with HDSDR.
A sound input can work for low frequencies (< 96 or 192 KHz depending on sound card), not enough for 300 KHz. Nevertheless it's possible to use it here. The effect occurs also at lower frequencies. I tried 10/20 KHz, it works, the problem is that it's more complicated to get a notch filter at this frequency, so I test mainly around 150/300KHz.

I have posted too quickly my previous results and possibly bad conclusions. I just get new amazing effects, with another ferrite of less resistivity. It's crazy. Here is new observations in "bulk" :
- it seems that the type of metal in contact does not matter (so is there really a junction?)
- it seems that there is no need for wide contact, sharp contact points are enough for effects as good or better (a point rather in favor of a junction). I now use small pliers with sharp teeth.
- magnetism finally plays a role! A neodymium magnet placed near the contact shifts the minimum, so that a DC voltage is required to compensate it, but the effect is neither necessarily the same nor the opposite when North and South direction is reversed.
Anyway, I don't understand anything at the moment.  I'm completely drowned in the flood of all the hypotheses to be made.

[Itsu]

F6FLT,

thanks for the info, i did also hear good reviews of the SDRPlay.
Its fun to play with these very broadband receivers.

Ok about your results of this weekend tests.
This reminds me of some tests Slider and me did on another forum using a piece of raw germanium and a sharp screw in a magnetic field.

It supposed to get you in contact with the "other side",  so beware of what you are doing   

Forum here:  http://www.itcbridge.com/forum/view_topic.php?id=1842&forum_id=32&page=1


Ok  let your new results sink in for a while it will be clear eventually i hope.

Itsu

[F6FLT]

...
This reminds me of some tests Slider and me did on another forum using a piece of raw germanium and a sharp screw in a magnetic field.

It supposed to get you in contact with the "other side",  so beware of what you are doing   

Forum here:  http://www.itcbridge.com/forum/view_topic.php?id=1842&forum_id=32&page=1
...

"After some moving around of the needle, I received the words 'Hey Mark' from an American sounding woman!!!!!!"
And no antenna?    Dead or alive?
May be a woman in her truck driving in a neighbourhood street while talking on the CB channel 19?   (Occam's razor explanation    )

[ion]

Interesting work

What role could the lowly copper oxide rectifier be playing in this? Maybe try stainless steel or gold contacts?

[F6FLT]

...
It was discovered that radio transmissions from Navy ships
developed Spurious Emanations in the form of Harmonics
and Intermodulation Products.  It was further discovered that
numerous corroded metallic junctions were the cause,
particularly at the Antenna Tuner/Couplers which were located
near the base of the Whip Antenna.  The very important
Ground connections had become corroded/rusted in
many instances and were acting just like a non-linear
diode.  The Whip Antenna (35') was used for radio
transmissions in the 2 MHz to 30 MHz range at power
levels up to 500 Watts.

Hi muDped,

I agree that many metal oxides have non-linear effects, including negative resistance such as Zn, and therefore intermodulation products and harmonics when at work in the radio field.
I thought of something like that for the doubling of the frequency I observe, which is why I talked about possible "junctions".

Nevertheless, it is less and less likely. My last tests with small pliers with sharp teeth rubbed well against the ferrite must have eliminated all traces of oxide, and no change of the results.
I have also replaced the ferrite by an ordinary resistance and the effect disappears, it's not an artifact, it really comes from the ferrite and it should be very easy to duplicate by every one.

I am now trying with more than two electrodes, and I will then try with other metals.
The first attempts are getting stranger and stranger.  I have separated the electrode providing the DC from the one providing the AC, and put one or two other electrodes for the ground. Whatever their order around the ferrite toroid, the effect is almost the same!
Another strange behavior is that the DC current doesn't depend on the path length. When the electrodes are just side by side, or when they are diametrically separated, almost same DC current depending only on the contact quality.

[F6FLT]

...
What role could the lowly copper oxide rectifier be playing in this? Maybe try stainless steel or gold contacts?

I will try with other metals because I believe less and less in a junction effect between ferrite and oxide or metal.
Another idea is also to use asymmetric electrodes, for example a wide and flat one, and the other sharp because currently, we have two equivalent contacts so that the current can flow, and therefore we can't study one contact at a time, which can mask effects that compensate each other. There is no shortage of work...

Update1:
- it doesn't depend on metals. Al, Fe Ok.
- very important: I tried with four electrodes along the ferrite toroid, two providing AC through capacitors, two providing DC. Order doesn't matter, it works. BUT one of the DC contacts must have continuity to ground for DC. It doesn't work with floating DC, I verified it with a battery instead of the power supply. So it's not a question of current, but a question of charges: the ferrite must not be neutral!
- AC is rectified (no DC is provided but 20v AC => 10µA DC in a 1K resistor connected to the DC terminals).
Perhaps I'm finally right with the Lorentz force. I'll be back after I check.

