Pulse Induction Metal Detector

[MarkMakies]

Following up on my earlier post about a rigid coil support (pasted below from a Google archive since the original post was deleted)…

I ended up using a 600 x 500 x 12 mm clear perspex sheet and had the supplier CNC two slots, 6 mm and 8 mm wide, and 6 mm deep, to house the coils. After some research, I found a much better epoxy than what was available at the local hardware store. This one is more fluid, wets well, and self-levels.

So far, I’ve added 10 turns of magnet wire in the transmit slot and poured enough epoxy to fill it to the top of the perspex (62 mL). However, I didn’t account for the wire volume, which led to a bit of an overflow. @Robert93820, you’re absolutely right about it being final!

For the receive slot, I lined it with copper conductive tape, which has adhesive on one side. It’s a bit fiddly, but the tape is fantastic. The adhesive is strong yet easy to reposition when needed, and the foil is durable—I only had a few small tears, and that was while trying to press it into a square-edged slot. The trick is to use plastic tools. Along the 1.7m lenfth my multimeter read 0 ohms - excellent.

The next step is to wind 50 turns of wire-wrap wire into the foil-lined channel, pour the epoxy, and complete the Faraday shield. My initial plan was to fold the tape over itself, but that probably won’t work due to the epoxy and tape width.

A simpler fix might be to lay another piece of foil, adhesive side down, against the foil lip. I tested the adhesive’s conductivity and found the resistance to be quite high until I applied a lot of pressure with the probes.

Before I proceed, I need to know:

If I press firmly while sticking it down, will it still create an effective Faraday shield?

Can anyone advise on this?

So far…

BTW, it’s still a pretty heavy unit, but I should be able to remove a significant amount of material without compromising the rigidity I need.

Roverling MkII and the Pulse Induction Metal Detector are almost ready to do some autonomous metal detecting. The trailer has been redesigned three times and a new desktop GUI provides so much more real time info at 20 samples per second over a LoRa link.

But I need some help. My previous coil housings were too flimsy and any bump, shake or flexing would result in a 20-100 mV change to my timed receive acquisitions, that are accurate to about 5nS. This is really bad when I need to detect down to 1mV changes. I’ve already managed to reduced electronic noise down to 100uV, so now it’s all physical.

In order to fix this problem, the coils you see here have been carefully epoxied (lots of) onto a 6.4 mm glass plate. The transmit coil is made of 10 turns of 24 AWG enameled magnet wire, whilst the receive coil is 50 turns of 30AWG Teflon insulated silver plated copper, AKA wire wrap wire. The receive coil has a Faraday shield made from conductive aluminium tape, but will be changed to copper in the next iteration. The trailer itself is made from PVC sheeting, and forms a bash plate that the glass sits on top of.

This system works really well now, with bumps and knocks no longer impacting the samples enough to cause a problem. However the new big problem is weight. The trailer is 6.8 kg without the battery, which has now been relocated to the front of Roverling for better weight distribution and to improve traction (front wheel drive).

But this setup is way too heavy. Roverling run time is reduced substantially and traction is a real problem, especially with the terrain around here. I know I could go back to the drawing board with Roverling, but I don’t want to go down that path.

What I really need is something that is very rigid, about 600x400mm, nonmetallic, light and not too costly, but I can’t think of anything suitable. I’ve been stumped for a while, so if anyone has got any ideas, I’d really like to know.

Thanks, Mark
PS. I’ll be offline until the New Year, so if you reply before then, I’m not ignoring you
PSS. Wishing everyone a great festive season

2 Likes

Robert9382023 December 2024 06:3015

Hi Mark
The “very rigid” in that size is the hard part. You could think polycarbonate but by the time you met the “rigid” criteria it would probably be pretty heavy.
Unless you can get some thinner material and stiffen it up with some angle extrusions or box section material.
Cheers Bob

2 Likes

Liam12034723 December 2024 10:2716

Hi Mark,

A cheap suggestion would be your geometries, if you add some spars along the top and bottom you could likely do away with the glass (it looks like this is the densest material).

If you can bind the two coils together and remove them from the glass, and replace that with some PC that will yield large weight savings.

(to put my thoughts into a photo, something like this : https://www.bunnings.com.au/all-set-100l-grey-and-green-heavy-duty-storage-container-with-flat-lid_p0406871)

1 Like

MarkMakies30 December 2024 00:4417

@Liam120347

(to put my thoughts into a photo, something like this : https://www.bunnings.com.au/all-set-100l-grey-and-green-heavy-duty-storage-container-with-flat-lid_p0406871)

Thanks for the suggestion @Liam120347. I have some of those at home, I get the idea, but these are very flexible.

MarkMakies30 December 2024 00:5218

@Robert93820

The “very rigid” in that size is the hard part. You could think polycarbonate but by the time you met the “rigid” criteria it would probably be pretty heavy.
Unless you can get some thinner material and stiffen it up with some angle extrusions or box section material.

