FKK coils, free beach movement :)

[Davor]

First off, I'm not joking.

I've gathered some initial experience with an unshielded coil and there is nothing wrong with it. You might have seen my coil in action, links to youtube are on IGSL page, and so far I did not notice any annoyance that would convince me to dress it.

My coil is center tapped and supplied to not so perfect frontend, hence all common mode troubles are tackled with by center tap alone. I have some goals on my mind that I wish to share here with intention to gather other enthusiasts with same level of optimism. Let me summarise what I wish to achieve, so far with VLF:
- unshielded coil;
- aperiodic operation - to promote less than perfect coils made by amateurs;
- low noise - of course;
- differential operation with good CMMR.

My "true differential op amp" was a failure due to the common mode asymmetry, and in meantime I reminded myself on some real differential solutions that take care of common mode as well.

So, shall we go FKK?

[Davor]

Now a bit about the balanced differential operation. In professional sound equipment it is used to discourage hum from the mains and various crackling sounds entering the system by virtue of cancelling the common mode signal. It is justified by long cables going everywhere and meeting different pieces of equipment connected to the ground taps that are on different potentials and so fort.

So, how can metal detectors benefit from this approach? Well, there are several effects that are introducing some interferences by electric field that are normally tackled with the coil shielding, and that would put a cork to it, unless ... you want to go FKK. In fact there may be some side benefits as well, but let's just uncover one thing at a time.

My first and main opposition to shielding is the effort on applying it.

The second order reason would be the eddy currents in the very shielding that kinda spoil my worldview of the equipment measuring the eddy currents - that just doesn't compute. Especially not the tin foil. Like going fur coat on a FKK beach. Kind of like the observer effect. You know the one where the measuring equipment - the coil - adds some of its own resistance to the targets, thus influencing their discrimination.

Anyway, to go FKK I must fix the common mode signal component, and key to that is balanced operation. There are a few approaches to that.

The obvious, but not the only solution is a center tap. It divides a coil into a self-transformer with both coils having L/4 inductance and near perfect coupling. So for a differential signal the voltage is added and coils' inductances are added and doubled by their mutual inductance, while for the common mode signal it is a short with no inductance. In a perfect world it would be just perfect because any unbalance further on would be ironed out by perfect transformation ... which is slightly spoiled by the coil resistance of the real world.

A perfectly balanced Rx frontend could do that as well even without the center tap, but having the center tap is still a bonus. Such frontend would obliterate common mode signal (high CMMR) and provide low noise gain for the differential signal. Simultaneously it would provide equal impedance on both inputs for differential mode, and high or equal impedance for common mode - I prefer low but equal. Such frontend acts as if there is a center tap even without it.

Differential amplifier assembled from a single op amp and having matching values of R1=R3 and R2=R4 is not balanced. It has high CMMR and high and equal input impedance for common mode, but for differential mode it's inputs impedances are very unequal. Without a center tap this configuration causes the coil to float wildly with the output swing. And it is noisy too. It doesn't make much sense as a frontend.

An instrumentation amplifier is an option, but usually a noisy one, and it costs extra. Using shunt resistors the input impedance can be low and equal for differential, and either low or high for common mode, low being a better choice. We've seen some implementations of instrumentation amplifier on this forum, and apart from noise there is nothing wrong with this approach. Say AD620 with 9nV/sqrt(Hz).

A better option is an integrated "audio preamplifier" which is very similar to an instrumentation amplifier, yet with much lower voltage noise and optimised for low input impedance. Think of SSM2019, and if picky THAT1510. They both go down to 1nV/sqrt(Hz) and both have true differential inputs.

My choice would be a semi-discrete solution as per the Graeme Cohen's preamp that employs current-feedback instrumentation amplifier (CFIA). Much of it is in THAT1510, yet I can find all the components I need in just about any shop, while THAT1510 is complicated to come by. I have it running in a LTspice already, but I'd like to make a few more touches before posting it here. It easily reaches below 2nV/sqrt(Hz) in a configuration with a typical MD coil, and provides perfectly balanced input. Not bad for ~4 bucks of parts.

