Progressive Or Thrash? How Metal Detectors Discriminate

Metal detecting is a fun pastime, even when all you can find is a little bit of peace and a whole lot of pop tabs. [Huygens Optics] has a VLF-based metal detector that offers much more feedback than just a beep or no beep. This thing is fancy enough to discriminate between types of metal and report back a numerical ID value from a corresponding range of conductivity.

Most pop tabs rated an ID of 76 or 77, so [Huygens Optics] started ignoring these until the day he found a platinum wedding band without looking at the ID readout. Turns out, the ring registered in the throwaway range. Now thoroughly intrigued by the detector’s ID system, [Huygens Optics] set up a test rig with an oscilloscope to see for himself how the thing was telling different metals apart. His valuable and sweeping video walk-through is hiding after the break.

A Very Low-Frequency (VLF) detector uses two coils, one to emit and one to receive. They are overlapped just enough so that the reception coil can’t see the emission coil’s magnetic field. This frees up the reception coil’s magnetic field to be interrupted only by third-party metal, i.e. hidden treasures in the ground.

Once [Huygens Optics] determined which coil was which, he started passing metal objects near the reception coil to see what happened on the ‘scope. Depending on the material type and the size and shape of the object, the waveform it produced showed a shift in phase from the emission coil’s waveform. This is pretty much directly translated to the ID readout — the higher the phase shift value, the higher the ID value.

We’ve picked up DIY metal detectors of all sizes over the years, but this one is the ATtiny-ist.

35 thoughts on “Progressive Or Thrash? How Metal Detectors Discriminate

    1. Indeed. But the basic principle for most detectors is looking for the return signal vs. the signal you sent out, compare amplitude and phase difference. I know of two other methods than injecting a single tone, one where what is near the transmit coil affects its frequency, another is where you send a pulse and look at what comes back.
      A thing I noticed in this video is that there appears to be a higher-frequency signal in the detection coil. I wonder if it is that the return signal has overtones due to the magnetic properties of the item, or if it is some local interference that it picks up.

      A word of caution for people is that in some areas, use of metal detectors may be restricted and may cause fines or other issues with law enforcement, so check local regulations.

      1. “A word of caution for people is that in some areas, use of metal detectors may be restricted and may cause fines or other issues with law enforcement, so check local regulations.”

        Oh. Why?

        1. Law in post-Soviet countries is generally that everything under ground surface – water, oil, treasures, ores, whatever – is property of state even when ground itself is private property.

          One of provided justifications is preserving history artifacts for archeology. Guys with metal detectors had been digging into what turned to be bronze age grave sites.

          (This protection sometimes goes too far and turns out to be counterproductive, but it’s another topic in itself”

    1. Is there any reason given, why they made such an extreme law? A restriction of use could somehow be understood, but to illegalize just the possession is really extreme – it’s no firearm or such.

  1. I’ve often wondered if there isn’t some science in metal detection that’s waiting to be discovered.

    For example, using an optical flow sensor (mouse sensor focused on infinity, like they use in drones for station-keeping) could give the processor information on how fast the coils are being swept over the ground.

    The magnetic field is subject to superposition, so there’s no way to get an image from a sensor at one point (only a single strength value), but along with the flow sensor you could map the magnetic field values at many points and put together a sort of “image” or map of field results over an area, then use image matching techniques to detect different shapes of objects. Possibly be able to distinguish between a ring and a rusty nail, and possibly sweep the coil over an interesting signal multiple times to get a higher-resolution image.

    1. I’ve been thinking about AI as an approach. Build a dozen or so recording detectors and give them to the local club if the members agree to carefully document what they found (how deep, weight, what kind of metal, etc.) Then train based on the data.

      You could also do controlled training with a sandbox and a large array of interesting metal objects.

      This seems like a perfect application

    2. One of the ones I’ve thought would be interesting to do is run basically three metal detectors strapped together (in such a way that they don’t interfere), and use it to triangulate things. Throw it on some sort of heads up display, like a Hololens.

