Breakthrough!!!!! Full discrimination with PI

[Tinkerer]

Experimenting with the TEM pulse method, I tried something different.
PI in differential mode.
The coil arrangement used is a TX coil of 450mm diameter and 300uH inductance, 0.8 Ohm resistance coil and cable.
The RX coil is 225mm diameter, 300uH inductance, center tapped.

The Flyback is captured with a 200nF capacitor and re conducted to the battery to save power, limiting the Flyback peak to 200V.

The PI runs at 4300 pulses per second, TX of 150uS. about 6.2A coil current at switch off.

Differential mode:

Attenuate the 200V TX Flyback to the same 30V level of the RX Flyback and then subtract the TX Flyback from the RX.
The resulting signal wave shape is shown in the picture annexed.

The target signal has a very good S/N and the sensitivity is high. To my great surprise, the FE discrimination is very good too.

Testing different metal targets, I found that the response is very different for different metals.

How does that work?????????????

Looking at the Lissajous curves, I found that the target response is in fact a phase shift as well as a amplitude change.

So this method does make it possible to discriminate targets by the phase shift, just like a VHF does.

The big difference is the amount of power in the coil.

By recycling most of the power of the TX, a very strong magnetic field with high di/dt can be built, that gives very deep ground penetration.

Tinkerer

[Carl-NC]

I suppose you were not using an IB coil arrangement, so how does subtracting RX-TX differ from an IB coil?

[Aziz]

Well done Tinkerer!

FE discrimination is possible due to coil coupling change between RX and TX coil. Even there is not obvious coil induction balance, it is still an induction balance principle. But more easy to realise.

The balance is made with linear combination the RX and TX voltage.

Good job.

Aziz

[WM6]

Well done Tinkerer!

FE discrimination is possible due to coil coupling change between RX and TX coil. Even there is not obvious coil induction balance, it is still an induction balance principle. But more easy to realise.

The balance is made with linear combination the RX and TX voltage.

Good job.

Aziz

Thank Tink!

Aziz, how ca we reach the same with mono coil, if even possible?

[mikebg]

Tinkerer, please show as circuit diagram how TX and RX coils are connected. Why RX coil is center tapped? Does it placed in induction balanced relative to TX coil? How capacitor 220nF is connected to TX coil?

[Aziz]

Thank Tink!

Aziz, how ca we reach the same with mono coil, if even possible?

This isn't possible, as the coil coupling change (TX coil = RX coil) end up in a TX inductivity change. So your reference TX voltage will also change and the response difference to a real TX/RX coil is smaller.

Ferro magnetic materials change the mutual coil coupling heavily (fundamental base for the FE discrimination). The fundamental TX frequency (+ harmonics) will be passed to the RX coil, which will change the balanced linear combination of the TX/RX voltages.

A balance is likely made like that:
a*Utx + b*Urx = 0

a, b = linear combination factors
Utx = TX coil voltage
Urx = RX coil voltage

Aziz

[Tinkerer]

I suppose you were not using an IB coil arrangement, so how does subtracting RX-TX differ from an IB coil?

When I use an IB coil arrangement, I have a TX coil, an RX coil and a Bucking coil. The Bucking coil is wound around the RX coil, with 180 degrees phase difference, so it reduces the Flyback to 0.005 %, making it possible to sample at any time, including during the Flyback, when the target response is very strong.

However, lots of people dislike the IB coil arrangement, because they think it is difficult to balance. This is true, any slightest shifting of any coil or even a lead wire during the assembly, shifts the balance.

It was only after I made an electronic adjustment circuit, that permits adjustment of the induction balance after assembly and even in the field, that the building of IB coils became very easy.

Still, people keep asking for a MONO coil solution. The closest to a mono coil solution that I could come, is a coil arrangement with separate TX and RX windings.

The separate windings, with a center tapped (Carl's suggestion, thanks) has several advantages. Among these is the possibility to make the RX coil windings of different induction, or with a different ground, or thinner wire with less capacitance etc. It also makes it easy to use a differential input preamplifier.

The differential preamplifier makes it easier to build a dedicated cable. Twisted pairs work quite well and can easily be made in short lengths. This is another important factor for using high coil current.

Then came the TEM TX METHOD (thanks Aziz for the help).

When using clipping diodes in traditional PI, to reduce the RX signal amplitude, the best part of the signal is wasted. The part where the discrimination happens and the greatest target response is eliminated. Only a very faint echo of the target response is left by the time the signal reaches near zero Volt.

So how could I sample earlier? How could I sample at the time when the target response is at it's peak?
Looking at the halve sine of the TEM Flyback and the same wave shape of the RX signal, the idea came to subtract the similar wave shape to get a "flat signal". But would the target response still be there?

Well, the target response is there, strong.

So I started sampling the RX signal at different times, using my standard targets. Using the TEM TX METHOD, I had found that it excited deeper eddy currents, that enhance the response of thick targets unlike the traditional PI, where the thin targets like foil are enhanced.

To my great surprise, I found that the DIFFERENTIAL PI METHOD (the difference between TX and RX coil) produced excellent target response for different types of targets at different times.
Excellent skin eddy current response at a certain spot in time as well as excellent deep eddy current response at different spots in time.

Then it also differentiates FE by making the FE response of the opposite polarity from non magnetic metals.

The response signal phase shifts. The two halve sine waves get distorted and phase shifted. Since the one is subtracted from the other and the difference is amplified, the resulting signal appears at different spots along the timeline.

Thus, sampling at several different times, several different response features like conductivity, thickness of the target, magnetic and non magnetic etc, can be gleaned.

