Square wave Pulse Induction

[deemon]

Yes , Davor , the main point of mine is having more information is better anyhow . For example , imagine that we have a ground with simple ferromagnetic properties . What we'll have on a classic IB device in that case ? It gives us only 2 channels ( X and R ) , and one of them is corrupted by the ground . But this X channel is necessary for 2 device features - ferro-discrimination and target TC measuring , and we have a problems . But in this "ultimate metal detector" we does have 4 channels , and we can do a trick - use this corrupted X channel ( T(0) in my terminology ) only for ferro/non-ferro discrim ( where we need only a polarity of the voltage of this channel ) , but the TC measure can be provided in PI-like manner , by the other 3 channels of the device - T(1) , T(2) , T(3) ) - being much more tolerant to the ground influence in this case .

Another possible case , where the ground has a "magnetic viscosity" and we have a problems with the "detection hole" produced by our GB system .... but this is a problem only in the PI device . In our new wide-band system we have this T(0) channel - and it will help us to "shift" this detection hole to the ferrous side , as you noticed here about IB systems . But for metal recognizing ( gold , for instance ) - we still can use a PI information , and forget about this problem at all

[deemon]

Let's continue , guys . After that portion of the theory , we can talk today about more "practical" things - I mean the construction of the balance chain of our "ultimate metal detector" - Version 3.0 . As I explained in the post 29 , and had been shown on the picture in that post - I'm using here an "integral form" of the principle described here - http://www.geotech1.com/forums/showt...3-mono-coil-IB . A "differential form" of the principle means that I add a voltage on the search coil and the derivative of the coil current , being applied in the proper amplitude and polarity in order to obtain a perfect balance . And , as I explained in that topic , I used one interesting trick - a bifilar wound coil . I need it to discard a wire resistance from the balance equation . That balanced coil system can really replace an every kind of 2-coils balanced search heads ( DD , concentric , etc ) and works completely in the same manner , but with mono coil - so it hasn't any conventional search heads issues ( bad directivity diagram , inversion zones , etc ) . So it works just like a DD , and gives a voltage on the output .... but in this square-wave wideband approach we needs a current output from the "shorted" coil .

And this is why I used here the second version of the principle - its integral form . I mean that I compensate now the current of the coil ( carrying now the target signal ) , adding the current proportional to the integral of the voltage on the coil - to the same current transformer . By the way , both variants of this balance principle are dual to each other ... On the simplified circuit in the post 29 we can see how I did it - I applied the coil voltage to a simple Op-Amp integrator ( converting it to a current on the input resistor ) , giving us a voltage on the op-amp output , and then converted this voltage to a current on the resistor being connected to the current transformer winding . On the other winding I did connect the "cold" lead of the search coil , so both currents meets in the transformer in proper polarity and amplitude , making a balance . Now we can see this circuit , on the top of the attached picture ( I/U integrator ) - the only difference is that I used a bifilar winding on the search coil , in order to discard the coil wire influence , as I explained before . And the circuit is simplified too - I didn't show all DC stabilizing networks on the Op-Amp integrator circuits , making the whole idea more clear . By the way , R1 resistor on the first and second pictures is voltage-controlled . In reality I used a photoresistor , being illuminated by a LED - changing the LED current I changed its resistance , so I could trim the integrator gain and used it to control the balance point by a servo loop ....

I tested this circuit , it works OK , but then I noticed the one but serious drawback of the circuit - I mean its noise performance . Of course , if we have an additional op-amp in the signal path , we must expect that it might give a noise "contribution" to the output signal , at least comparable to that of the front-end preamp and I/U converter ( U2 on the attached picture ) .... but situation appeared to be even worse As we can easily see on the picture , the compensation output signal goes to the current trans via the variable resistor R2 , connected to the coil winding with the same turn number that output winding does have . So the compensation signal is "virtually connected" just on the U2 (-) input , where the negative feedback resistor R3 goes too . So , now we have an "unwanted" R-divider in the feedback loop ( R3/R2 ) , and the whole circuit begins to amplify the noise voltage of our integrator U1 ( Unoise1 on the circuit ) by the ratio of this divider , in practice it was about 10 .... of course it's bad . But we have even more trouble - the noise voltage of the U2 amp ( Unoise2 ) does have the same amplification

