Square wave Pulse Induction

[mickstv]

Hi deemon, thanks for uploading the pictures. Over the next few weeks I'll rough up a circuit and see if I can get the TX section working as per your schematics.


cheers
Mick

[Waikiki_Sweep]

Pulse Induction detectors magnetizing minerals in ground by powerful impulses. Magnetic field from the magnetized ground makes false signals. Square bipolar wave should demagnetize it and reduce ground effects. Brilliant!!!!

[deemon]

Pulse Induction detectors magnetizing minerals in ground by powerful impulses. Magnetic field from the magnetized ground makes false signals. Square bipolar wave should demagnetize it and reduce ground effects. Brilliant!!!!

Of course , it depends on the ground magnetization speed ... For example , if we have a mineral that needs a big time to magnetize - this bipolar pulse technology will give a significant advantage . But if we have a mineral with fast reaction ( like a kind of "diluted ferrite" ) , another feature must work - I mean this constant current pulse approach , that I used in this design as well as in previous one ( recuperative PI deviсe , from another my topic ) . You see , if we maintain the coil current constant after current reverse - when the target response being received - we don't change the magnetization of the ground near the coil during all the receiving interval , thus discarding its influence . But the main problem in my "ground rejection" study is an absence of the ground "bad enough" All my circuits can easily reject all kinds of our local ground and stones , but people talks that Australian ground must be much worse .... so I cannot know if it would be such effective in that case ....

[deemon]

By the way , I spent a lot of time on experiments in my home lab , with different signal processing algorithms ( see in post 29 ) , and achieved a good results . For example , the third variant ( version 3.0 ) is capable to discriminate colored and black metals ( ferro-discrimination ) . So we have here a "hybrid" of a pulse and IB technology in mono-coil device , getting closer to the "ultimate metal detector" , as I told before I took some photos of the signals from the different ferro and non-ferro targets - I will upload them today ....

[WM6]

Excellent work deemon, go ahead to finalisation.

[deemon]

The only thing that I need to finalize it is a time ...

Here are the photos that I told about . The signal was taken after the preamp stage in the VERSION 3.0 configuration , from the targets of different time constants and ferromagnetic properties . Mono coil is connected to the power chain that described in this topic , with additional compensation network , pulse frequency is 2 khz . Lower trace on the scope is a coil voltage ( flyback pulses ) .

The first is the idle signal , without the target .



This is the signal of a big and thick aluminium plate .



Here is the signal of the copper ash-tray .



The signal of the bronze statuette .



Coke 0.33 can .



Thick aluminium foil .



Thin aluminium foil .

[deemon]

Now let's go to the ferromagnetic targets .

The first is the signal of a big ferrite stick .



Big iron scissors .



Tin can , the coil is perpendicular to the can bottom .



The same tin can , the coil is parallel to the can bottom .

[deemon]

As we can see on the photos , the signal in this configuration ( version 3.0 ) does exist even on the flyback interval - because of the full coil balance we don't need here to suppress or limit signal during a flyback . So it might be interesting to inspect this interval more thoroughly .... so let's do it

The signal of the typical colored metal target with medium time constant ( 0.33 coke ) , flyback interval is magnified by the scope .



And here is an example of a typical black metal target ( tin can ) .



As we can see - the aluminium target response signal does simply "grow" from the zero line during a flyback , closely to a half-sine law - and then decays to zero with exponential law ... the behavior of the iron target is more complicated .

[Aziz]

Oh man!, you are screwing Mr. Candy's latest rectangular wave patent. *LOL*
Thanks for your contribution.
Cheers,
Aziz

[deemon]

Oh man!, you are screwing Mr. Candy's latest rectangular wave patent. *LOL*
Thanks for your contribution.
Cheers,
Aziz

Yes , Mr.Candy missed the train with his patent .... what a pity

By the way , I didn't finish yet . In the next post I'll explain all this discrimination algorithm . It's based on the same correlation technology that I used in my first recuperative device . And I'll explain even more - how to "fill the hole" between IB and PI technologies , and how to obtain all their advantages in one device - an ultimate metal detector

[Aziz]

Yes , Mr.Candy missed the train with his patent .... what a pity

By the way , I didn't finish yet . In the next post I'll explain all this discrimination algorithm . It's based on the same correlation technology that I used in my first recuperative device . And I'll explain even more - how to "fill the hole" between IB and PI technologies , and how to obtain all their advantages in one device - an ultimate metal detector

Oh man!, I'm doing the same at the moment ("fill the hole" between IB and PI). But in a very KISS and simplified design.
A new ultimate sound card detector controller. *LOL*
Aziz

[deemon]

Oh man!, I'm doing the same at the moment ("fill the hole" between IB and PI). But in a very KISS and simplified design.
A new ultimate sound card detector controller. *LOL*
Aziz

