SPICE Simulation Challenge

[Davor]

I think I was on the right track with my previous Fe model. It was based on a negative inductance, something that naturally happens in vicinity of transformers, and it placed the response in a proper quadrant. However, response was completely wrong. Instead of a response circling from left to right with frequency, it did the opposite. But at least the quadrant was correct.

Now I have a small correction, and it seem OK. It should be properly tested, of course. It is based on a inductance difference, which effectively gives an equivalent of μr >> 1
Also it is rotating as it should, and it ends at high frequencies as mostly resistive, yet still on ferrous side.

If only I could produce a Cartesian plot in AC simulation... Here are the magnitude/phase, real, and imaginary outputs. I'd say it is close enough, but please knock yourself perfecting it. It should be possible to curve-fit some materials with responses measured on professional equipment.

Here goes:

[green]

So my conclusion (at the moment) is that real effective ferrous / non-ferrous discrimination cannot be achieved with a mono coil.
Now, go and prove me wrong ... not with an LTspice simulation, but with some real hardware.

P.S. I'm talking about a standard PI setup here, and not other variations that use truncated sine or triangular waves, etc.

My attempt with a mono coil. Thought I could see a difference in the beginning based on what I saw with the IB figure8 coil. Didn't see a difference with the mono coil so repeated test with IB figure8. Can see a difference. Any suggestions why I don't see a difference with the mono coil? Maybe something to try?

[Qiaozhi]

My attempt with a mono coil. Thought I could see a difference in the beginning based on what I saw with the IB figure8 coil. Didn't see a difference with the mono coil so repeated test with IB figure8. Can see a difference. Any suggestions why I don't see a difference with the mono coil? Maybe something to try?

The same reasons I presented in post #30. The effects of redistribution and absorption are simply not visible when using a mono coil, and can only be seen with a balanced coil arrangement.

[Davor]

But the whole point is that PI is inherently an in-phase detector. The whole difference in balanced vs. mono coil is the possibility to look at the time where excitation field still collapses, and quadrature response still creeps through.
However... there is a difference in in-phase frequency response of ferrous materials against the non-ferrous. Up to ~ 30kHz the Ferrous in-phase response is a low pass, while non-ferrous is a high pass. Because EF knocks the response tail to zero, a Fe response, with a fat low pass tail, may sink below non-Fe responses for the early samples. And there you have it.

I believe my simple model from post #31 may prove useful. The only thing I'm missing are the different material data. You may find some data scattered on Internet, but fun part is that you may tune this model to accommodate. See this paper: Effect of nanograin?boundary networks generation on corrosion of carburized martensitic stainless steel but especially the Figure 3, and compare those to my model. I'd say I nailed it

[Davor]

By in-phase i meant resistive - in PI there is little point speaking in terms of phase.

[Carl-NC]

Personally I do not think the difference between ferrous and non-ferrous targets can be attributed to an inductance change, nor to a shift in resonance. It's the way the magnetic field is distorted by these different targets that is the root cause. This field distortion causes an imbalance in the coil coupling.

Correct.

If you have a VLF detector with a balanced coil system, the RX loop has a residual signal when no target is present. If a non-ferrous target gets near the coil, there is a redistribution of the magnetic field that causes an imbalance. For a concentric coil, this results in an increase in the amplitude of the RX signal, and all target types cause a left shift in the phase. If a ferrous target gets near the coil, there is some absorption of the magnetic field that also causes an imbalance, but in this case the RX signal experiences a decrease in amplitude. The phase shift is used by the GEB channel to detect the presence of a metal target, whereas the amplitude changes are used by the DISC channel to indicate ferrous or non-ferrous.

If an IB loop has a perfect null, there is no residual signal and either magnetic or eddy responses will cause an increase in RX amplitude. It's only when the coil is mis-nulled to one side or the other that there is a residual signal and a target causes a decrease in amplitude. Whether it is ferrous or non-ferrous that does that depends on which side of null the coil starts out at. This is true for either DD or concentric.

Both magnetic targets and eddy targets cause a distortion in the TX field. In VLF, the magnetic response is due to the gap in the B-H curve which produces a phase response that is limited to 0-90 degrees (all my phases are WRT the TX coil current). Ferrite is 0 deg because it (ideally) has no B-H gap. Eddy responses incur an automatic 90 deg phase shift in developing the surface EMF, so those target responses are limited to 90-180 deg, a fortuitous coincidence.

