Taking a look at GROUND BALANCE on the bench

[Qiaozhi]

No, I do not want to copy a Minelab machine. I want you to show me a working analog circuit of what you say. Show me a PI circuit with frequency domain target signal extraction.

I get the distinct impression that you're going to have a long wait.

[Davor]

Point is that each subsequent sample in a PI system has more and more reduced frequency content, so in fact you do have a kind of frequency domain processing whether you want it or not. Reality check... to detect small gold you need short delays, e.g. high frequency content.
With enough samples, and some proper weighting function, you could process samples by comb filtering method and learn more about target tau... which is not unlike nowadays rigs with several samples.

[Tinkerer]

The gold prospecting machines operate at different frequencies depending on nugget size.
For large nuggets, the operators uses low TX frequency. See such machines of MINELAB.
However we live in 21st century and even in the second decade. We should use DSP. Note that patent pending for US4506225 is written in 1981.

No, I do not want to copy a Minelab machine. I want you to show me a working analog circuit of what you say. Show me a PI circuit with frequency domain target signal extraction.

[mikebg]

WE NEED DSP !

Would you be so kind and show me an analog circuit to do that?

Tinkerer

The gold prospecting machines operate at different frequencies depending on nugget size.
For large nuggets, the operators uses low TX frequency. See such machines of MINELAB.
However we live in 21st century and even in the second decade. We should use DSP. Note that patent pending for US4506225 is written in 1981.

[Tinkerer]

Would you be so kind and show me an analog circuit to do that?

Tinkerer

[mikebg]

PHASE DIFFERENCE GND - TGT

....This is a time domain detector. If you know how to extract the target signal out of the signal shown, by using frequency analyzing, please explain how to do it directly, not with theoretical graphs and mathematical formulas, because I do not understand what you are saying.

A gold target, (pure resistive, non permeable) and a ferrite (non conductive, permeable) target gibe a response that is 180 degrees apart.....

Thinker, this is not true. Your oscilloscope shows that all metal detectors are time domain instruments. We all live in Time domain, but you are using a term "phase", attributable to Frequency domain. Can you explain how the phase angle appears on your oscilloscope? Please don't use your oscilloscope and Time domain for analysis.

The most suitable tool is Frequency domain because it removes the mist of math operation "convolution". Another tool is the DSP, but it uses terms as "deconvolution" and "impulse response" (see patent US
The visual Frequency analysis shows that if we usе suitable frequencies, we can make the phase difference between gold nugget and ferrite halfspace less than 180 deg. See how we can obtain phase difference almost 90 deg or 135 deg.
At frequency A we have ferrite response 0A1 and gold response 0A3. The phase difference is almost 90 deg. This is ideal for synchronous demolator.
At frequency B we have ferrite response 0B1 and gold response 0B3. The phase difference is 135 deg. This not ideal for synchronous demodulator, but visual analysis shows that at 45 deg and 135 deg it operates well.

http://www.geotech1.com/forums/showthread.php?16793-Frequency-Domain-and-Time-Domain-relationships&p=164634#post164634

[Tinkerer]

Tinkerer
The first 30 usec of your traces look different than what I would expect. Have you tried reducing the amplifier gain to maybe 4? I think the no target signal should start at .6 volts times the gain and decay from there. Maybe I'm wrong.

You are right, the wave form looks different from the traditional PI wave form. The traditional PI uses clipping diodes at about 0.6V and high gain in the preamp.

The TINKERERS uses an induction balanced coil to get rid of the Flyback voltage spike of several hundred V and uses no clipping diodes.

This is why, with the TINKERERS, you see the part that is normally clipped and the part that is usually lost in the preamp saturation.

You would need the usual 1k gain in a fast traditional PI preamp to see the response of the 1" square alu foil.

Tinkerer

[green]

Tinkerer
The first 30 usec of your traces look different than what I would expect. Have you tried reducing the amplifier gain to maybe 4? I think the no target signal should start at .6 volts times the gain and decay from there. Maybe I'm wrong.

[Tinkerer]

Here are a few more pictures.

All have the ground, as the earlier ones, but then I added a target also.

1) Ground only
2) Ground and ferrite
3) Ground and 1"square alu foil that has a TC of 10us
4) Ground and silver dollar

Remember this is at the output of the preamp, that has a gain of only about 10. therefore the change in the signal wave form is very small for the alu foil target. Look at the tips of the sinusoid at the transient.

I find that I am showing the full dynamic range that this screen shot can show, the minimal, just barely visible deflection for the alu foil and the maximum of the silver dollar that just hits the edge of saturation on the negative rail.

Taking samples at several spots along the time line, we can obtain information about some characteristics of the targets.

We can also add or subtract or multiply or divide the results of the different samples we can obtain additional information. DSP would be good for that.

