GROUND BALANCE ( without the ground :-) )

[moodz]

It would be great if detectors just didnt see the ground at all ... just turn them on and detect away.
Here is a thought experiment.
One way of looking at this mathematically : a detector is working by "solving" a set of simulateous equations.

for example

f(T) = yes or no : funtion of ( Target ) = 1 for target present OR 0 for no target present

f(GR, GB) = 0 : function of ( Ground Response GR ) = 0 (if target = 1 OR 0) and GB = b where b is the Ground Balance variable control setting and GR is the actual "ground target" ie ground intensity ( sand, clay, ferric oxides etc )

Now the usual way a detector solves the f(GB) aka "ground balance" is by bringing the detector coil near a ground with no target ( f(T) = 0 ) and adjusting manually or automatically the GB setting till f(GR, GB) = 0 ie ground balanced.

BUT and it turns out its actually a big BUT.

Once the detector is balanced and GB is set to some value and there is no ground and no target near the coil ( ie holding detector off ground in mid air ) the equation for f(GR, GB) = 0 is still being solved by the detector even though GR = 0 ( no ground present ).
A mathematician looks at this and says to himself "hang on -- you mean the equation still solves if one of the supposed key dependant variables is zero ??? and not only that but when the detector is near a real ground and f(GR, GB) = 0 the equation is solved for a whole range of grounds ( ie the detector is moving across variable ground ).
So this means in short you dont need that variable ( ie the GR value ) because when balanced f(GR, GB) = 0 for a whole range of grounds including no ground at all ( ie the detector is balanced )

In summary where this is leading is that you can ground balance a detector without the ground ! It was just a matter of realising the above.

Yup you could say that using lookup tables or "canned" data could be used "preset" the detector so that you an "predict" the GB setting .... that is not the optimal solution.

I can say that I can prove the GB point can be solved algorithmically without going anywhere near a ground.

so the new formula is now f(GB) = 0 where a method is used that solves only for GR = 0 ( ie no ground near the coil ) and that GB ( the GB setting ) is thus solved for all GR ( all grounds) within practical limits.
You just turn on your detector and it just is ground balanced alerady. No pumping LOL.

There is a f(GB) function running to solve the GB setting all the time ... it just does not need to be "calibrated" against a ground.

So thats the maths ! The methods ( two proposed ) well they might be patents .... I will be testing it out for real.
If I am wasting my time or "it was already done" then let me know.

How did I find this .... I actually dreamed it .. believe it nor not. I was working on a coding bug for the auto ground balance on the FPGA detector and I saw the solution in a dream and I am writing this at 5.20 AM. The bug was not a bug! Its working on the hardware. Time will tell.

[Carl-NC]

The equation for a single frequency sinusoidal VLF is something like this:

f(GR,GB) = mag(GR)*sin(ph(GR) - ph(GB))

Let's say the phase of the ground is 2°, so you adjust the GB phase to 2° and the sin() term is 0 regardless of the magnitude of GR. But if the magnitude of GR is zero then f = 0 for any setting of ph(GB). Offhand, I don't see a way around this.

The same is true for PI subtractive GB which relies on the GR decay rate:

f(GR,GB) = GR(t0) - k(GB)*GR(t1)

If GR = 0 then the multiplier k(GB) doesn't matter and f = 0 always. But when GR != 0 then f = 0 only for the correct k(GB).

What do you see that I don't?

[boilcoil]

Yes Moodz, but you will need to generate a whole family/library of curves, collected in a common point "air" and branched to different points "ground", so that when you bring the coil to the ground, the detector can choose a curve from the library to work on.

[moodz]

Yes Moodz, but you will need to generate a whole family/library of curves, collected in a common point "air" and branched to different points "ground", so that when you bring the coil to the ground, the detector can choose a curve from the library to work on.

This could work if the detector quickly matches the ground to the curve but on the fly how will possible targets ( eg small / deep ) targets not confuse the system.... my solution does not use this.

[moodz]

The equation for a single frequency sinusoidal VLF is something like this:

f(GR,GB) = mag(GR)*sin(ph(GR) - ph(GB))

Let's say the phase of the ground is 2°, so you adjust the GB phase to 2° and the sin() term is 0 regardless of the magnitude of GR. But if the magnitude of GR is zero then f = 0 for any setting of ph(GB). Offhand, I don't see a way around this.

The same is true for PI subtractive GB which relies on the GR decay rate:

f(GR,GB) = GR(t0) - k(GB)*GR(t1)

If GR = 0 then the multiplier k(GB) doesn't matter and f = 0 always. But when GR != 0 then f = 0 only for the correct k(GB).

What do you see that I don't?