In PI, the same mechanisms cause off-time decay responses. Eddy responses are pretty close to exponential. Magnetic responses are not, they are still controlled by the magnetic lag dictated by the B-H curve. Ferrite again has no gap and therefore no magnetic lag; when the TX field collapses, the ferrite response collapses along with it. Targets with a B-H gap are not instantaneous, they have a time lag in their collapse. Then there are targets with magnetic remanence, which is yet another mechanism.

[Davor]

The B-H curve becomes a B-H line for small H, yet even the weak ferrous responses produce 90-180? shift.
When you observe the inductive frequency response curves of steels, you'll find a resistive magnitude stronger at lower frequencies, followed by decline, and finally with a constant resistive response at high frequencies. It may convince you the remanence is dominant in the softest of ferrites, but that would explain viscosity far better than ferrous behaviour.
In impedance plane you may notice the steel responses working as if the non-ferrous response is merely translated to the ferrous side (by means of μr). Even the lift-off curves look as if they are merely translated from the non-ferrous quadrant.
The main difference is that resistive response at high frequenvies of non-ferrous has a flat top, while ferrous has a flat bottom. The big difference is at low frequencies, and I may already see implications for PI.

[Carl-NC]

The B-H curve becomes a B-H line for small H, yet even the weak ferrous responses produce 90-180? shift.
When you observe the inductive frequency response curves of steels, you'll find a resistive magnitude stronger at lower frequencies, followed by decline, and finally with a constant resistive response at high frequencies. It may convince you the remanence is dominant in the softest of ferrites, but that would explain viscosity far better than ferrous behaviour.
In impedance plane you may notice the steel responses working as if the non-ferrous response is merely translated to the ferrous side (by means of μr). Even the lift-off curves look as if they are merely translated from the non-ferrous quadrant.
The main difference is that resistive response at high frequenvies of non-ferrous has a flat top, while ferrous has a flat bottom. The big difference is at low frequencies, and I may already see implications for PI.

My understanding is that the B-H curve tends toward a tilted ellipse at small signal, sort of like a Lissajous pattern. Which still results in a 0-90 deg phase range. Many steel targets -- especially flat steel, like bottle caps -- have a dominating eddy response that pushes the overall response to the non-ferrous side. For a ferrous test target I like to use a nail-on-end, which has practically no eddy component.

[Davor]

Bottle caps are all tin or zinc plated. When those eventually rust, there's no non-ferrous response. But the geometry plays a distinct role, and nail-on-end, or pliers-in-hand work ferrous every single time.
For that reason we may only combine different target simulants and observe their contributions to the outcome... a bottle cap may work as both ferrous and a non-ferrous target, and if your detector discriminates, and is fast enough, you'll hear them both as da-deee-da while passing over it, and you'll spare yourself a futile digging.

[ivconic]

"...Bottle caps are all tin or zinc plated. When those eventually rust, there's no non-ferrous response. But the geometry plays a distinct role..."

Bottle caps are tough to beat trash in the real world searching.
If we speak on common bottle caps produced in Europe.
In most of the cases; rusty or not: those will produce spurious and pretty suspicious response, tending to non-ferrous loookalike.
More rust on it; more "crispy" will be the ends of response. Rise doubts, yet reminds on some other desirable targets.
I've been fighting with this so often on my sites. Deus is having nice little feature "Silencer", adjustable from -1 to 4.
To get more better performances on all other targets; it is desirable to set it on -1. Especially if you search for smallest coins (like i do).
But once you step on "bottle cap field" it is a horror. So than i usually do set it up to 4. It reduces the error in significant percentage.
And now we came to the second part of this; the geometry. It reduces the error... much better if geometry is not perfect circle.
If it is remained circle still; "silencer" is struggling.
And now i would add a third factor too; the orientation and position in soil... of such bottle cap.
If it is leveled and parallel to soil surface; it will produce crispy non ferrous response. If it is positioned in some angle; non ferrous response will move to more accurate ferrous response; as the angle is larger.

Silencer at 4 and bottle cap parallel to soil surface = crispy and "chirpy" non ferrous response.
Silencer at 4 and bottle cap by 90 degree to soil surface = perfectly accurate ferrous response.

And this is only a part of a story. True completely only in case of modern produced bottle caps plated with zinc or other similar material.
Much older bottle caps, produced 40-50 years ago; add to this story more details and aspects.
I could write quite a study only on bottle caps as main subject.
Don't know about other modern md models; but on Deus, Gold Max Power and Gmaxx II there is mentioned "Silencer" option.
Interesting to play with it and it's setups. By my humble opinion it is the one of most important features at those machines.

