felezjoo PI(the best pulse induction metal detector that I made until now)

[geoscash1]

Could it be a problem with the coil cable, I have all the coils built with speaker cable. It worked well for the Clone Pi AWR. The auto delay is 19, 390mh, 1.2 ohms.

That is something youll have to try to work out. Do you have the capability to test your coil parameters (scope and test equip)?
I will go check my docs and see what the coil parameters were, they are definitely specific to this detectors code and will set the coil alarm if not within the paraneters.
Edit: some screen shots of adc parameters.

[Aadu]

Thanks. That's helpful.

[geoscash1]

Thanks. That's helpful.

Your welcome. Hopefully it helps a little and you can get it fixed.

The software has some pretty extensive features in it even in the demo/rental mode.

[Aadu]

Hello. Question about the resistor. In all the diagrams R5 is 2.2k, but on the printed circuit boards it is 10k. Why was it changed?

[boilcoil]

Hi Aadu,
Your question is the reason I read this topic, which I had not paid attention to until now.
The topic is quite long, almost a movie saga.

Indeed, the FELEZJOO project is a very interesting project and directly proves the advantages of microcontroller machines over analog ones. There is no need for complex hardware in MCU control, because almost everything is virtual.

Regarding your question about R5 - its value is not critical. It serves to turn on the key mosfet and this load only the input capacitance of the IRF740/840 and the output capacitance of the driver transistor-BS170. It does not matter how long the IRF740 will turn on, because the pulse duration is long enough. So whatever value you put in the range you quoted, the TX stage will work. A smaller value will turn it on faster and there will be smaller transient losses in IRF740, but in this case this does not matter at all.

Regarding the "magic numbers" - these are three consecutive 10-bit samples - the first with a small delay, say 15-30 us, and the others are offset from it by 100-160 us. The first sample is used as a sense sample and is combined with the other two to perform the so-called discrimination, with the MCU calculating the slope of the falling curve between different samples. That is why these numbers are so important, not for the sensitivity of the device, but for discrimination and grounding (in reality, the device does not have a real ground balance, but tries to compensate for the ground effect). Without a real ground balance it is impossible to work on mineralized soil (this is the reason why searchers fly the coils high above the ground).
These "magic numbers" are the result of three things - the characteristics of the coil, the gain of the preamplifier, and the energy in the coil, (determined by the inductance of the coil and the current through it). You can achieve these numbers with ANY coil and ANY opamp, which must have a slew rate greater than 13V/µs and a higher bandwidth. To achieve this, you have to first model the decay curve through proper damping and pulse duration, and select the appropriate preamplifier gain..
I saw a proposed solution by placing a series trimmer in the clamp diode circuit. This is not the most optimal solution, because by changing this trimmer you change both the damping conditions in the first stage of damping and the preamp gain. A better solution is to put a constant resistance of 300-600 ohms to the clamp diode circuit, why not and 1k, and a variable trimmer, say 1.5 kilohms as the main damping resistor. Separately, the preamplifier gain can be adjusted using the feedback resistor.
Of course, setting the samples with other delays would help a lot, but as far as I understand from what I've read, this is not possible from the device interface.
Maybe the device will have a different sensitivity, but you will be able to build it with more popular elements and non-critical coils.You can even try shielding the coil to allow for quieter operation.

There was a question why a linear stabilizer of 12V is required, and why not working directly with 12V. The reason is that all calculations and algorithms by the author were made for a strictly defined system - <coil-impulse-preamplifier>. Any change would lead to incorrect results, so he requires a stabilized voltage on the TX module. Of course, you can lie the device that the voltage is 18V by changing the resistances in the divider to the ADC of the MCU, but in real operation, any change in voltage will lead to unstable operation and incorrect results.

Obviously, to overcome the limitations of the 10-bit ADC (if the device is not very fast in operation, perhaps the author also applies oversampling to improve the resolution), the preamplifier must have a high gain to provide acceptable sensitivity, but this leads to problems with the first sample. I suppose this is the reason it works best with spider coils, which are essentially very fast coils and can provide a fairly fast decay to the first sample.

I would still say that this is a really good project, with good ideas and concepts (I was especially impressed by a good idea for automatic alignment of the first sample, based on the ADC result). In any case, if the project was more open about the software, it would be really wonderful.

[Aadu]

Thank you for the very thorough answer. This is a good detector but the big drawback is the response to the ground. Not suitable for clay soil. I have tried different shielding, it is not good. If I could find a solution to this, it would be a great detector

[Laith]

Anyone has make edit on felezjoo to make it more stable ??

[Laith]

Anyone test opa627 ???

[sabios]

I'm having a problem with the coil.
The coil I'm using is 1.8 ohms and 400uH. Do you think that's the problem?
By the way, I'm using this one.
FELEZJOO Rev A-2 (UPDATED 8TH OCT 2021)

[sabios]

What do you think of this BS170?

[TCHASS]

Salut les amis. J'ai un sérieux problème avec mon FelezJoo Pi. Le détecteur démarre correctement mais le seul problème c'est l'erreur '' Coi Error ''. A part le son à l'allumage et à l'extinction, rien ne se passe.

[Qiaozhi]

Salut les amis. J'ai un sérieux problème avec mon FelezJoo Pi. Le détecteur démarre correctement mais le seul problème c'est l'erreur '' Coi Error ''. A part le son à l'allumage et à l'extinction, rien ne se passe.

Please read the forum rules -> Basic Rules of the Forums
and make our posts in English.

[sabios]

I just finished this project, and after the tests I've done, I'm not sure if it's correct.

1. It detects a coin at about 20-25 cm.
2. It detects a can at about 45 cm.
3. My coil is made of 0.8 mm magnet wire, is 28 cm long, has a resistance of 1.6 ohms, and an inductance of 400 uH.
4. I used a CA3130 and a BS170 with the drain and source pins reversed, both purchased from AliExpress.

5. To fix the coil error, I replaced the BS170, IRF840, and also the 1N4148 diode, and that solved the problem.

Another question: what would be the best coil for detecting objects at greater depths, for example, for detecting coins and small and large objects?

[sabios]

I've been doing some tests and I have to say:

1. When doing some tests, the first thing I noticed is that it detects stones around here, and I don't know if that's normal.

2. I did a test with a can of tuna, and it detects it at about 70cm with a 30 x 45 coil. The main thing is, it's normal that it detects stones.

[sabios]

I made this modification to the circuit board I'm working on. What do you think?

By the way, I'm still having the same problem, since the detector is picking up the stones, which is annoying.
It might be the cable I'm using, since I couldn't find RG58, so I'm using one from an old microphone.
Also, I installed the 2W potentiometer on resistor 11, and it seems to be working, but the stone detection problem persists.