Update2: see attached file
- just a small piece of ferrite between two clamps works perfectly. The coil placed next to it, serves as a probe for the scope.
- by changing the frequency to 119 KHz and an AC voltage 10Vpp, I get a clean and well balanced signal showing the rectification with respectively a positive polarization current of +1v, and a negative of -1v.

[F6FLT]

Here is the simplest method to see the non-linearities of a conductive ferrite: a simple 10 ohm resistance in series with the ferrite. The voltage across the resistance gives the current. That's all!

Measurements made at 100 Hz. That doesn't really change from the measurements above 100 KHz.

In general, non-linearities occur for high current or voltage values, with clear threshold effects. This is not the case here. At 0.5v AC, we still see non-linearity. Below, it seems to fade, but difficult to be sure, because the current measurement is very noisy.
So I still don't have a theory.

[muDped]

Are you able to measure the resistance of the ferrite
material?  Does it vary in response to temperature or
voltage?

Apparently metal oxides are known to have non-linear
resistance characteristics.

I can see the non-linearity in your photos above but I
am unable to read all of the small print in your comments.

Are you applying a DC Bias somehow with the AC component
superimposed?
- - - - - - - - - -

Duh!  Senior Moment overcame.

I was able to download your graphic file and view it enlarged
so now am able to read your comments.  Well done!

I wonder what a DC plot of Voltage and Current would reveal?
Is it some form of Varistor?

[F6FLT]

Are you able to measure the resistance of the ferrite
material?  Does it vary in response to temperature or
voltage?

Apparently metal oxides are known to have non-linear
resistance characteristics.

The resistance depends on the ferrites and the contacts. Usually it is in the range 700 Ω -> 10 KΩ. It doesn't depend on temperature, and almost not on the voltage in the range 0-20V (not tried upper).

What is surprising is that the resistance is practically independent of the ferrite length between the two contacts. Whether the two contacts are diametrically opposed on the ferrite core, or just side by side, the resistance hardly changes. Is it the contact that makes the resistance, and that the mass of the ferrite is much better conductive? I don't know, it's just a possible explanation.

I can see the non-linearity in your photos above but I
am unable to read all of the small print in your comments.

You can download it and open it in Paint, it should be readable.

Are you applying a DC Bias somehow with the AC component
superimposed?

I did it in this test, the generator allows it, it's the value "offset".
(I used the other channel in my previous tests).

[F6FLT]

...
I wonder what a DC plot of Voltage and Current would reveal?
Is it some form of Varistor?

Good idea, I should have started there.

Here is the slope for a -10 -> +10v voltage ramp (yellow, 3v/div). Current is in mV/10Ω (i.e 5mA/div).
We see that the non-linearity is not much and you are right, it looks very similar to that of a varistor.

[Verpies]

There is an old paper by two polish guys who apply electrical current to a piece ferrite in order to change its magnetic reluctance.

See: "Konrad and Brudny in "An Improved Method for Virtual Air Gap Length Computation," in IEEE Transactions on Magnetics, Vol. 41, No. 10, October 2005."

Also, a Bulgarian inventor Valeri Ivanov also claims an electrically controllable reluctance switch.

A piece of ferrite that changes its magnetic reluctance without mechanical drag-back penalty is the Holy Grail for magnetic motor and generator designs.

For example a patent US9742252B2 uses this method to make a switched reluctance AC generator:

Another electrical means of implementing a reluctance switch is the placement within the primary magnetic path of certain classes of materials that change (typically increase) their reluctance upon the application of electricity. A different way of implementing a reluctance switch is to saturate a sub-region of a primary magnetic path by inserting conducting electrical wires into the material comprising the primary magnetic path.

[JimBoot]

Good idea, I should have started there.

Here is the slope for a -10 -> +10v voltage ramp (yellow, 3v/div). Current is in mV/10Ω (i.e 5mA/div).
We see that the non-linearity is not much and you are right, it looks very similar to that of a varistor.

Does ferrite conductivity still matter at this stage of the experiment? Are you still only using ferrite that directly gives a reading on the ohm meter?  Edit. Have you tried ferrite ring magnets? Lowest I can get on the one I have here is about 350k.

[ion]

For good contact I usually paint the surface of interest on the ferrite with a silver conductive paint and embed some fine wires into the paint. First coat is usually lightly diluted with acetone to bond into the surface. You can also just bond the copper contacts using this method, provided they are not oxidized.

As I remember, ferrites with Mn and/or Zn were the low resistivity types, around 10k or less with others (not containing Mn, Zn) in the megohms.