[Tinkerer]

Now a bit about the balanced differential operation. In professional sound equipment it is used to discourage hum from the mains and various crackling sounds entering the system by virtue of cancelling the common mode signal. It is justified by long cables going everywhere and meeting different pieces of equipment connected to the ground taps that are on different potentials and so fort.

So, how can metal detectors benefit from this approach? Well, there are several effects that are introducing some interferences by electric field that are normally tackled with the coil shielding, and that would put a cork to it, unless ... you want to go FKK. In fact there may be some side benefits as well, but let's just uncover one thing at a time.

My first and main opposition to shielding is the effort on applying it.

The second order reason would be the eddy currents in the very shielding that kinda spoil my worldview of the equipment measuring the eddy currents - that just doesn't compute. Especially not the tin foil. Like going fur coat on a FKK beach. Kind of like the observer effect. You know the one where the measuring equipment - the coil - adds some of its own resistance to the targets, thus influencing their discrimination.

Anyway, to go FKK I must fix the common mode signal component, and key to that is balanced operation. There are a few approaches to that.

The obvious, but not the only solution is a center tap. It divides a coil into a self-transformer with both coils having L/4 inductance and near perfect coupling. So for a differential signal the voltage is added and coils' inductances are added and doubled by their mutual inductance, while for the common mode signal it is a short with no inductance. In a perfect world it would be just perfect because any unbalance further on would be ironed out by perfect transformation ... which is slightly spoiled by the coil resistance of the real world.

A perfectly balanced Rx frontend could do that as well even without the center tap, but having the center tap is still a bonus. Such frontend would obliterate common mode signal (high CMMR) and provide low noise gain for the differential signal. Simultaneously it would provide equal impedance on both inputs for differential mode, and high or equal impedance for common mode - I prefer low but equal. Such frontend acts as if there is a center tap even without it.

Differential amplifier assembled from a single op amp and having matching values of R1=R3 and R2=R4 is not balanced. It has high CMMR and high and equal input impedance for common mode, but for differential mode it's inputs impedances are very unequal. Without a center tap this configuration causes the coil to float wildly with the output swing. And it is noisy too. It doesn't make much sense as a frontend.

An instrumentation amplifier is an option, but usually a noisy one, and it costs extra. Using shunt resistors the input impedance can be low and equal for differential, and either low or high for common mode, low being a better choice. We've seen some implementations of instrumentation amplifier on this forum, and apart from noise there is nothing wrong with this approach. Say AD620 with 9nV/sqrt(Hz).

A better option is an integrated "audio preamplifier" which is very similar to an instrumentation amplifier, yet with much lower voltage noise and optimised for low input impedance. Think of SSM2019, and if picky THAT1510. They both go down to 1nV/sqrt(Hz) and both have true differential inputs.

My choice would be a semi-discrete solution as per the Graeme Cohen's preamp that employs current-feedback instrumentation amplifier (CFIA). Much of it is in THAT1510, yet I can find all the components I need in just about any shop, while THAT1510 is complicated to come by. I have it running in a LTspice already, but I'd like to make a few more touches before posting it here. It easily reaches below 2nV/sqrt(Hz) in a configuration with a typical MD coil, and provides perfectly balanced input. Not bad for ~4 bucks of parts.

It will be interesting to see your circuit.
Environmental noise is ever present. Any way to reduce it is a good way.

Coil building and shielding is an art. It can get extremely complicated, but does not necessarily need to be. Good results can be achieved with simple methods if one understands the basic criteria.

