External Shield?

[Nova]

G'day all, i am a very curious person by nature and get a lot of ideas that pop into my head and about 2 months ago the idea popped into my head about " would it be possible to add an external shield to my mono coils for my gpx5000" as in some areas the emi is shocking and for me to run the machine at all i must severly dumb down the settings, which in the end is pointless going over ground in weak settings that newer tech has also hit.
So, i consulted AI and it said that it could be done and it explained to me how to do it, i bought MG Carbon paint made 2 outer plastic covers made from poly carbonate plastic to fit over my Coiltek 11in mono and my 14in Detech mono coils, i made a C shape with about 10mm spacing to break the carbon curcuit added a stainless steel bolt to attach the drain wire and have made sure the carbon registers between 1.000 to 5.000 k/ohmns, the AI told me i could attach the drain wire to the armwrest, but i put this past Carl in a private message and he told me i would have to join the wire onto pin 1 or 5 to get into the curcuit.
But , he also told me that if the shield has been done properly in the actual coil i PROBABLY wouldn't get any benefit at all, the question i really want an answer to is have any of you actually made one yourselves and if so did you get any benefit out of it at all?, Carl said he has never actually tried adding an external coil shield but his knowledge of the science in general tells him that there probably won't be much if any benefit at all, the GPX 5000 if the emi can be greatly reduced will match or even exceed the ZVT tech in many instances, so what say you?

[Op04]

Hi,
I think you need this one

[Nova]

I used that exact paint, doesn't answer any of my questions though, but i believe i have found an answer to at least one question, "Magnetic EMI" these shields are practically useless, again according to AI , this is the EMI i'm having problems with, when i put my Audio into deep i get a rapid on and off pulse especially higher up on the slope of the goldfields, i'm not concerned about the occasional spike but i might just have a solution already, something that i have already made, i just need to test thoroughly , again AI says it should help to block this type of EMI, and i checked to see if the info it was giving me was documented in general electrical engineering and it is, so, i will probably not worry about this type of shield as the coil already has a similar one inside, i need a different route, so i will test and test.....

[Op04]

Wow ... my memory that can needs at least 3 layer,, need at least 80 ohm, how many layer you make?

[Carl-NC]

An electromagnetic wave consists of an electric field and a magnetic field at the same time. Coil shielding is only designed to block the electric field component. If you block the magnetic field component, the coil won't work. If you are in an area where magnetic field noise is severe, the only solution is to use a noise-cancelling coil such as a figure-8, DD/butterfly, DOD, or tophat.

[Tinkerer]

This is one of the most genuinely misunderstood topics in PI design, and the confusion usually comes from treating "shielding" as one thing when it's really about selectively blocking electric fields while letting the magnetic field pass completely untouched. Let me lay out the physics first, because once the core principle is clear, every design rule falls out of it. The master principle: block E, pass H


A metal detector coil's entire reason for existing is the magnetic field. The TX field must reach the target to induce eddy currents, and the target's response field must reach the RX winding. So whatever you wrap around the coil must be transparent to the magnetic field while still intercepting electric fields. These two goals seem contradictory — a conductor blocks fields — but they're separable because of how each field is blocked:

• An electric field is stopped electrostatically. Free charges in the conductor rearrange themselves to cancel the field inside — a Faraday cage. This requires no current flow around any loop. A high-resistance, broken, or even patchy conductor still does this well.
• A magnetic field can only be opposed by a circulating current (Lenz's law). If the shield has no closed conductive loop around the coil, no circulating current can flow, so the magnetic field passes as if the shield weren't there.

That single asymmetry is the whole game: the shield must be a Faraday shield that is electrically broken so it cannot act as a shorted turn. Everything else is detail. Factor 1 — EMI (external E-field into the front end)


Your RX front end is a high-impedance, high-gain node — a preamp at a few hundred to ~1000× staring at microvolt-scale decay tails. Any external electric field (50 Hz mains and its harmonics, broadcast RF, switching-supply hash, atmospheric noise) capacitively couples onto the windings and gets amplified right along with the signal. A grounded Faraday shield around the windings intercepts that displacement current and shunts it to the shield reference before it ever reaches the wire.

Worth being explicit: the shield does essentially nothing against magnetic EMI (power-line H-fields, etc.) — but you couldn't shield that with thin non-ferrous material anyway, and blocking it would mean blocking your own signal. So the shield's EMI job is purely electric-field, and that's exactly the component you can and want to kill. Factor 2 — Capacitive coupling to ground


This is the subtler one and it's specific to sweeping a coil near earth. The windings sit at an AC potential during operation and form a capacitor with whatever is underneath — soil, wet grass, dew, rain film, saltwater, your own leg. As you sweep, that capacitance changes: distance to ground varies, moisture varies, mineral conductivity varies. Changing capacitance means changing displacement current, and that current is indistinguishable from a target response — it shows up as falsing, drift, and the classic "goes crazy in wet grass / morning dew / rain" behavior. Saltwater is the extreme case.