Tinkerer

[Tinkerer]

TEM TX METHOD

Tinkerer, please show as circuit diagram how TX and RX coils are connected. Why RX coil is center tapped? Does it placed in induction balanced relative to TX coil? How capacitor 220nF is connected to TX coil?

Hi Mikebg,

here is Aziz's simulation circuit of the TEM TX METHOD.
http://www.geotech1.com/forums/showp...&postcount=819

The 220n capacitor size depends on the amount of joules you run in the TX pulse. I prefer to use several smaller capacitors in parallel, it is easier to make a precise peak Flyback voltage.
You can chose the peak Flyback voltage at any voltage level you can get film capacitors for, of the pulse type, but you must stay below the Mosfet avalanche voltage.

Important is the duration of the Flyback. The duration gives you the slope or di/dt or slew rate. If the slope is very steep, the eddy currents are enhancing the surface response. If the slope is less steep, the eddy currents penetrate deeper into the target, persist for a longer time and can give you information on the volume of the target.

It amounts to a target frequency response.

I like the center tapped RX coil and differential preamplifier for it's common mode noise cancelling characteristic and the fact that one can use a different ground with it.

Tinkerer

[Aziz]

Coil Coupling Factor

Hi all,

a quick calculation with my coil software gets the following coil coupling factor of TX to RX coil:
if Radius RX = 0.5 * Radius of TX coil
and Lrx = Ltx (=300µH)
k=0.1

If the TX flyback voltage is max. 200V, then the max. induced voltage at RX would be 20V (=k*200V=20V).

The coil coupling factor is calculated using the mutual coil inductances:
TX inductance only L1=300µH
RX inductance only L2=300µH
Total inductance (setting RX = TX, connecting now both TX coils in serial) gets
Ltot = 660µH

The following formula calculates coil coupling k:
Ltot = L1+L2 + 2*k*sqrt(L1*L2)
k = (Ltot-(L1+L2)) / (2*sqrt(L1*L2))
k = 0.1

Aziz

[mikebg]

RX front end

Tinkerer,
Is this your RX front end?

[Tinkerer]

Full discriminating PI

Hi Mikebg,

Attached is the circuit as I have it now. It is not complete because the free Multisim Analog version does not give more parts.
I will draw the RX input part and send it in the next post.
There is a screen shot as well as the Multisim file.
This is just the first try. It is by no means perfect. A lot of things can be improved. I welcome any constructive input.
I am sure that with your help (and help from others, everybody is welcome) we can develop this idea into a very powerful PI detector.

Tinkerer

[Tinkerer]

Differential PI

Here is the other part of the front end circuit.
Again, there are many parts that can be improved. This is just a first try to see if the idea works. There are many ways this can be done much better.

I hope to hear many suggestions for improvements.

Tinkerer

[Tinkerer]

FULL DISCRIMINATION PI

As mentioned before, there are many things to improve.
The first one I want to address, is the filtering.
We want the TX pulse filtered so that no traces of target signal remain in it.
In the RX we want to keep the target response intact.

This sounds simple, but it is more complex that it seems at first sight.

Lets look at SPOT_27-33uS, shown in the picture below. The picture shows the signal wave shape and the sampling pulse at a spot in time, 27uS after TX switch off, until 33uS.

This sample got me baffled at first. At this spot, a massive lead musket ball of 20mm diameter gives a very weak response even close to the coil.

Yet, a 1 Inch square piece of foil that has a TC of about 10uS, gives a good response. Why?

Looking at the next picture we get a clue. Where the halve sine of the TEM Flyback gets abruptly stopped, we see a little oscillation. Not all that much, but let's remember that there are a lot of amps in that pulse.

These oscillations have a frequency of about 80khz, up there where the skin effect starts getting important.
This is a frequency that is exciting small targets and foil really well.
It is not a good frequency to induce eddy currents in massive targets that have a very long TC, like a heavy musket ball.

So we have several different target exciting frequencies in our TEM TX METHOD. Each distinct frequency is ideal for exciting optimum response in a different type of target.

So before we start designing the filters, we better find out which frequencies we want to keep and which ones we want to attenuate.

Tinkerer

[Tinkerer]

In the last post we have looked briefly at the spot 32. We have seen that at this spot, we get a good response for short TC targets, specially very thin targets. We will come back to this spot or point in time, later. For now we have a look at another interesting spot.
The peak of the TEM FLYBACK.
This is a very interesting moment in time. The Flyback has been absorbed by the TEM capacitor. The capacitor is charged to the peak 200V.
For an instant, there is no current flowing in the coil, then the capacitor discharges through the coil, back to the battery.
What is the target response at the zero current moment?
We are going to take 3 different samples:
Center peak. We make this sample short, only 2uS to catch mostly the zero current moment.
5uS before peak. This is the last third of the capacitor charge slope. The initial current rush has already diminished.
5uS after peak. The current has inverted. It is now flowing out of the capacitor and back to the battery.
In the picture below we see the timing of the sample pulse at the peak of the TEM pulse.
The results are as follows:
12mm Steel ball….. 6mV negative
13mm lead ball……. 8mV positive
Crown cork (thin steel) flat…….. 8mV positive
Crown cork vertical.. 20mV negative
From the above numbers we can see that the response of the magnetic targets is negative and the non-magnetic targets is positive.
However, the thin steel gives a positive response when presented flat to the coil. Therefore, the discrimination at this spot in time is not perfect.
In the next post we will look at other different spots along the timeline.

[porkluvr]

I know I will probably be embarrassed by the obvious simplicity of your answer, but I can't stand it any more.


Tinkerer, WHAT IS "TEM"?