Tinkerer

[Davor]

Interesting - yes. There are some conflicting requirements to reconcile before committing to some design, especially if the path was never trodden before. Of course, I'll have to test it before, which I'm not quite able on my li'l island as yet, but in a week or two I will.
If you care to try it yourself, and have the necessary supplies - be my guest. Of all people you surely do understand a good preamp in a well thought of project. There are great similarities with your TEM solution pre.
Here it goes, with values as for IGSL with Musketeer coils, operating frequency is ~8kHz and the Rx coil is 15mH. The upper circuit is a standard IGSL pre designed as a differential amp and in semi-resonant operation. The lower one is a current state of my design, CFIA with garden variety transistors. Please note that you may use much worse opamps than the 5532s shown here with marginally worse results. When normalised for gain, I get ~1.4nV/sqrt(Hz) input equivalent. It could get lower with a coil that has lower resistance. I pumped up the gain to compare gains at the similar noise levels. CFIA supplies ~11.5dB more gain, and a bit less noise. In LTspice, of course. Must try the real thing.

Try playing with this model. V1 is a differential exciter, V2 is a common mode exciter. The IGSL equivalent is provided with values ready for worst case analysis, just remove the asterisk in front of the .step param run...

Just ask

[moodz]

.... no one listens on this forum when you try to tell them the advantage of differential coils / amps / EMI rejection ... you name it LOL




This amp has a gain of around 200 with a real bandwidth of at least 1 MHz and fairly good phase response. Noise is around 1nvrootHz with matched, thermally coupled Jfets for J1 and J3.

moodz.

[simonbaker]

First off, I'm not joking.

I've gathered some initial experience with an unshielded coil and there is nothing wrong with it. You might have seen my coil in action, links to youtube are on IGSL page, and so far I did not notice any annoyance that would convince me to dress it.

My coil is center tapped and supplied to not so perfect frontend, hence all common mode troubles are tackled with by center tap alone. I have some goals on my mind that I wish to share here with intention to gather other enthusiasts with same level of optimism. Let me summarise what I wish to achieve, so far with VLF:
- unshielded coil;
- aperiodic operation - to promote less than perfect coils made by amateurs;
- low noise - of course;
- differential operation with good CMMR.

My "true differential op amp" was a failure due to the common mode asymmetry, and in meantime I reminded myself on some real differential solutions that take care of common mode as well.

So, shall we go FKK?

Sounds fine to try, see how it works.

I've asked in several threads about exactly what shield is doing. Some people mention EMI suppression but I don't understand that because seems you'd also shield target signal same and not improve S/N -- also, some commercial designs seem to use "paint-on" shields where are fairly high resistance and wouldn't shield EMI. Wet grass mentioned -- perhaps center-tapped coil will handle that. Some people mention ground "capacitance" or something, not sure what that effect is. People swear by experience shield is necessary, and I'm sure many of those coil designs were center-tapped. But let's find out more by trying.

-SB

[Qiaozhi]

Sounds fine to try, see how it works.

I've asked in several threads about exactly what shield is doing. Some people mention EMI suppression but I don't understand that because seems you'd also shield target signal same and not improve S/N -- also, some commercial designs seem to use "paint-on" shields where are fairly high resistance and wouldn't shield EMI. Wet grass mentioned -- perhaps center-tapped coil will handle that. Some people mention ground "capacitance" or something, not sure what that effect is. People swear by experience shield is necessary, and I'm sure many of those coil designs were center-tapped. But let's find out more by trying.

-SB

The coil shield is also known as a Faraday Shield. Although it can block electromagnetic interference to some extent, this is not the main purpose. It is there to block external electric fields (both static and non-static). However, it does not block relatively slowly changing magnetic fields; which is lucky because that's what a metal detector needs to be able to do its job. Any electric field that builds up around the coil, as it is moved over the ground, will be distributed throughout the shield material, and this cancels the effect of the field within the shield's interior. Hence, since the coil is within the shield, it is not affected by the external electric field. The reference to a capacitor has to do with the fact that the shield acts like one of the capacitor plates, whereas the ground is the other plate. Without a shield, the coil may also react when you bring your hand near the coil. In this case, the hand and the coil itself form a capacitor.

