Experimenting with metal detecting

[ppTRN]

Hi, I want to understand more about metal detectors and possibly design one of my own. It will be uC based, but the problem I am having difficoulties to overcome is how to amplify a signal to feed to the searching coil (some 20mH, 1Ohm resistance). I literally have no idea about which circuit to use. Keep in mind that i do not want it to amplify just a sine wave, but i want to be able to feed my own signal to the coil, possibly generated by the uC, and I want the signal to be as strong as possible.

Edit: bandwith of the signal Is no more than 50kHz

If it wasn't for the low resistence of the winding i would even have used a power amp like TDA2030 or something like that.
You guys have any suggestions?

[Infraviolet]

"Metal detector" describes a whole family of devices which operate by a variety of different principles.

The simplest trick, though with perhaps the shortest ranges is to have an inductor set up as part of an LC oscillator, when metal comes near the inductor eddy currents induced in to it have an effect where the inductance of the inductor gets changed by upto a few %. The LC oscillation the changes frequency.

Your plans sound like you'd rather use an emitting coil and receiving coil instead and look for signals in the receiving coil "reflected" off the metal. If you simply need to amplify a signal from the MCU, probably a digital signal anyway as most microcontrollers have ADCs but not DACs (some will let you use PWM to fake an analog voltage, but that is only for things like motor speed or LED brightness where a fast flicker looks like a continuous medium level on human timescales), how about just a transistor acting as a switch to control larger signals? NPN as a low side driver?

[zrq]

I had some fun experimenting a computer sound card as DAC-> Power Amp (headphone amp) -> Pre Amp (for mic) -> ADC for induction balance metal detection, and managed to achieve reasonably good result with trivial DSP. I simply changed the winding turns to make it have around 6 ohm impedance, so can be comfortably driven with audio amplifiers. BJT input stage headphone amplifiers also have nice performance at low input impedance.

[ppTRN]

"Metal detector" describes a whole family of devices which operate by a variety of different principles.
[....]
Your plans sound like you'd rather use an emitting coil and receiving coil instead and look for signals in the receiving coil "reflected" off the metal.

That is correct, i want to develope a metal detector with a double D coil

If you simply need to amplify a signal from the MCU, probably a digital signal anyway as most microcontrollers have ADCs but not DACs

I will use a Teensy4.1, so a really powerful device, that even if it does not have a DAC, it will be fairly easy to use it with an audio IC to reproduce sound stored in a uSD. Said sounds will be the waveform i want to feed the TX coil.

I simply changed the winding turns to make it have around 6 ohm impedance, so can be comfortably driven with audio amplifiers. BJT input stage headphone amplifiers also have nice performance at low input impedance.

I guess i'll try to increase the resistance of the coil and use an audio amp, but this will decrease the bandwith of the signal. Not that big of a problem, but i was really hoping to reach 50kHz

[jmelson]

I had a Vietnam-era US Army mine detector.  It used vacuum tubes, but the way it worked (I think) was there was a square transmitting coil with 4 round coils inside the square.  The 4 coils were connected in a plus-minus configuration so their signal would cancel out when no metal was near.  Any metal nearby would disturb the field and the transmit signal would not cancel out.  The signal would then be hetrodyned down to the audio range (or maybe the main oscillator was actually IN the audio range.  Detection was with headphones.  Anyway, the thing was totally amazing!  It would detect a sewing PIN at over a foot.  There would be a null in the signal when the object was exactly in the center of the 4 coils.
Jon

[Conrad Hoffman]

You probably won't (and shouldn't!) build this, but there's some interesting information in this old DIY article:

[ppTRN]

I had a Vietnam-era US Army mine detector.  [...]  Anyway, the thing was totally amazing!  It would detect a sewing PIN at over a foot.  There would be a null in the signal when the object was exactly in the center of the 4 coils.
Jon

That is outstanding, old tech, expecially military one, never ceases to amaze me.

You probably won't (and shouldn't!) build this, but there's some interesting information in this old DIY article:

You're right, i'll never build this hahahah but I really appreciate the infos, really intresting paper, thank you.

I really want to experiment with target locating using some kind of DSP, we'll see once i build the coil

[ch_scr]

The book "Inside the Metal Detector" from Overton and Moreland has a nice analogue design that is well explained and the book is easy to read.
I felt like it gives a good foundation of knowledge on the topic.

[metrologist]

I just wanted to buy one, but there is such a broad range and price.

Minelab Equinox 800 for $900

Garret ACE 400 for $357

and $70 cheapies

[Alex Eisenhut]

I live near a few public parks, I wonder if I could pick up loose change and whatnot around here. Are the Aliexpress detectors any good?

[ppTRN]

I live near a few public parks, I wonder if I could pick up loose change and whatnot around here. Are the Aliexpress detectors any good?

I think so, loose change will be at 15cm of depth max, so no problem with that. The thing you will find really annoying will be the lack of good discrimination between "good & bad" metals which will resoult in a lot of digs for junks.

I just wanted to buy one, but there is such a broad range and price.

That's the reason of my intrest on the topic. I would really love to have one and since i am an EE, why not make my own version?