Even though you see diagrams for discriminating metal detectors as a polar phase diagram with a single vector drawn on it the signal from the RX coil is actually a composite vector made up of multiple sinewaves with the same frequency but different amplitudes and phases.

For example a simple one for an RX coil on an IB .... with two phases that we might call IB ( tx imbalance ) and GR which is a ground response vector.

E1 might correspond to the GR and E2 to the IB vector.
The result is E0 the resultant vector.

So my solution rearranges ( using hardware) the below diagram to do the following things ..

1. Solve E1 and E2 phase and amplitude using only E0 and knowing how a ground signal will affect the result.
2. Use a method to solve for E2 such that any changes in E1 will have no impact on E2.

All the designs I see in VLF for instance aim to solve for E0 and calculate a VDI which is resultant of superimpose vectors E1,E2 .... En in a real system

The solution is to NOT use an FFT / Goetzel etc ... hint hint.

Below is just a general representation of the problem ....

[boilcoil]

Ha, ha you shot me.
If you say the principle, it will be very interesting.
But think about it - is it wisely to share it now?

PS: Oh, yes, you did it

[moodz]

Ha, ha you shot me.
If you say the principle, it will be very interesting.
But think about it - is it wisely to share it now?

PS: Oh, yes, you did it

LOL ... in short maths is not patentable ... methods are.

[JoyJo]

The equation for a single frequency sinusoidal VLF is something like this:

f(GR,GB) = mag(GR)*sin(ph(GR) - ph(GB))

Let's say the phase of the ground is 2°, so you adjust the GB phase to 2° and the sin() term is 0 regardless of the magnitude of GR. But if the magnitude of GR is zero then f = 0 for any setting of ph(GB). Offhand, I don't see a way around this.

The same is true for PI subtractive GB which relies on the GR decay rate:

f(GR,GB) = GR(t0) - k(GB)*GR(t1)

If GR = 0 then the multiplier k(GB) doesn't matter and f = 0 always. But when GR != 0 then f = 0 only for the correct k(GB).

What do you see that I don't?

And why multiply the "soil response" by the sine of the ground angle?

[Carl-NC]

E1 might correspond to the GR and E2 to the IB vector.
The result is E0 the resultant vector.

The White's XLT & DFX had a GB procedure whereby you lift the coil in the air and pull the trigger; then place the coil on the ground (away from metal) and pull trigger. With the coil in the air the detector would grab the X&R data which would represent E2. With the coil on the ground it would grab the X&R data which represents E0. From that it could calculate true ground phase. Most detectors don't do this, you can bob the coil over the ground and effectively get the same results. The null vector (E2) ideally will never change but can with temperature, but in a well-designed detector this will be minimized to the point that it is in the mud.

All the designs I see in VLF for instance aim to solve for E0 and calculate a VDI which is resultant of superimpose vectors E1,E2 .... En in a real system

VDIs are mostly taken from the 2nd derivative which will have removed all of E2 and most of the static effect of E1. You'd think that leaves just the target vector, except that mineralization can also rotate the target vector in a dynamic way. I don't know if any detectors try to correct for this, I expect that MF models attempt to do so.

[Carl-NC]

And why multiply the "soil response" by the sine of the ground angle?

VLF detectors have quadrature demods, so they effectively take the sine and cosine of the incoming signal to create the resistive (R) and reactive (X) signals. for GB we want the R signal to be zero in the presence of ground, even if the strength of the ground varies. So we adjust the sine() term to be zero.

[moodz]

So can we conventionally adjust the sine() to zero without a ground sample ..?

[Carl-NC]

If you don't know what the loss angle of the ground is, I don't see how. Maybe you see something I don't.

[boilcoil]

This is too complicated for me, but I think that in order to fix the system correctly, you need at least 2 points - air and ground. Without any of these points, the system is undefined.

[JoyJo]

VDIs are mostly taken from the 2nd derivative which will have removed all of E2 and most of the static effect of E1. You'd think that leaves just the target vector, except that mineralization can also rotate the target vector in a dynamic way. I don't know if any detectors try to correct for this, I expect that MF models attempt to do so.

Can multi-frequency, simultaneous operation on 3 or 5 frequencies contribute to alleviating this problem? It seems to me that multi-frequency will not solve the soil issue completely.

[Carl-NC]

Can multi-frequency, simultaneous operation on 3 or 5 frequencies contribute to alleviating this problem? It seems to me that multi-frequency will not solve the soil issue completely.

In some cases, yes. Non-viscous ground (ferrite-like) has a loss angle and magnitude that is largely independent of frequency. So, assuming there is no null vector component, you can subtract two frequencies and cancel ground without knowing its loss angle or strength. Viscous ground, though, has a loss angle that slightly increases with frequency so that won't work.