Hey Bob,

I’ve taken a plunge, and for the first time collaborated with an AI system to come up with a suitable design, taking into account weight, rigidity, fabrication and cost. In my next iteration I will be trying 10mm perspex, with grooves routed on both sides to house the coils and material removed where it is not needed whilst maintaining rigidity. Grooves will then be filled with epoxy.

1 Like

Robert9382030 December 2024 01:1419

Hi Mark
Sounds like a solution.

@MarkMakies

Grooves will then be filled with epoxy.

You should make sure the whole thing works as expected before you go mad with any epoxy. That stuff can be pretty final.
Cheers Bob

2 Likes

[Robert93820]

Hi Mark
That copper tape is readily solderable so you might be able to solder a strip on top.
Cheers Bob

[Jeff105671]

@MarkMakies

If I press firmly while sticking it down, will it still create an effective Faraday shield?

If you look at similar commercial products you will see that they cut small tabs from the edges that are folded up from the existing layer and soldered to matching tabs folded down from the upper layer, then folded flat when gluing down the new layer. This is repeated several times along each edge. I would guess you would do it by cutting and bending the tabs in both layers, inserting spacers (such as foam strips) between the layers, positioning the new layer, soldering, removing the spacer strips, and pushing everything flat.

If you can solder at two points, above and below the midpoint, you can then fold at the midpoint as you push the layers together. Fiddly, but possible.

[MarkMakies]

Getting ready to tune and field test, again. I haven’t completed the shield yet, but it might not be such an issue, thanks @Jeff105671, I’ll be following your advice

[MarkMakies]

It’s not pretty but after 2+ years we’re ready for the first field tests.

Rather than complicate the testing by using the autonomous Roverling, we’ll start with this 2.5m ‘pusher’, where I can record and analyse data in real time.

Time to go bury some treasures and trash and see how it goes.

[Robert93820]

Hi Mark
Good luck.
Cheers Bob

[MarkMakies]

I have now developed a Python-based GUI to interface with the RP2040 acquisition MCU. The timing between the removal of the transmit pulse and the subsequent acquisition of the receive pulse is accurate to within 15 ns. Once the coils and circuitry have warmed up, the standard deviation of the received signal is typically better than 100 µV.

The slope seen in the data below is due to the system still warming up. Nonetheless, it’s clear that ferrous metals (steel spanner) are detected as positive spikes, while non-ferrous metals (piece of copper pipe) produce negative spikes. The polarity difference happens because ferrous metals store magnetic energy, reinforcing the decay field (positive spike), while non-ferrous metals induce opposing eddy currents, weakening the field (negative spike).

Through testing, I’ve found that the optimal setup is a 40 µs transmit pulse, generating a peak current of 7 A. Sampling at 8.4 µs provides the best results, as this is when the 100V+ signal has been critically damped to around 3V. The system runs at a 5 kHz pulse rate, with a decimation filter of 256, providing 20 samples per second.

I’ve already detected a few large underground spikes but haven’t dug them up yet to see what they are.

[Robert93820]

Hi Mark
That is a positive result. If the sensitivity is good enough you could find all sorts of things.
Well done.

Ferrous material will increase the inductance of a coil and non ferrous will decrease it. Mostly in tuneable inductors the adjusting material is Ferrite or in larger cases iron but in the smaller variety I have seen (very occasionally) brass used.
Cheers Bob

[MarkMakies]

@Robert93820

Ferrous material will increase the inductance of a coil and non ferrous will decrease it. Mostly in tuneable inductors the adjusting material is Ferrite or in larger cases iron but in the smaller variety I have seen (very occasionally) brass used.

So if it is brass you screw it out and if ferrous you screw it in to get the same result?

[Robert93820]

Hi Mark
Basically yes. Depends of course where the starting point is. If a brass “slug” is already out screwing it out further may not make any difference. Same with ferrite. I think the ferrite might be more sensitive to change than the brass. I am not sure about that. Probably depends a bit on the actual material

Some things like tuneable small transformers had 2 ferrite cores, one for each winding. Some had only one which would increase the resonant frequency of one coil and reduce it in the other. The main function here was to trim the coupling and hence the response curve of the whole thing. In my time the 2 ferrite cores were the most common.

A common example of tuneable inductors was some of the older car radios with preset tuning points. The tuning was done with the inductor component instead of the usual tuning capacitor. The presets were entirely mechanical where a button was linked to return the inductor ferrite core to a preset position.
Cheers Bob

[MarkMakies]

First ‘treasure’ found after 1 hour of searching. Looks like a brass dipstick, about 20cm down. Unfortunately it broke during extraction.

It has some markings but I’ve been unable to find out more about them.

As far as I can tell it’s pre 1940 dipstick from a vehicle or Ag equipment

[Kimmo]

Mind you don’t go stepping on anyone’s toes… I hear that detectorising caper can be a cut-throat business.