mono coil IB

[deemon]

Some time ago , when I was developing my PI metal detector , I suddenly found one idea that seemed to be very useful , not in PI but in IB metal detectors , and now I want to talk about it . I mean inductive balance with a single coil . Of course , everybody knows what a difficult task is to make a good and reliable search coil for IB device . We need to wind 2 coils with special shape , place them precisely and so on ... and sometimes a bad thing can happen - when we finish a coil and find that balance is bad And many times I thought how can I separate transmit and receive signals using only one coil ? Of course there are some well-known solutions like Wheatstone bridge , for instance .... but the same problem with them is need to simultaneously balance two parameters - coil inductance and coil wire resistance . Inductance is not a big problem , but resistance of a copper wire is thermal dependent parameter , and balance will go away with every temperature change . Another thing that I don't like in those circuits is need to use several precision components in the bridge .

But the circuit that I talking about doesn't have this problem and also is extremely simple . What we need is to wind a usual round coil , the same kind that we use in well-known PI devices , but with 2 bifilar wires ( thin and thick ) and connect them to the simple transformer wound on the little bobbin - we can place in near the printed circuit board of our MD . As we can see on the picture , balance equation does contain 3 inductances but no resistors at all , so we need to balance only one parameter and doesn't have a problem with thermal drift of the wire resistance . So I made some experiments with a real coils and found that this circuit really works and gives a good balance . In order to simplify a balance procedure I used a little and thin ferrite stick that I move into the compensation transformer bobbin . By the way , we can use this circuit in the multiple-frequency IB detectors or even in PI-IB devices . As for me , I have a plan to make a multi-frequency IB device with very interesting algorithm , using this balance circuit . But it needs some time to think about ...

[Carl-NC]

Years ago White's made a "Coinmaster TR" which had a non-IB TR coil. I believe the null was provided in a manner similar to your Basic Variant. I've thought about doing a single coil TR but haven't worked up a solution, I really like your Simple Coil Variant approach.

[Aziz]

Hi deemon,



indeed, some really good fresh ideas. I like it too. Thanks for sharing your ideas.
Cheers,
Aziz

[deemon]

Thank you , Carl , it's very interesting info . When I made my experiments , I tried to find something like this , in order to compare the efficiency , but cannot find anything . But on the other hand , this solution is so simple that it's hard to believe that nobody tried this before Anyhow , it would be interesting to find the circuit of this White's device ....

[Davor]

There is a point for consideration, but perhaps of lesser importance: ferrite tends to produce harmonics at ~-50dBc. In case sampling is done with perfect symmetry it is not that much of an issue at all because 2nd harmonic is cancelled.

I like these ideas. In fact I was considering making impedance transformers for adjusting Tesoro to Musketeer coils etc, but this idea of yours is way better than that.

[deemon]

Yes , I tried this circuit on ferrite cores too , and it worked worse than on the usual "air" coils . And what I noticed - if I use a search coil ( L1 and L2 ) on the ferrite , then it's better to wind L3 and L4 on the ferrite core too , and to use the same kind of ferrite . When I did it - I found some harmonics compensation , but anyhow - it wasn't good . By the way , the dominating harmonic frequency was not the 2-nd but the 3-rd .... And another fault - the balance with ferrite cores was quite unstable and had some drift due to transmit amplitude changes , temperature and external magnetic fields .

So the best way to use these circuits is to wind all the coils without any ferromagnetics inside , except only the very small ferrite rod that we can use for fine tuning L3 and L4 coils when approaching the balance point . In my experiments I wound L3 and L4 on the simple plastic or wooden bobbin . And L1 and L2 I used to wind just the same way that in usual PI devices does , but with bifilar instead of single wire . And with this kind of the coils I had very good balance , of course .

[Carl-NC]

I was thinking about the basic variant, and whether L3 & L4 could be placed inside the coil housing. But they would probably need to be placed orthogonal to L1/L2 to avoid coupling. Then that brought back memories of the old Daytona humpback coil, so I pulled up the patent and that's exactly what they were doing:

http://www.geotech1.com/pages/metdet.../US4345208.pdf

[deemon]

I was thinking about the basic variant, and whether L3 & L4 could be placed inside the coil housing. But they would probably need to be placed orthogonal to L1/L2 to avoid coupling. Then that brought back memories of the old Daytona humpback coil, so I pulled up the patent and that's exactly what they were doing:

http://www.geotech1.com/pages/metdet.../US4345208.pdf

Looks interesting .... but they came to exactly what I wanna go away - very complicated construction of the search coil My idea was quite opposite - to simplify the coil itself and to transfer all the balance functions to another independent part ( compensation transformer ) that can be adjusted when the coil is just finished and working . And I decided to place this transformer not in the coil case but in the electronic block , closer to the PCB - because the compensation transformer has high output impedance due to relatively high inductance of L4 , so the cable capacitance may cause an undesired receiver signal phase shift . And of course , if we place this trans near the coil - we must prevent induction from the coil field ....