It's very good , Aziz - we'll be able to compare our approach and the basic ideas . By the way , although I do my device in analog , all my signal processing can be easily transferred to digital domain . The only thing that I cannot "digitize" is my power chain

[deemon]

So guys , let's continue
After watching the signals being received from the real targets , we must think how to deal with them Our system is bipolar , so the target response does present in the both half-periods in the opposite polarity , and the first thing that we must do is to "rectify" them - I mean that we need to invert every odd half-period pulse on the preamp output , in order to transform our signal to the unipolar shape . We can do it easily , using the inverting stage ( op-amp based ) and the simple CMOS switch - as shown on the upper block diagram on the picture . After doing this we'll find that the signals has two main components - AC and DC . As we can see from the picture - DC component gives us a very useful information ( that absent in the classic PI devices ) , whether the metal target is colored or black . But why we are able to get it here ? It's just because our system is balanced and we can transmit and receive at the same time ( like we do in IB devices ) , so while the black target increases the coil inductance , the colored target does the opposite thing - it decreases inductance , and they produces the opposite polarity of the unbalanced signal . And when we apply the signal to our "full-wave active rectifier" - we are transforming this unbalance signal polarity to the DC component , that can be measured and analyzed for the black/colored ( ferro/non-ferro ) discrimination purpose . So we can see that our system behaves here just like the IB system , and can do the same things .... but with mono-coil search head - it's a substantial advantage

But what about another component of the signal - AC component ? As we can see from the picture - it's a well-known exponential decaying function of target response , the function that everybody can see in a classic PI machine . And this signal doesn't depend on a target magnetic permeability , being equal for the colored and black metals , but directly informs us about another important target property - a target time constant . Short ( fast ) exponential decay means that our target does have a little TC , and vice versa . And now we must solve another task - to determine the target TC . Here we can use the technology that I already tested in my previous device , I mean my recuperative PI machine described here - http://www.geotech1.com/forums/showt...y-recuperation . I used there a kind of correlation approach , based on the target response exponential function ( T(x) on the picture ) being expanded in the Taylor series . In that device I had been used two correlator blocks ( TILT and PARABOLIC ) , but in order to increase a precision it's better to use three of them , adding a CUBIC function correlation too . But don't forget that in reality our series always begin with the constant term . If we write it in the mathematical form - T(x) = T(0) + T(1)*T1(x) + T(2)*T2(x) + T(3)*T3(x) ..... and so on . In theory this series does have an infinite terms quantity , but in a real device we can cut them , because every next term has less and less magnitude , and T3 ( cubic term ) is really quite enough . And we can see here that in my previous device I used only T1(x) - LINEAR ( TILT ) function , and T2(x) - PARABOLIC function . But what is a T(0) ? T(0) is that constant term of the series that I told about here .... and now we can understand that T(0) is nothing else but a DC component , being selected from my signal , just after the "signal rectifier" and a low pass filter And this is why my new device can do the thing that previous one couldn't - I mean the ferro/non-ferro metals discrimination . Previous device , being closer to a classic PI - ignores this DC component ( that depends on the immediate reaction on the coil current and can be obtained only by transmit and receive at the same time , the thing that my old device cannot do ) . In another words , this new coil balancing technique allows us to obtain ALL the target information in one device , using the mono-coil search head .

Then I will compare a different technologies ( IB and PI ) in terms of "information quantity" being received from the target .
To be continued ....

[deemon]

And now we finally can understand , how this wide-band technology does correlate with the well-known previous principles , I mean inductive balance and pulse induction . As I had shown before , if we excite the target with this square wave pulse train - we can obtain the same exponential decaying target response as we can get from a well familiar PI device . And what can those devices do ? In the simplest PI machines ( with one sample pulse ) we can operate only in the "all-metal" mode , and cannot determine both the metal type ( ferric and non-ferric ) and the target TC too . When we add more samples , we can measure target TC more or less precisely , but a ferro-discrimination is still impossible . Using my terminology , I can say that those devices can utilize only the "highest" terms in the Taylor series of the target response function ( T(1)-T(3) ) , but without T(0) . If we take into account the fact that this T(0) signal depends on magnetic permeability of the objects near the coil ( as I explained before ) - we can easily understand why all PI machines does have a good "ground tolerance" . It's just because the ground has a ferromagnetic properties and might give a most strong reaction on the T(0) signal . But if the metal detector cannot receive it - it has very weak reaction to the ground , of course . And we must notice that the simplest PI device cannot separate T(1)-T(3) terms , receiving them "in general" , receiving only the sum of them , more or less effectively , and this is why they cannot distinguish a big piece of metal from the little one . But my new circuit ( VERSION 3.0 ) can do this and it can obtain the T(0) term also , so it can do the thing that PI device with mono-coil never did - I mean ferro-discrimination function . In another words , I can say that my device "includes" PI principle in it and all PI devices are a kind of "subset" of my device . It's become obvious if I "cut" my circuit , I can get a simple PI device with "all-metal" operation ( VERSION 1.0) , or a more complicated VERSION 2.0 ( see a block diagram in the post 29 of the topic ) , being equal to a PI device with a TC measuring possibility . And we can easily see that 1.0 version receives all the exponential decay in general ( as I explained in the post 34 ) , and the 2.0 version can "disassemble" this function and analyze its parts , measuring the target TC .