So I discarded this circuit and started to think how to get rid of this additional noise . The next solution that I tried was the second circuit on the picture ( U/I integrator ) . I'd never seen this circuit before , but it's a kind of a "dual" circuit to a classic op-amp integrator . Everything is "vice-versa" here - current instead of voltage , L instead of C ( and bifilar winding is in the game too ) .... but anyhow it does exactly what I need - it takes the voltage and gives the current , so I don't need to convert it on the resistor and can supply it directly to the current transformer compensation winding . The situation seems to be better here - because we don't have a resistor ( and this integrator has a current-source output property ) we don't have any amplification of the U2 noise ..... but with the U1 noise the situation is worse . Its input noise is just applied to the L2 , being converted to an output current just like a signal does , so we already have its noise amplification on the output of the circuit . It can be clear when we calculate the real values of the parts of the signal path

And after all this stuff I found that the best solution is the simple L-balance circuit , similar to that I'd shown in the "mono coil IB" topic . It doesn't have any active components in the balance chain , so it gives "near ideal" performance - and the best theoretical sensitivity . But if I use the simple L2 coil to compensate L1 search coil current - how I can electronically regulate it for my servo-loop purpose ??? In the first and second circuits I used a photoresistors like a controlled resistance ( and I cannot say that they are free of problems too ) - but what can I use here for the coil ? And then I remembered one forgotten thing , I mean a magnetic-controlled inductance . There were many kinds of this beast , used in magnetic amplifiers , power regulators and other stuff .... so I had chosen the best of them - I mean the controlled inductance with orthogonal winding on the toroidal core . The working coil is wound in a well-known manner , but the control winding is "immersed" just in the core body .... By the way , the core material isn't a ferrite , but a kind of a melted iron powder . It haven't any noise and works very good .... the only possible problem - nonlinearity due to the magnetic core saturation - doesn't make any trouble here , because my signal is a square current wave - and the working intervals are nothing but a steady current half-periods , so we don't need a very strict linearity requirements here .


To be continued ....

[Davor]

Good to see you back

[deemon]

Hi Davor

It's a pity , but I haven't enough time to finish a working device , based on this idea .
But I'll finish it , anyhow ...

[deemon]

I just sent a reply to the private message from mickstv , but received this notification

The following errors occurred with your submission

• mickstv has exceeded their stored private messages quota and cannot accept further messages until they clear some space.

Because the question was about this square wave TX circuit , I post the answer here -

Hi Dmitry, Thanks for the reply. Just one other question that you may have posted, If I connect a cro across TR1.2 (no TX transistors fitted) what would be the p-p voltage of the square wave.

This voltage is easy to calculate - the transformer divides input voltage proportionally to its turn ratio ( shown on the circuit ) . If you connect the power supply generator to +12 V battery , you'll see 12 volts (peak to peak) square wave on the primary coil of the TR1 ( TR1.1) . So the voltage on the TR1.2 ( turn ratio 1/5 ) must be 12/5=2,4 V ( p-p ) . And on TR1.3 and TR1.4 it must be 12/4=3 V ( p-p ) . The signal shape must be ideal square wave on the all windings of the transformer .

P.S. Anyhow , it's better to ask and answer the questions here - all information about the circuit may be useful for somebody else .... and the forum topic does have enough free space

[mickstv]

I just sent a reply to the private message from mickstv , but received this notification

The following errors occurred with your submission

• mickstv has exceeded their stored private messages quota and cannot accept further messages until they clear some space.

Because the question was about this square wave TX circuit , I post the answer here -



This voltage is easy to calculate - the transformer divides input voltage proportionally to its turn ratio ( shown on the circuit ) . If you connect the power supply generator to +12 V battery , you'll see 12 volts (peak to peak) square wave on the primary coil of the TR1 ( TR1.1) . So the voltage on the TR1.2 ( turn ratio 1/5 ) must be 12/5=2,4 V ( p-p ) . And on TR1.3 and TR1.4 it must be 12/4=3 V ( p-p ) . The signal shape must be ideal square wave on the all windings of the transformer .