    3. You might have better luck looking for new engineering than new science. ;)

      That said, I think what limits advancements is that when you start adding more techniques it starts to become as difficult as ground-penetrating radar, and then people realize, “hey why don’t we just use ground-penetrating radar?”

  2. With the advances in signal processing and other electronics technology, I wonder if the old BFO or Beat Frequency Oscillator metal detector system could be improved upon? BFO was the original technology used for portable landmine detectors. That was reduced in size to make it operable for longer time on batteries, and small enough to be carried around by a single person.

    VLF and other designs superseded BFO in the latter half of the 1970’s and much effort was put into designing analog circuitry to add discrimination to them to try and reject junk like nails, chunks of iron etc. The old ring shaped pull tabs that came off soda and beer cans proved to be a nigh unsolvable problem because their magnetic signature is very close to American nickel coins (and other coins of similar size made of nickel based alloys) and gold or platinum rings. Detectorists would use rings, pull tabs and other objects to test and listen to subtle changes in the sound with headphones and how the analog meters twitched to get a feel for whether they’d picked up a ring, a US nickel, or another damn pull tab.

    Much more recent digital signal processing can tease out the fine differences in magnetic signatures to better discriminate between targets that have similar signatures, along with fairly accurately identifying various types of coins. YouTube videos of people detecting on sites like parks and school grounds that have been in use and undisturbed for decades still often find their fancy detectors fooled by old ring style pull tabs that have been lurking beneath the surface for 35+ years.

    But back in the day, BFO was still the depth king, able to pick up metal objects at greater depths than any other method. So relic hunters who were interested in everything didn’t care about trying to decide if they’d found a gold ring or a chunk of something else that bent the magnetic field in nearly the same way.

    Another thing modern electronics has brought to metal detecting is zero motion pinpointing. Without that feature, the detector coil has to be in motion in order to detect the field distortions. Swing it left and right, move it fore and aft and narrow down approximately the position of the target. Some detectors with zero motion mode have a hole centered in the coil to stick a probe in the ground so you know where to dig when the detector is set aside.

    1. I always think this is a much overlooked approach, going back down the tech tree finding an older branch and seeing what we can do with it today. Some technologies “won” because of marketing or patent encumbrance or the competitor being too technically immature for the time.

  3. Why is 15.3kHz used ?
    Is it a sweet spot ? Is it because it gives some kind of an optimal ground penetration depth with the strength of magnetic field that can be generated by the batteries used for a reasonable duration ?

    I’m guessing the penetration depth of soil would be similar to the eddy current skin effect, where the lower the frequency the deeper the depth, and the higher the frequency the shallower the depth of penetration.

    The way, after watching the above video, that I now see a metal detector working is like a transformer, the primary signal coil in the detector transfers energy to the short circuited secondary coil (the object to be detected). The sensor coil in the detector is exactly overlapping the signal coil so that by default all magnetic flux is 100% cancelled. And the sensor coil can either detect distortions in the flux from the signal coil caused by the object to be detected or the weaker returned magnetic flux from the short circuited secondary coil (the object to be detected).

    Is 15.3kHz used because decreasing the frequency lower, would mean that any signal returned by an object deeper would be attenuated so much by the thickness of soil above to be less than the thermal noise power within the circuit used for detection. Basically that there is no point in lowering the frequency used (increasing the penetration depth) without increasing the power by multiple order of magnitude.

    I also wonder if you used a massive extremely short duration magnetic pulse, instead of a fixed amplitude sinusoidal magnetic field, if deeper detection could be achieved. It would require much more signal processing, but the amplitude returned would give an indication of depth, the phase difference would reveal the material and the decay profile an indication of the size of the object. It almost sounds like something that could benefit from using a trained neural network to profile objects. Maybe even add a 6-Axis MEMS (3-axis gyroscope and a 3-axis accelerometer) for additional inputs to the neural network to help identify objects.