Tinkerer

[Tinkerer]

mikebg,

I don't understand what you mean. Could you explain?
I posted pictures of the wave forms, so that you can explain on these wave forms how to extract the target signal and how to remove the ground signal.

This is a time domain detector. If you know how to extract the target signal out of the signal shown, by using frequency analyzing, please explain how to do it directly, not with theoretical graphs and mathematical formulas, because I do not understand what you are saying.

My statement above is based on the following observation:

A gold target, (pure resistive, non permeable) and a ferrite (non conductive, permeable) target gibe a response that is 180 degrees apart.

An iron target has conductivity and has permeability. Sometimes, depending on the shape and orientation of the target, the response is very near equal from the conductivity as from the permeability, then the net signal is nearly 0.

I appreciate your help

Tinkerer

[mikebg]

.....We could also try to add a short TC resistive target. I can see some change with a 1" square alu foil, that has a TC of about 10us, however, at the gain of 10, the change is very little and probably difficult to see on the picture.

Suggestions??????????? Tinkerer

I have a suggestion:
Let somebody explain two figures.
The first one is a block diagram of principle on which received signals are formed. It explains what means "difficult ground".

http://www.geotech1.com/forums/showthread.php?19942-X-vs-R&p=163793#post163793

The second figure explains why the conductivity generates frequency dependent resistive and reactive components, and when magnetic permeability contains resistive component.

http://www.geotech1.com/forums/showthread.php?19942-X-vs-R&p=163886#post163886

These two figures contradict your statements in post # 1:

I open this thread to discuss the matter of GROUND BALANCE as seen during bench testing and simulations.
Difficult ground is the one that gives a resistive response and also a reactive response.

The resistive response is due to the conductivity of the ground.
The reactive response is due to the magnetic permeability of the ground.
The combination of the 2 responses in a wide variety of proportions, amounts to the generalized HOT GROUND.
....Tinkerer

[Tinkerer]

Thank you mickstv.

So what do we see here?

1) We see the dynamic range near maximum. With +/-5V supply, the pre-amp saturates at about 4V. If we were to use +/-12V, we could increase the gain considerably. No need to increase the dynamic range, this is already extreme.

2) We see 2 spots along the cycle time, where there is no signal change. If we sample at the place where the traces crosses over, or at the end of TX, the changes are minimal.

We can also look at the crossover as the pivot. If we sample before the crossover, the signal change is negative relative to the NO ground signal, after the crossover, it is positive.

3) The biggest change in the signal is at the first positive swing of the signal, at about 6us. If we sample there, taking a 5us sample window, we get the highest amplitude.

Now, please suggestions for more tests.

I could think of adding a strong ceramic magnet to the cannon ball. This should increase the reactive response, without increasing the resistive response.

We could also try to add a short TC resistive target. I can see some change with a 1" square alu foil, that has a TC of about 10us, however, at the gain of 10, the change is very little and probably difficult to see on the picture.

Suggestions???????????

Tinkerer

[mickstv]

Hi Tinkerer, here you go.

[Tinkerer]

Hi Tinkerer, when you post the pictures I'll superpose them using gimp.


Cheers
Mick

With a gain of only 10, the bricks alone could not be seen, while keeping a full wave form on the screen. Therefore I added an antique 6 pound cast iron cannon ball to simulate a very hot ground.
I set the distance of the ground target to be such as to give a somewhat similar target response amplitude as the silver dollar at the center of the coil, that is, close to saturation of the pre-amp, to represent the maximum dynamic range.

The vertical setting is 1V/div.

The horizontal setting is not calibrated, but the full cycle is about 100us, for 10,000 pulses per second.
Here are some screen shots:

NO ground_1, transient only
NO ground_2, full cycle
With ground_1 transient only
With ground_2 full cycle

Your offer of superposing the pictures is much appreciated. I am going to wait, to see the superposed pictures, before taking more pictures, as we may want to make changes to the setup.

All the best

Tinkerer

[Tinkerer]

That's not necessarily true, if the powder is baked in place by heat from say a bushfire it would have been created with domains aligned with the earths magnetic field and since Australian dirt is often well packed, hard as buggery and too dry for worm activity it probably still contains a lot of order even decades later.

Perhaps a good approximation of this type of hot ground would be, find a patch of neutral ground, at least few square meters, throw down some iron oxide pigment and an assortment of different sized brick fragments for some hot rocks. Insert a few targets, pack it all down and light a fire on top. Let it cool down, maybe walk on it a bit to throw in a bit of surface disorder.
I'm not sure if this fits into your definition of 'bench testing', its more 'simulated field testing' but should be far more useful. What your trying simulate is a target in a changing matrix, waving the same piece of matrix under the coil doesn't simulate that well.


Midas

I agree on your description of the hot ground.

The question is to look at the changes in the signal wave form on the oscilloscope, generated by the hot ground. To do that in the field is too complicated.