P.S. Anyhow , it's better to ask and answer the questions here - all information about the circuit may be useful for somebody else .... and the forum topic does have enough free space

Thanks

[deemon]

By the way , as we a talking about a CC compensation feedback ( tilt control servo loop ) in the topic about GPZ 7000, I want to show how I usually do this in my CC transmitter . My idea is to measure the coil current just before the flyback interval and the current just after it , compare them ( subtract and integrate ) and perform a control action to the kick circuit ( increase or decrease the kick voltage ) , in order to maintan both currents ( before and after ) equal . So , it's a classic servo-loop . The most interesting in this circuit is a current sense method - I use here the same energy exchange mechanism as I do in the power chain . Energy that has been stored in the little probe coil - then being converted to the capacitor voltage during a quasi-resonant process ( so this voltage is directly proportional to the coil current at the moment of the power chain switch ) , and I utilize the fact that switching ON a positive current gives the same coil response that switching OFF a negative current ( and vice versa ) - the principle is shown on the pictures .

[deemon]

And this is the circuit of the real transmitter ( working and tested ) :

[BILLY]

And this is the circuit of the real transmitter ( working and tested ) :
[ATTACH]32372[/ATTACH]

That is a great picture. Lets see a picture of the actual detector along with some pics of how good it works. Thanks

[mickstv]

Hi Deemon, in post 51 you have a picture showing the discrim_idle, were you still using the current transformer method ? also just wondering if you could post the preamp part of the schematic (surrounding the current transformer) and explain how you got the output shown in the first picture on post 51.


Thanks
Mick.

[deemon]

Hi Deemon, in post 51 you have a picture showing the discrim_idle, were you still using the current transformer method ? also just wondering if you could post the preamp part of the schematic (surrounding the current transformer) and explain how you got the output shown in the first picture on post 51.


Thanks
Mick.

Of course I can , Mick - I didn't do it only because it haven't sense without a working power chain . But now , when you have the power chain working - it's a time to make a perfect balance I need a little time to draw and scan it - so tomorrow I will post the circuit ....

[mickstv]

Of course I can , Mick - I didn't do it only because it haven't sense without a working power chain . But now , when you have the power chain working - it's a time to make a perfect balance I need a little time to draw and scan it - so tomorrow I will post the circuit ....

Thanks Deemon. I'll also post up some oscilliscope pictures over the weekend.

[mickstv]

The power supply is 12v p-p generated using a high side half bridge + micro. The top waveform is TX current measured across a 0.1 ohm resistor.

[deemon]

Here is the circuit of the balance chain . The circuit has been tested and really working , all the pictures of balanced signal in this topic has been provided by this manual balance chain . As we can see , it works as I explained before , the only "strange" thing here is TR2 transformer with two variable resistors ( VR2 and VR3 ) . The function of these parts is compensation of the parasitic signal being produced by eddy currents in the search coil wire . The coil "feels itself" , as I explained before , and this is why we need to cancel this unwanted exponential decaying signal , adding the anti-phase correction signal from the TR2 transformer . Adjusting the duration ( time constant ) and amplitude of this correcting signal we are able to suppress it completely .... by the way , in the previous post by Mickstv we can see the pictures of the real signal - and a very little "spikes" on the very beginning of the every half-period of current wave ( upper trace ) is what I'm talking about .

C1 and R2 are compensating another issue in the real world - I mean a stray capacitance of the search coil winding and a parasitic conductivity of its wire isolation . Athough both factors are noticeable only in a flyback pulse intervals ( that must be blanked during the signal processing ) , being proportional to a flyback pulse voltage and its derivative - we need to suppress them too , in order to avoid the preamp overload by this high-speed signal .... so in reality we can obtain a perfect balance in a wide frequency range - from 1 kHz or even less ( square wave fundamental frequency ) up to several hundreds of KHz . And due to a symmetrical preamp configuration we can obtain the output signal either single polarity or balanced , if we need it for a further processing .

[deemon]

Just forgot to add two parts to the circuit I mean the protection diodes ( D1,D2 ) connected in parallel to TR1.3 ( current transformer output ) . Although they are not needed in a normal operation , it's better to keep them in order to prevent the op-amp inputs from burning by excessive voltage in case of preamp overload or other possible situations during our experiments ....