A Graph Plotting Metal Detector

Metal detectors can be a great source of fun, and occasionally even found wealth. They allow the detection of metal objects at a distance, enabling hidden treasures to be discovered. They’re also highly critical to the work of minesweepers and unexploded ordnance disposal teams. [Andrius] wanted to add such a device to his kit when motorcycling through the woods of Lithuania, and thus decided to undertake a build of his own. (Editor’s note: original link went bad, this is through the Wayback Machine.)

The detector is a thoroughly modern one – fans of the 555 may want to look away now. A Collpits oscillator, built from two transistors, is used to generate a frequency that is passed through the detection coil. This frequency is measured by an Arduino that plots a graph of the received frequency on an OLED display. As the coil is passed near metal objects, the oscillator frequency changes, and this is visible on the frequency plot on-screen.

Not only is it a quick and easy build that is achievable from what are now junkdraw components, it’s also one that would be readily usable by the hearing-impaired, too. It’s a great project to tackle if you’re looking to get to grips with basic oscillators, frequency measurement, or just microcontroller programming in general.

Still need more inspiration? We’ve seen a similar concept executed before.

31 thoughts on “A Graph Plotting Metal Detector

      1. As much as I would like to agree with the “Buzzword” reply there actually is space for some AI here. Some modern metal detectors attempt to determine the size and material of the thing detected. If I could train an AI to differentiate between quarters and bottle caps or determine that something is gold rather than tin that would be super-compelling.

      1. Whenever I think of metal detectors, I think about some dude on the beach combing for lost jewelry. So something like a roomba you plop down on the sand seems like a neat project idea. Plop it safely away from the water, fire it up, it’d go back and forth like mowing the lawn. Could even have it spin in a circle whenever it found something metal in the sand leaving a lil’ crop circle so you know where to dig. Spend more time napping while it works.

        1. I’ve thought about using a metal detector to locating a plastic water line. Simply drain, then feed a flexible metal line all the way down to the meter, then trace.

          1. @Ren
            Oh the EPA would love that.* But in finding a leak a clever way is shut off one end, drain, and pump in some air. Turns it from a hard to find water leak, to an easy to find air leak. Less messy too.

            *Had one suggest water dowsing.

          2. We did this (pressurizing with air) once to find a leak in heating pipes. As it was the ground floor, the water did not show up anywhere, it just got lost into the earth. But with the air we could hear the hissing.

    1. I think there is a method for metal detection which excites the tank circuit with pulses from a microcontroller and looks at the waveform of the decaying oscillation to measure changes of Q. Unfortunately this was about 10 years ago at a company which made really mine detection equipment and memory is quite shady now. I remember the test piece was a metal pin about 15mm long and 2mm diameter which was detected over several 10cm of distance, but not sure any more.

  1. Most commercials metal detectors are way overpriced / Most hobbyist detectors are way too rudimentary.

    If I understand right, being able to use multiple frequencies is desirable, but then the coil is not tuned at its best, and the performances suffer from it.

    1 – Would it be a solution to have different coils with different diameters and different resonance frequencies to achieve a *good* multifrequency detector ? Or one coil with several taps in the middle, like an transformer than can provide different voltage ?
    2 – Different coils for TX and for RX is the best, right ?

    3 – Also, if I get it right, the real deal is to be able to filter the background noise out, and this changes according to the geology of the terrain, humidity, and so, right ? How is it done ? with a kind of envelope follower filter, or pid algorithm ?
    It filters out *progressive* changes and amplifies sudden changes ?

    4 – Is there any other feature that the high-end metal detectors use ? (except the GPS mapping / touch screen UI, that are trivial to implement for us electronic hackers) What makes a metal detector real added value ? What improvement could bring for example a DSP ?

    5 – what shape is the TX coil signal ? always a sine ? pulse ? would some more complex signals (freq modulation for example) improve the detection ?

    1. “1 – Would it be a solution to have different coils with different diameters and different resonance frequencies to achieve a *good* multifrequency detector ? Or one coil with several taps in the middle, like an transformer than can provide different voltage ?”

      Some metal detectors I’ve seen have a large coil and above that have 4 coils (about 1/4th the size of the large coil). I thought I’d find one on Amazon to show you, but after 6 pages, didn’t see what I was thinking about, but they do show a bunch with large outer coil and a smaller inner coil.

        1. +1
          Good link, saves me going off on a tangent explaining this to a tech. I wrote a design brief some decades back on GPR like signal analysis, theory sound but, way too many other things on my plate, thanks also for link looks comprehensive and thought provoking too in terms of revisiting with newer tech from 30 yrs back, cheers.

      1. I’ve seen a different approach to the pinpoint-problem too. You have one device with the large coil to sweep the floor and cover a large surface area quickly. Once you have found some signal with that, you have a smaller handheld device to pinpoint the signal location further and test the dirt while digging around on the floor.

    2. To answer question 3. There is no need for PID. It is not a controller but a detector. I have made one last year similar in principle to the one described here but no display. It sound a tone when something is detected. To avoid false trigger it filter out oscillator drift by only considering fast frequency changes. It count frequency at short interval and check for count difference between successives samples.

      source on github: https://github.com/picatout/metal_detector

    3. I’ve wondered about a few of these things as well. Would it be possible to use three different coils to triangulate an object, assuming all three coils were firing at slightly different intervals as to not interfere with each other?

  2. If you can get to the two end points with water still in the pipe then apply a suitably high voltage short rise/fall time pulse at regular intervals whilst you traverse the area with a good quality conventional metal detector. Ie you’d be picking up disturbance in the eather ;-)

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.