The shield fixes this by interposing a fixed capacitance. Once a grounded shield sits between windings and earth, the winding-to-shield capacitance is constant (fixed geometry, fixed potting), and that's what the coil now sees. The shield-to-earth capacitance still varies with wetness — but because the shield is a low-impedance ground reference, that varying current drains straight to ground instead of modulating the signal node. The coil's capacitive environment goes from "whatever the ground is doing" to "a stable, known value." Why you must break the loop (and why PI cares more than VLF)


Here's where most homebrew shields go wrong. If you wrap the coil in a continuous conductor that closes on itself around the coil's circumference, you've built a shorted single turn coaxial with your coil. During the TX pulse that loop gets induced, and after turn-off it carries its own decaying eddy current with time constant τ = L/R. Three bad things follow: it loads and distorts the TX, it attenuates the field reaching the target, and — worst for PI — the shield itself becomes a large, close, permanent phantom target whose decay lands right in your sampling window.

This is why PI is far less forgiving than VLF/IB here. You're working in the time domain, directly sampling decay from ~10 µs out. A shield eddy current with a meaningful τ sits in the middle of that window as a constant offset that also drifts with temperature (as the shield's resistance changes). The math is the clean way to see it:

• Graphite/carbon conductive paint, high sheet resistance — say the shield loop, if closed, would be R ≈ hundreds of Ω, L ≈ order of µH → τ = L/R ≈ tens of nanoseconds. The eddy dies before your first sample. Harmless.
• Closed copper foil, R ≈ milliohms, L ≈ µH → τ ≈ milliseconds. Catastrophic — a huge, slow phantom target dominating early time.

So two independent things save you, and good designs use both: keep the resistance high (which shortens τ toward zero) and break the loop (which forbids the circulating current entirely, τ irrelevant). The Faraday/E-field function needs neither low resistance nor a closed loop, so you give up nothing.

Let me show the construction topology, since the "broken loop" is the part that's hard to picture from words:



V

visualize

V

visualize show_widget

The green E-field arrows terminate on the shield and drain to ground; the purple H-field passes straight through the plane of the coil because the shield's gap forbids any circulating current from opposing it. That single break is the whole trick. Design best practices


Break the loop — always. For a wire bundle wound into a ring, you spiral the shield material around the cross-section of the bundle as you go around the coil. Where the wrap comes back to its starting point, leave a few-millimetre gap and do not join the two ends. Some builders bridge the gap with a small insulated overlap (a flap of shield material with a dielectric layer between) so E-field coverage stays continuous across the gap while DC/eddy continuity is still broken — but a plain gap works fine and is simpler.

Favour high sheet resistance. Graphite/carbon-loaded conductive paint is the classic PI choice precisely because its high resistance drives the shield's time constant into the nanosecond range, so any residual eddy dies before your first sample. Nickel/silver conductive paints are lower-resistance and give better E-field attenuation but bring you closer to phantom-target territory — use them only if you keep the loop rigorously broken. Solid foil (aluminium/copper) gives the best E-shielding but is the most dangerous: a closed foil loop is a milliohm shorted turn with a millisecond τ. If you use foil, the gap is non-negotiable and the overlap must be insulated.

One ground, one drain wire. Connect the shield to the analog/shield ground reference at exactly one point — a single bare drain wire pressed into the paint or soldered to the foil, carried down the coil cable's shield to the electronics ground. Multiple connection points reintroduce loops and defeat the whole exercise.

Manage the added capacitance. The shield forms a capacitor with the windings (that's the stabilising mechanism from Factor 2), but that same capacitance loads the coil — it lowers self-resonant frequency, changes TX damping/ring, and slows RX recovery. In a PI detector, where fast decay and short recovery are everything, this is a real cost. Keep the shield spaced off the windings with a potting/insulation layer, and don't over-cover: full geometric coverage for E-field integrity, but no more shield mass than you need. This is the tension to tune — more shield means quieter but slower.

Coil cable is part of the shield system. Use shielded (coax or shielded twisted pair) cable to the coil. Tie the cable shield to the coil's Faraday shield at the coil end and to ground at the electronics end, arranged so you don't create a loop between the two grounds. The cable shield does the same E-field job for the run between coil and box.

If your coil is balanced (separate TX/RX), keep the shield symmetric across the RX winding so it doesn't unbalance the null — an asymmetric shield capacitance shows up directly as a residual.

The mental model to carry away: you are building a Faraday cage that is deliberately incompetent as a transformer secondary. It must be a good electrostatic screen (full coverage, grounded) and a hopeless magnetic loop (broken, high-resistance). Every rule above is just one of those two requirements in a different costume — and in PI specifically, the magnetic-loop failure mode is the one that silently destroys early-time sensitivity, so when in doubt, add resistance and cut the loop.