    1. 15khz is the multifunction frequency.

      There are lots of single frequency detector that aren’t 15khz. Around 7.5khz is another popular frequency because it works well in soil and finds things many people are looking for; coins, jewelry, landmines, bullets, etc.
      What frequency you use depends on what you’re looking for where.
      Coins in a lake? higher is better. UXO in an old battlefield? lower. Jewelry or coins on land? mid range. Precious metals? high again.
      As to your transformer analogy. Sometimes. That depends on how your detector works. Some work on phase variance, some still use beat frequency (BFO), some on pulse induction which uses a momentary induced magnetism like you described.

  4. Supreme kudos to [Kristina Panos] for the best article title in Hackaday history, IMHO.
    As for which is better, progressive or thrash, the jury is still out in my brain. I love them both and have done so for at least 30 years. I’ll let you know if I ever pick a favorite.

    I had an Ace 350 detector from back when Garrett first started making the “DD” style of coil, and I loved it. For what was essentially a toy metal detector (brightly colored plastics, cheap electronics, etc.), the darn thing worked exceptionally well. I was sad to have to sell it, but that’s life sometimes.

    1. I on the other hand, seem to have a “toy” that’s nothing more than a toy. One of the old Radio Shack detectors, now, back in the day, a family member had one, and performance was great, just bare bones performance, no bells and whistles, when he bought an upgrade costing wayyy more, the RS was picking up things at a couple of inches deeper. This one I picked up looks pretty close to it, but despite all appearances of working, can’t find anything smaller than a door hinge, any deeper than lying right on top of the ground. One of these days I’m gonna get the ambition to replace every component in it, and then see if it’s worth crap.

  5. There are a few research ( and hobbyists bordering on research) groups that create maps like what you describe. Mostly used for meteorite hunting and probably a few NGOs doing UXO clearing.
    As with any remote sensing, so long as your travel speed is sufficiently below the sample rate you should be able to create a reasonably precise map for whatever buried goodies you’re looking for.

  6. I use to do a little metal detecting, but moved on to other interests. What I learned though, is that the most productive ares, are also the trashiest areas. Metal detecting isn’t just the equipment, it’s more of a craft. You really never know what’s in the ground, until you dig it up, which can be a lot of work. You learn over time what the display is telling you, and what the headphones tell you, and you get a pretty good idea of how deep, how large, and what might be down there. Even then sometimes you get surprised, sometimes disappointed. It’s a hobby, you can work as hard as you want, or walk around all day, waiting for a sure thing. Like most things, the harder you work, the more productive you are. Most of the discrimination, and other features are marketing, to sell machines. Like anything else, the newest, most advanced, and most expensive, sell well. People perceive an advantage from getting the ‘best’ offered. Mostly, it just renews interest, and motivates more effort, which will of course show as being more productive, for a while, as the new features are thoroughly tested. And of course, by the time interests starts to slow again, a new, better machine hits the market…

    There is always going to be limits on depth. Magnetic fields go in all directions, not just down. Shielding can only help so much. The greater the power, the more area, and potential more items are in the area of influence. Everything in that area, will have some effect on the readings you get. Might be a gem in the middle of a bunch of junk, but you won’t know, because you will only read the trash.

    1. That’s a professional geolocator that’s packaged as a consumer metal detector to sell it to rich people who want a fancy toy.

      The point is, regarding range, that you can use it to find ore deposits. It doesn’t mean it does a better job of finding small items that people normally are hunting for.

      They’re a bit like a cross between a ground-penetrating-radar (note frequency up to 2Mhz) and a fish finder. So you can’t take an underground image like with a regular ($25k+) ground-penetrating radar, but you can map the pings in different spots to understand how different minerals are distributed.

      Normally they’re combined with fancy geology software that costs more than the detector.

      Probably not an improvement over a (cheaper) metal detector with a fancy DSP, unless you’re just looking for the best spot to pan for gold.

Leave a Reply

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