But what I think - it's a good way to cancel any magnetic influence to our compensation trans ( L3 , L4 ) , if we wind it on the torus core . Torus transformers has a little leackage field , this is why them being used in the audio , for instance . So if we take a little plastic torus ( a few centimeters diameter ) and wind our L3 and L4 coils on it - we'll be able to place it even in the center of the search coil , and the coil field will be compensated and not affect the balance .

[Tepco]

Looks interesting .... but they came to exactly what I wanna go away - very complicated construction of the search coil My idea was quite opposite - to simplify the coil itself and to transfer all the balance functions to another independent part ( compensation transformer ) that can be adjusted when the coil is just finished and working . And I decided to place this transformer not in the coil case but in the electronic block , closer to the PCB - because the compensation transformer has high output impedance due to relatively high inductance of L4 , so the cable capacitance may cause an undesired receiver signal phase shift . And of course , if we place this trans near the coil - we must prevent induction from the coil field ....

But what I think - it's a good way to cancel any magnetic influence to our compensation trans ( L3 , L4 ) , if we wind it on the torus core . Torus transformers has a little leackage field , this is why them being used in the audio , for instance . So if we take a little plastic torus ( a few centimeters diameter ) and wind our L3 and L4 coils on it - we'll be able to place it even in the center of the search coil , and the coil field will be compensated and not affect the balance .

Very nice idea, monocoil IB , only, this is eternal source of trouble.
Just some time ago, one (well known, from MD book) Russian monocoil IB is mentioned on this forum:
http://www.geotech1.com/forums/showthread.php?t=19313 - post151642
applying opposite amplitude and phase to substract TX component.

I want to make some useful input here, regarding your circuit and add few more comments.

First, on your “basic variant” circuit, searchcoil is bifilar, so L1 and L2 are almost 100% mutually coupled having same inductance, and balance equation says, in essence, L3 and L4 must be the same value, L1*L2= L3*L4, (or in practice, inductances of coil and transformer should be about the same, but L1\L2 and L3\L4 well matched ) Then, L1\L3 ratio is stated to be 10, what of this is correct? Also why need to use different wire diameters on bifilar L1\L2 coil? Source TX and appropriate RX impedances must be taken into account too.

Whatever you do with standard monocoil (not bifilar), will end up in some sort of bridge-balanced circuit, one arm of bridge being searchcoil. (like your “simple coil variant”) This arrangement is to some degree similar to circuit commonly called “line hybrid (in telephone)” or “4 to 2 wire converter”, for TX cancellation but with two major differences:

• In order to get close to even questionably balanced VLF coil, you need to achieve something like 60dB hybrid (RX\TX) isolation, more if possible, and maintain it with reasonable stability.

• line hybrid usually works with resistive, or somewhat reactive load (then balancing network needs to have same characteristic), you use almost purely inductive load, producing almost 90deg phase shift, so you need to compensate. Typical problem with bridge circuits of any kind, one arm must be matched with exactly identical impedance in another to achieve balance. Chance that you can make circuit (using some small inductor, gyrator or whatever) to exactly mach searchcoil parameters and retain good enough stability is slim. All this can be done purely solid state, without transformer, using pure resistance in one bridge arm, and you can provide artificial phase shift to compensate, then you will need separate amplitude and phase balance adjustments.

End result is, you can replace most precise and highest stability part of MD, balanced coil system easy to build, with monocoil and some electronics with questionable stability hardly to ever mach mechanical design. Even tuned operation is problematic this way, put aside multifrequency or square wave. Worth trying, but maybe not way to go.

[Davor]

Just some time ago, one (well known, from MD book) Russian monocoil IB is mentioned on this forum:
http://www.geotech1.com/forums/showthread.php?t=19313 - post151642
applying opposite amplitude and phase to substract TX component.

It was not the same. Tx signal was extracted from pre-PA stage there, which left the PA free to generate all-you-can-eat kinds of trouble. This is different.

[deemon]

First, on your “basic variant” circuit, searchcoil is bifilar, so L1 and L2 are almost 100% mutually coupled having same inductance, and balance equation says, in essence, L3 and L4 must be the same value, L1*L2= L3*L4, (or in practice, inductances of coil and transformer should be about the same, but L1\L2 and L3\L4 well matched ) Then, L1\L3 ratio is stated to be 10, what of this is correct? Also why need to use different wire diameters on bifilar L1\L2 coil? Source TX and appropriate RX impedances must be taken into account too.