[Davor]

This amp has a gain of around 200 with a real bandwidth of at least 1 MHz and fairly good phase response. Noise is around 1nvrootHz with matched, thermally coupled Jfets for J1 and J3.

In reality it would not have that little noise with any realistic coil because of the coil resistance, and that is the ultimate limit of low noise design here. Otherwise, to reach the coil noise, you must match it to an equally low noise preamp, and that would be it. To reach any better you'd have to use thicker wire for your Rx coils.
This way or another, such low noise design will give you extra 10dB or so of system gain, and you'll reach maybe 25% deeper. Not bad at all for 3 bucks worth electronic parts.

...some commercial designs seem to use "paint-on" shields where are fairly high resistance and wouldn't shield EMI...

But such solutions do provide a capacitive screen against any common mode electric field. Air resistance is measured in teraohms, so every paint-on shield with several hundred kohms will do, and on the plus it would not produce any measurable eddy currents at any meaningful angle, while tin foil produces a lot of eddy current responses in ... tin foil and small gold angle range. Not quite good news for gold prospectors.

Without a shield, the coil may also react when you bring your hand near the coil. In this case, the hand and the coil itself form a capacitor.

Well, mine doesn't, and it is not quite optimal at the moment - my preamp I currently use sucks. Winding coils bifilarly with a center tap reduces the most of such effects significantly. With proper preamp that provides proper loading for both common mode and differential signal, and with considerable CMMR, will surely outperform any shielded coil for a simple reason of having no parasitic matters in close proximity.


Yet another thought - Tx coil is usually not balanced, yet it provides some swing against the system ground. That may account for the residual imbalance that you simply can't overcome by geometry alone. Besides, a second harmonic generated by asymmetric oscillator is a door for entering various problems of the PWM kind to your detector. I'll give this some thoughts.

[simonbaker]

The coil shield is also known as a Faraday Shield. Although it can block electromagnetic interference to some extent, this is not the main purpose. It is there to block external electric fields (both static and non-static). However, it does not block relatively slowly changing magnetic fields; which is lucky because that's what a metal detector needs to be able to do its job. Any electric field that builds up around the coil, as it is moved over the ground, will be distributed throughout the shield material, and this cancels the effect of the field within the shield's interior. Hence, since the coil is within the shield, it is not affected by the external electric field. The reference to a capacitor has to do with the fact that the shield acts like one of the capacitor plates, whereas the ground is the other plate. Without a shield, the coil may also react when you bring your hand near the coil. In this case, the hand and the coil itself form a capacitor.

Right -- I just don't see how electric charge slowly building up on the coil being a problem unless there is a sudden discharge -- which then would seem a little hazardous to the circuitry, although people have claimed that op amps are protected against a big common mode spike. So if big discharges are possible in some circumstances, I'd think a shield would be recommended, because even a center-tap ground probably wouldn't help due to the high frequency of the discharge.

-SB

[simonbaker]

But such solutions do provide a capacitive screen against any common mode electric field.

Exactly -- but not EMI, so serves as some confirmation that shields are not for EMI, along with Qiaohzi's statements.

What do you think about possibility of large static discharge to coil?

-SB

[Davor]

You glow blue for a very short period, and after that your hair style becomes entirely different
There is a large static tension at about 10 000V/m on nice weather, but then again, air resistance is also very high, so don't worry. Problems with static electricity are related to the phenomena where that natural tension powers up effects similar to the condenser microphone against the surface, e.g. grass. So you swing your coil, and by virtue of distributed capacitance and a large electric field, the roughness of grass cover provides your coil with some common mode voltage.
It is not much of a question what to do IF something like that happens, because it just does. Much better question is how (else) we can deal with that, and why is such approach good, and even better - what other choices we may apply.