But what about another well-known technology - IB ? How can we compare this new approach with it ? Maybe it can do something that this new device cannot ? In order to make it more clear we must make a simple "mental experiment" . Just imagine that we add a narrow-band filter in my signal chain after the preamp stage . Everybody knows that a square wave signal can be presented as a sum of a harmonic frequencies with odd numbers . For example , if my device works at 1 khz PRR - it actually transmits on air a number of sine waves - 1 khz , 3 khz , 5 khz , and so on ... and all the transmitted frequencies being received and used in the correlation algorithm . But if we install a filter , tuned on the fundamental frequency ( 1 khz ) in the receiver signal chain - we'll cut all the higher harmonics and leave only the 1 khz sine wave in the receiver , and the result would be the same as we'd transmit ( and receive ) only this frequency , like IB devices usually does . And what we'll have on the our demodulator channels ? In the classic IB we transmit the sinewave , and receive the target response - the same sinewave but with the phase shift , caused by the target . But how we can measure this phase shift ? We use a pair of synchronous detectors ( analog multipliers ) with low-pass filters on the out . In fact , they are the same correlators that I use in my circuit . And on the reference input of the first multiplier we send the sine signal ( in phase with the transmitter wave ) , and on the second multiplier reference input - we send the cosine signal , having a 90 degree phase shift with the transmitter wave . By the way , this cosine function is an integral of the sine function on the first channel reference . And the output signals of those multipliers ( correlators ) we usually call I and Q signal ( in phase and quad phase ) .... but in metal detectors we usually call them X and R signals . X signal is proportional to the "instant reaction" of the objects , without any phase shift - it depends of the target ( ground ) magnetic permeability , and R signal is proportional to the "shifted reaction" , due to the eddy currents in the target . And further we can use both signals for the discrimination purpose . Polarity of the X signal can show whether the metal is black or colored ( ferro-discrimination ) , and X/R ratio can show the phase angle of the target signal , depending on its time constant ( TC ) .

And what we'll have in my device with rejected signal spectrum ? We can easily understand that my T(0) channel would operate as an X channel of the classic IB device . It would be because we have only one sine wave on the correlator input , and the reference , although it's a square wave here , is in phase with the transmitter wave . So we'll obtain here nothing but the same X signal on the T(0) demodulation output . And what about the next term - T(1) ? This demodulator does have the reference wave ( triangle ) , being the integral of the T(0) channel ( square ) .... so we'll have the same operation as we have in the classic IB . The first T(0) channel will take all the "sine" signal component and give us a classic "X signal" , and the second T(1) channel will take all the "cosine" compopnent of the signal , producing the classic "R signal" . And now we can understand an interesting thing ( the same to that I did before with PI principle ) - that we can "cut" the circuit of my VERSION 3.0 device , and obtain the device that very similar to the classic IB - we really can get all the information from the target that classic IB can get . And we can claim that all IB devices are the "subset" of my "square wave" principle . Simply look to the target signal expansion terms - classic IB uses only T(0) and T(1) terms , classic PI uses T(1) , T(2) and T(3) , but my device can use all of them - it can consume MORE information at once . So we can make one circuit , and then use it at the same time like IB metal detector , or PI metal detector , or combine all information in one algorithm - we can do what we want with this information . And this is why I call this circuit an "ultimate metal detector" - because it can do the things that can do all kinds of detectors , but in one circuit . And I think that it's good , of course

[Davor]

I must congratulate you for perfect reasoning. You managed to overturn several PI and IB myths in a single post. That said, I must confess I'm still a CW IB junky.

So how exactly does the IB CW GB compare with PI gb? They both create a detection hole. But unlike PI detectors, CW IB hole due to GB resides exclusively on the ferrous side.

Typical CW IB gold detectors (like Goldbug) are very much like PI detectors as they concentrate on "R" component, and don't bother with discrimination that much. They could discriminate, but the traditional CW IB detector has a hole due to the discrimination around wet salt response, which is too close to small gold response, so again - why bother with discrimination. IMHO there is a way to get discrimination here without compromising detection.

Typical PI is deaf for poor conductors, as it requires some guard time for it's input circuitry to start signal processing. It may be beneficial on wet salty sand, but not elsewhere. In cases where you need GB, you insert two detection holes in your response, one at ferrous, and the other at non-ferrous quadrants, as the PI system is non-quadrature.

Your approach may bridge these two generic approaches together. I'm looking forward to the future developments.