Balance equation is calculated in supposition that L1=L2 , because of bifilar winding . First we know L1 value ( for example 1,0 mH ) , then set L3 arbitrarily .... let it be 0,5 mH , and after it we can know L4 - 2,0 mH ( 1,0^2=0,5*2,0 ) . If we set L3=0,01 mH , we shall get L4=100 mH , and it will balance too , of course . But you see - in the first case we'll get too much voltage loss on L3 , and in the second case it will be too much L4 inductance ( and too much wire capacitance ) .... and this is why I set the preliminary condition L1/L3 = approximately 10 . But of course it can be 8 or 13 and so on .... And if we have L1=1,0 mH and set optimum L3 value is 1,0/10=0,1 mH , then L4 must be 10,0 mH for the balance .

About the wires - transmit current goes through L3 and L1 , so it's better to use thick wire for them , of course .

[Carl-NC]

Looks interesting .... but they came to exactly what I wanna go away - very complicated construction of the search coil .

I agree, it looks difficult to build. I think Wilson-Neuman must have also agreed, as they abandoned the humpback coil in later versions of the Daytona.

[Aziz]

Hi all,

the new ideas look quite easy to build. But they aren't IMHO! They can't achieve the mechanical induction balance quality. And they are limitted to few applications only (simple TR, single frequency VLF).

Sorry, but I'm not convinced at all after doing some spice simulations. But I do not want to prevent & discourage you to try it out. Please report your experience here.

Cheers,
Aziz

[ivconic]

Just a suggestion, don't throw stones on me now!

[deemon]

Whatever you do with standard monocoil (not bifilar), will end up in some sort of bridge-balanced circuit, one arm of bridge being searchcoil. (like your “simple coil variant”) This arrangement is to some degree similar to circuit commonly called “line hybrid (in telephone)” or “4 to 2 wire converter”, for TX cancellation but with two major differences:

• In order to get close to even questionably balanced VLF coil, you need to achieve something like 60dB hybrid (RX\TX) isolation, more if possible, and maintain it with reasonable stability.

• line hybrid usually works with resistive, or somewhat reactive load (then balancing network needs to have same characteristic), you use almost purely inductive load, producing almost 90deg phase shift, so you need to compensate. Typical problem with bridge circuits of any kind, one arm must be matched with exactly identical impedance in another to achieve balance. Chance that you can make circuit (using some small inductor, gyrator or whatever) to exactly mach searchcoil parameters and retain good enough stability is slim. All this can be done purely solid state, without transformer, using pure resistance in one bridge arm, and you can provide artificial phase shift to compensate, then you will need separate amplitude and phase balance adjustments.

End result is, you can replace most precise and highest stability part of MD, balanced coil system easy to build, with monocoil and some electronics with questionable stability hardly to ever mach mechanical design. Even tuned operation is problematic this way, put aside multifrequency or square wave. Worth trying, but maybe not way to go.

Yes , if we use standard monocoil ( with single wire ) - we need to compensate its inherent time constant ( L/Rwire ) , it's a fundamental requirement if we have only "electrical" access to the coil . It's just because we can't separate voltage drop on the "ideal inductance" and additional wire resistance when we measure the voltage on the coil leads . But the situation becomes different when we can feel the magnetic field of the coil .... and that is why I used this bifilar winding . This additional coil will feel the same magnetic flux as the main winding , and according to Faraday's law the voltage on this additional coil ( L2 ) is just equal to the voltage on the internal "ideal coil" without wire resistance voltage drop in the real coil . And this fact allows us to exclude the wire resistance from the balance equation in the first and second variants of the balance circuit . But in the third - simple coil variant - we cannot get this "internal voltage" , so we need to "simulate" it by RC network with the same time constant that the coil does have . Of course it's a drawback of this variant , because we need to maintain this RC time constant precisely equal to that of the coil .... but when we do it - the balance quality is just the same that in the first variant , and I tested it experimentally , it works OK .

About the principle of compensation in this class of balance circuits - it's based not on subtraction the voltage drop on the two arms of the Wheatstone bridge or hybrid network , but on subtraction derivative of the coil current function ( proportional to the coil voltage according to Faraday's law ) from the real coil voltage that we can get from L2 winding . As we can see - this compensation transformer ( L3 , L4 ) is a real differentiator . Unlike the usual transformer that has voltage input and voltage output ( or current input and current output ) - this one has current input and voltage output ( proportional to the input current derivative ) , and this output voltage we subtract from the voltage on our bifilar-wound L2 . This subtraction cancels the input transmit voltage , but external magnetic field produces the induction current in L1 coil - and this current produces ( after differentiation ) the signal voltage on the L4 , that haven't been cancelled and appears on the output . And this output signal is proportional to derivative of the external magnetic field , just on the leads of a real unloaded receive coil , so this circuit after being balanced behaves itself like balanced 2-coil system , double D for instance , but without its drawbacks like corrupted directivity diagram and phase reverse on the sides of DD coil . And it of course must be good for metal discrimination purpose , so I decided to try my balance circuit in my new IB multi-frequency metal detector .