[Qiaozhi]

If an IB loop has a perfect null, there is no residual signal and either magnetic or eddy responses will cause an increase in RX amplitude. It's only when the coil is mis-nulled to one side or the other that there is a residual signal and a target causes a decrease in amplitude. Whether it is ferrous or non-ferrous that does that depends on which side of null the coil starts out at. This is true for either DD or concentric.

Perhaps I should have been more specific when I stated that non-ferrous targets cause an increase, and ferrous targets cause a decrease in amplitude. I was talking within the context of Tesoro-like detectors, such as TGSL and Raptor, where the coil is mis-nulled to one side, and the direction of amplitude shift was with reference to commercial Tesoro coils.

So yes, what you said above is absolutely true. This is also the point I've been trying to get across to coil builders who insist on attempting to create the deepest null possible, and then wonder why they cannot get the coil to ground balance, and discrimination doesn't work.

Both magnetic targets and eddy targets cause a distortion in the TX field. In VLF, the magnetic response is due to the gap in the B-H curve which produces a phase response that is limited to 0-90 degrees (all my phases are WRT the TX coil current). Ferrite is 0 deg because it (ideally) has no B-H gap. Eddy responses incur an automatic 90 deg phase shift in developing the surface EMF, so those target responses are limited to 90-180 deg, a fortuitous coincidence.

I'll do some more hacking around with the LTspice core model, and try to create a realistic ferrous target. However, I'm not hopeful based on my previous simulation tests.

In PI, the same mechanisms cause off-time decay responses. Eddy responses are pretty close to exponential. Magnetic responses are not, they are still controlled by the magnetic lag dictated by the B-H curve. Ferrite again has no gap and therefore no magnetic lag; when the TX field collapses, the ferrite response collapses along with it. Targets with a B-H gap are not instantaneous, they have a time lag in their collapse.

OK ... so maybe not all is lost. That's something to look at.

Then there are targets with magnetic remanence, which is yet another mechanism.

I'm not sure how you would detect that, unless it's possible to measure the coercive force required to kill off the residual magnetism. Will have to think about it some more.

[Davor]

George, I checked your model, and the ferrous part is wrong. I attached a part of your schematic that needs a correction, and I don't expect it to work properly on the first run, but it will respond in a correct quadrant. Also you should try it in AC simulation prior to attaching it to a narrowband RTX.

[Qiaozhi]

George, I checked your model, and the ferrous part is wrong. I attached a part of your schematic that needs a correction, and I don't expect it to work properly on the first run, but it will respond in a correct quadrant. Also you should try it in AC simulation prior to attaching it to a narrowband RTX.

I don't understand how that could work, as you've removed the target inductance and replaced the mutual coupling capacitance with a resistor and capacitor in parallel.

This method, that uses a combination of controlled sources, is an overly complex solution which attempts to remove the necessity to define coupling coefficients between the various inductors. The simple reason for this is because LTspice does not support the use of its Chan core model together with mutual inductance. The actual target model is the part in the middle, and the two outer sections define the mutual inductances between the TX and RX loops and the target. Although you can now use the core model with this setup, the results are not correct. I suspect this is because the target inductance is calculated implicitly from the core parameters and the geometry, whereas the mutual inductances are calculated from an explicit target inductance value. Hence I've decided to revert back to the use of coupling coefficients, and somehow defining a model for the ferrous target to give a more realistic magnetic response.

As far as I can see, the simple LR target model works just fine for non-ferrous targets that produce an eddy response.

Also, when I tried to use the overly complex model in a PI simulation, there were some difficulties connecting both ends of the model to the a single TX loop. Therefore it would be much easier if it was possible to use the original setup. The problem is not so much getting the correct phase, but to also produce a decrease in amplitude so that it can simulate correctly in a Tesoro-like detector circuit.

[Davor]

As ever you like it. Good luck obtaining correct phase response.

As for steel frequency response, seek a paper "FREQUENCY-DEPENDENCE OF RELATIVE PERMEABILITY IN STEEL" by N. Bowler. That paper alone is enough to sink a lot of PI nonsense related to Fe.

[Davor]

BTW, a resistor and capacitor in parallel to a current source = leaking integrator. A coil parallel to a current source = differentiator, and not a target inductance.