FWIW

[F6FLT]

Does ferrite conductivity still matter at this stage of the experiment? Are you still only using ferrite that directly gives a reading on the ohm meter?  Edit. Have you tried ferrite ring magnets? Lowest I can get on the one I have here is about 350k.

I only use ferrites that gives a reading on the ohm meter. Resistance around 1 KΩ to 3 KΩ, depending on the type of ferrites (that one of higher permeability presents the lower resistance).
Whatever the ferrite and the contact, V/I slopes are similar.

[F6FLT]

For good contact I usually paint the surface of interest on the ferrite with a silver conductive paint and embed some fine wires into the paint. First coat is usually lightly diluted with acetone to bond into the surface. You can also just bond the copper contacts using this method, provided they are not oxidized.

As I remember, ferrites with Mn and/or Zn were the low resistivity types, around 10k or less with others (not containing Mn, Zn) in the megohms.

FWIW

I have no conductive paint for trying your method.  It was also my first idea: the more the surface, the best the contact, that's why I coated the ferrites with copper. Then I noticed that even if I pressed the surfaces hard with pliers, I could achieve the same result with much less surface area, with loose clamps but having sharp teeth.

What surprises me the most, but I don't know if it's specific to ferrites or if it happens for all bodies with comparable resistivity, is that the potential of any point on the ferrite is about half the voltage (AC or DC), regardless of the position of the two electrodes and the probe, even when the voltmeter probe is really very close to an electrode. The potential varies by less than 8% along the ferrite

For example, with my toroidal ferrite and diametrically separated contacts (5cm), one contact powered by 8V and the other connected to ground, the potential near the 8V contact is 4.65V (even at less than 1mm). Near the ground contact it is 4.38V, and all intermediate values are obtained by sliding the voltmeter probe on the ferrite on whatever half circle between the two contacts.
When the two contacts are very close, the potential near the 8V contact is 4.15V. Near the ground contact it is 3.97V, and we also have all the intermediate values by sliding the voltmeter probe on the ferrite, making almost a full turn.

So everything happens as if the ferrite was excellent conductor, and that the resistance was mainly at the contacts. But if this were really the case, these ferrites would give rise to strong eddy currents and the electronics industry would not use them.

[F6FLT]

There is an old paper by two polish guys who apply electrical current to a piece ferrite in order to change its magnetic reluctance.

See: "Konrad and Brudny in "An Improved Method for Virtual Air Gap Length Computation," in IEEE Transactions on Magnetics, Vol. 41, No. 10, October 2005."

Also, a Bulgarian inventor Valeri Ivanov also claims an electrically controllable reluctance switch.

A piece of ferrite that changes its magnetic reluctance without mechanical drag-back penalty is the Holy Grail for magnetic motor and generator designs.

For example a patent US9742252B2 uses this method to make a switched reluctance AC generator:

I observed some magnetic effects but very slight, it may be a measurement bias.
In any case, Murphy is still there, and changing a reluctance of a magnetic circuit when a flux passes through it costs energy.

[F6FLT]

A little review to finish as far as I'm concerned, this thread on conductive ferrites.


The conductive ferrites had a resistance on the order of KΩ between two contacts.

A non-linearity V/I is observed, similar to that of varistors. This non-linearity makes it possible to obtain signals with a frequency twice that of the input signal.

The resistance between two contacts is highly dependent on the contact pressure.

In my last tests using contacts with very tight hose clamps, the resistance dropped to less than 50Ω. Under these conditions the non-linearity is very low.
Contrary to appearances at first, these ferrites finally have a high conductivity.

No magnetic effects were observed except experimental bias. Even with a sensitive differential circuit (contacts powered symmetrically and signal detected by a scope probe placed at equal distance from the two electrodes), a neodymium magnet moved along the ferrite does not cause any change in the output signal.


Concerning my goal of the other thread on the Archimede's screw, one of the possibilities of which would be to use the Lorentz force in a dielectric to reveal electrical effects, these ferrites as dielectrics are not suitable, nothing indicates the slightest effect of this kind.

[F6FLT]

I just did some additional testing with conductive ferrites and confirmed the nonlinear effects of 2019, and the points made above.

By biasing the ferrite even with a small current (500 µA to 2 mA), in the same direction as the AC current, the nonlinearity is increased. We can easily obtain the double of the AC frequency and we can make HF mixers (but not very efficient, the power ratio of the signal at the output frequency to the one at the input different frequency is low).
If the DC current is too high, the resistance of the ferrite decreases and the non-linearity fades.

Finally I obtained an equivalent schematic of the ferrite (when it is not polarized and therefore with weak non-linear effects) : see attached picture.
A triangular signal allows to observe the effect, and the similarity of the signal between the one with the ferrite and with the R and C components.