[Infraviolet]

You should note, some types of metal detectors require a change in the presence of metal to actually make a detection, this sort can also make some attempts to identify the type of metal. The LC frequency change type  on the other hand can work in a static configuration, giving a different output signal (depending how you process it) when metal is or isn't present and (again dependent on signal processing circuitry) an analogue level to indicate something about distance to or size of the metal object. Might not be such an important diference for a detector you are sweeping back and forth over the soil, but was an important choice when I was designing a metal detecting proximity sensor as a non-contact endstop.

There is a youtube channel where someone seems to have been DIYing just about every different metal detector type, he fitted each of them in to the same design of handle structure and compared them. Can't remember what the channel was called though.

[ppTRN]

some types of metal detectors require a change in the presence of metal to actually make a detection

[...]

There is a youtube channel where someone seems to have been DIYing just about every different metal detector type, he fitted each of them in to the same design of handle structure and compared them. Can't remember what the channel was called though.

I am making a non-motion metal detector, and i think I know what videos you are refering to, but hust for reference i will link them here:

This is my favourite, by Huygens Optics: https://youtu.be/EcuTsifSgBs
Another intresting one by Diodes Gone Wild https://youtu.be/eCRgNr6jj2g & https://youtu.be/-czOV-hKSlA

But again, my issue is to feed a custom signal to the coil, i want to try what append in a multi frequency system

The book "Inside the Metal Detector" from Overton and Moreland has a nice analogue design that is well explained and the book is easy to read.
I felt like it gives a good foundation of knowledge on the topic.

Almost forgot this, i will surely take a look at it whenever i can

[ch_scr]

[...]

Almost forgot this, i will surely take a look at it whenever i can

The guys that made the book also have a website with some of the material, not as well structured or easily consumed,
but still a lot of good information in one place: www.geotech1.com
Especially the "Coil basics" might be worth a look, to not have to re-invent the wheel.

[RoGeorge]

I think this one worth a try, pointed by Gyro here https://www.eevblog.com/forum/projects/portable-proton-magnetometer-teardown-and-tech-info-wanted/msg4079602/#msg4079602 from Practical electronics 10/1970 at page 33 of 92:
Proton Magnetometer, based on NMR! 
https://worldradiohistory.com/UK/Practical-Electronics/70s/Practical-Electronics-1970-10.pdf

Might worth mentioning to check the local laws regarding metal detectors and prospecting.  Owning and/or using a metal detector might be illegal in some regions without a special permit.  We have such laws here, in Ro/EU.

[ppTRN]

Quick update:

In the picture attached you can see the coils set up. The central one is the TX coil, the others are RX. At first i made the coils all the same with 100 turns of 0.5mm copper diameter. I had to decrease the turns of the transmitting because i had an inductance of 22mH. Now it has 30 turns for a measured inductance of just above 300uH. I tried to drive it with a sine wave amplified by a TDA2030. I measured, at 15kHz, a 10Vpp voltage at the RX coil. I found pretty easy to null the voltage at the RX coils just by moving the RX.

Now that it is all set up, i was expecting to read a voltage at the TX coils when a metallic object passes nearby. NOT A CHANCE! It did not work at all. I must do further investigations. Thanks for your help, i'll keep you posted.

[ppTRN]

So, quick update on the project (I do not think it is necessary to create a new thread, right?). Here's a list of what I observed in the last few weeks:

1. When the RX coil is placed righ over the TX one, the coupled voltage is detected to up to 1 meter of distance.
2. The diameter of the wire does not seem to be really important.
3. Just like a transformer, if the RX coil has more turns it will boost the detected voltage. My TX to RX ratio is 30:75.
4. Adding a capacitor (matched to the coil inductance and frequency) in parallel to the RX coil significantly increases the amplitude of the RX voltage, but does nothing about sensitivity.
5. Balancing the coils is not that difficult, but with the capacitor in parallel to RX i cannot drop down to 10mVp.
6. Even trying to amplify the RX voltage, it get "more responsive" but not more sensitive in terms of distance.
7. Coins, keys and other small object that even the cheapest detector would detect, my porototipe doesnt.
8. Lower frequency from 5 to 10kHz that should be more effective in the distance does not work. I found that 15 to 22kHz is the best frequency to operate.

What I think I learned:

a. Many Induction balance detectors does not put much power to the RX coil, but highly amplify the RX coil.
b. A simple and cheap audio amplifier will drive the coil widouth any issue. I currently use TDA2030A and achive almost 30Vpp to the RX coil.
c. A matched capacitor on the output is impossible to connect since the amp output will start oscillate.
d. I do not think that i need to increase the RX power.
e. Conditioning the singal seems to be useless.

I am feeling pretty discouraged right now and really do not know what to do to obtain some decent resoult. I will keep triying and keep you posted. In the meantime I attach to this post an updated image of the coils and some extra info:

RX coil: 75 turns of 0.5ish mm diameter, not shielded (a cable wraps it and it is not closed to itself), measured induction is 2.42mH;
TX coil: 30 turns, same as above, i did not measured the inductance but i expect some 1mH;

Anything you can think of, please share!