Your instinct about the hand test is exactly right, and it's the more important of the two things you said. Let me take them separately because they stimulate different physics. Why the hand test is confounded


A hand couples to the coil through two independent paths, and the shield only addresses one of them:

• Capacitive/electrostatic path — the hand changes the coil-to-ground capacitance, and (if it's grounded through your body) injects mains hum and acts as an antenna for ambient E-fields. This is precisely what the Faraday shield exists to kill.
• Magnetic/eddy path — tissue is a lossy conductor (~0.5–1 S/m of saline), so the TX pulse induces genuine eddy currents in it. This is a real target response, and the shield does nothing to it by design (broken loop, magnetic field passes).

So "the detector sees the hand" is ambiguous: a perfectly shielded, perfectly functioning detector should still see a hand as a weak conductive target. You can't conclude anything about the shield from a positive result. There's a PI-specific wrinkle that cuts slightly in your favor: tissue's eddy time constant is very short (roughly τ ∝ μσL²), so at typical sample delays of tens of µs most of the eddy response has already decayed and much of what you see is the capacitive path — but that depends entirely on your sample timing, so it's not a property you can rely on across detectors. The test conflates the two paths and its result shifts with grounding, mains proximity, and delay. It's a rough go/no-go at best.

You can rescue it, though, by exploiting the one thing that separates the paths: grounding. Capacitive coupling depends strongly on whether the object is at a defined potential; eddy currents don't care at all. So compare a grounded conductor (your hand touching your body) against the same conductor isolated on a foam stick or dry glove. If the response collapses when you isolate it, you were looking at capacitive leakage. If it barely changes, it's eddy — and not a shield problem. Watching the preamp on a scope for a 50/60 Hz component during the approach tells you the same thing: mains-correlated wobble is a shield leak; a clean TX-synchronized transient is an eddy target. Why the charged-pipe test is actually the better one


This one is genuinely valid, and for the right reason: a charged plastic pipe is a dielectric carrying bound surface charge. It has no free carriers, so it sustains essentially no eddy currents — it produces a strong quasi-static E-field and nothing on the magnetic path. That's the whole point: it isolates the exact coupling the shield is supposed to block, with none of the eddy confound that ruins the hand test. A well-grounded shield terminates that field and drains it away, so the preamp should stay quiet. A reaction means E-field is reaching the windings — a gap, a coverage hole, or a bad shield ground. Null result = good shield is a clean inference here in a way it never is with the hand.

A few caveats so you read it correctly:

• It's a very stringent field (kilovolts), far larger than any ambient EMI. That's mostly a feature — it stresses the shield hard and finds marginal leaks — but don't panic over a tiny blip that a real 50 Hz E-field would never produce. A shield that stays quiet under the charged pipe is comfortably fine for ambient EMI.
• It's transient and humidity-dependent. The charge bleeds off (that crackle is it leaving), and on a humid day you can't build much. It tests the approach and any motion of the charge, not a steady state — which is fine, since your preamp responds to change anyway.
• On your differential front end it tests shield + CMRR together. With the cross-coupled ADA4807 diff topology feeding the ADS8881, a field that couples roughly equally to both inputs gets rejected by common-mode rejection even if the shield is leaking. So a null at the ADC output could be a good shield or good CMRR masking a leak. To isolate the shield specifically, probe the single-ended coil node ahead of the diff amp, or deliberately unbalance one side.
• The corona/discharge injects broadband RF alongside the electrostatic field, so a reaction might be RF rather than pure static — but that's still an E-field the shield should handle, so it's still diagnostic of a leak.

The single most diagnostic move


Whichever stimulus you use, the cleanest way to attribute a reaction to the shield is to toggle the shield ground while the source is present. Bring up the charged pipe (or the grounded hand), then lift and reconnect the shield's single-point ground and watch the delta at the preamp. If grounding the shield makes the reaction vanish, the shield is doing its job and the coupling was capacitive. If grounding it changes nothing, either the shield isn't coupling to the windings at all or the ground return is bad. That before/after subtraction removes almost all the ambiguity, because it isolates the shield's own contribution rather than asking a yes/no about a confounded stimulus.

If you want a number rather than pass/fail, drive a wire from a signal generator (or just use a mains lead as a 50 Hz E-field source) held a fixed distance from the coil, and measure the coupled amplitude at the preamp with the shield grounded versus floating. The ratio is your shielding effectiveness in dB, repeatable across builds.

One last thing neither test covers: the charged pipe says nothing about the other shield failure mode — the shield accidentally forming a closed loop and becoming a phantom eddy-current target. That one doesn't show up under a static field; you catch it by looking at the early-time baseline and its temperature drift with the shield added versus removed, swinging the bare coil in air and watching for a constant offset in the decay. So think of it as two separate audits: the charged pipe (and the ground-toggle) for E-field integrity, and the baseline/decay check for magnetic self-response.

[Nova]

3 coats yes, Carl, yes am aware that those coils can do that, but supposedly at the expense of 25 to 35% depth, not a very good trade off in my opinion, and Tinkerer just mind raped me by posting that LOL, i won't mention what i am doing, but i have tried it a little already, with seemingly little effect to the coil field, okay, fair enough the science is solid, but sometimes you can find little holes in it that can allow a little movement, and to find this out for myself i have to test this thoroughly and then know for myself !!!