Regarding EMI, true, shielding will reduce influence of the E field component, but not the H component. For a vertically polarised signal our coils are just perfectly aligned with the axis perpendicular to the ground. Hence, for far field sources of interference the metal detector coil is a perfect antenna - shield or no shield. That kind of EMI can be cancelled by using differential coils, just like humbucker pickups do, however, such coils would require a 4 quadrant discrimination to work properly ... it is another story.

[simonbaker]

You glow blue for a very short period, and after that your hair style becomes entirely different
There is a large static tension at about 10 000V/m on nice weather, but then again, air resistance is also very high, so don't worry. Problems with static electricity are related to the phenomena where that natural tension powers up effects similar to the condenser microphone against the surface, e.g. grass. So you swing your coil, and by virtue of distributed capacitance and a large electric field, the roughness of grass cover provides your coil with some common mode voltage.
It is not much of a question what to do IF something like that happens, because it just does. Much better question is how (else) we can deal with that, and why is such approach good, and even better - what other choices we may apply.

Regarding EMI, true, shielding will reduce influence of the E field component, but not the H component. For a vertically polarised signal our coils are just perfectly aligned with the axis perpendicular to the ground. Hence, for far field sources of interference the metal detector coil is a perfect antenna - shield or no shield. That kind of EMI can be cancelled by using differential coils, just like humbucker pickups do, however, such coils would require a 4 quadrant discrimination to work properly ... it is another story.

If the common mode voltage due to E-field is not varying too fast, then I'd agree the center tap ground of the coil might be all you need for that. Even the lopsided RX grounding of the original TGSL coil should keep the coil at PCB ground for slowly varying E-fields.

And I agree totally with your antenna description -- noise with that polarization is received just like the target signal, shield irrelevant.

But is there more to shields than that? What kind of shields do the latest Tesoro coils use, anyone? Wouldn't MD makers abandon shields if they added nothing? Or have they?

-SB

[Davor]

I have no idea about Tesoro. I've seen a short video showing their facilities and coil production. Their coil winding machine was making a monofilar coil using a self-adhesive wire so I'm pretty sure that particular coil was not center tapped.
It is interesting to see how mutual inductance in case of a center tap actually cancels the coil inductance for a common mode signal. With good coupling you are left with much less than 1% of the normal coil inductance, and a half of the coil resistance. In effect the stray capacitances with that half resistance form a high pass filter, so your observation of slow E field contributions is correct.

I'd say that a center tap is a cornerstone of any serious FKK Rx coil. It may not be perfect due to the coil resistance, but it works, and it is not prone to components tolerance problems.

[simonbaker]

It is interesting to see how mutual inductance in case of a center tap actually cancels the coil inductance for a common mode signal.
With good coupling you are left with much less than 1% of the normal coil inductance, and a half of the coil resistance. In effect the stray capacitances with that half resistance form a high pass filter, so your observation of slow E field contributions is correct.

That's an interesting point I hadn't considered, so actually you're saying that high frequency common mode signals will be effectively grounded also (in addition to low freq signals) because there is no net inductance blocking the path to ground. You gave me credit where I didn't deserve it.

However, I could imagine a case where one side of the coil has a static discharge (high freq), and might not be effectively grounded. Perhaps that never happens, or perhaps the mutual inductance coupling again creates a low inductance path to PCB ground.

-SB

[Jerry]

Well, mine doesn't, and it is not quite optimal at the moment - my preamp I currently use sucks. Winding coils bifilarly with a center tap reduces the most of such effects significantly.

Hi Simon,

Could you elaborate a little on the actual construction & winding of your coil I understand what bifilar winding are but do not understand the arrangement of the center tap or if the individual wire strands are twisted together or not prior to winding.

I picture two wires wound parallel and end c. and d. connected to make the center tap and ends a. and b. being connected to the cable to the detector.

a. >-------------------------------------------< c.
b. >-------------------------------------------< d.

Jerry