Non motion detector

[ivconic]

Maybe this will be interesting for someone.
For many years I have been planning to try to make one real non-motion detector.
But there is always something going on to delay the start of work.
These days I have finally started doing something.
To reduce the possibility of error and downtime; I chose the Tesoro platform for further development.
I started with the schematic for Tesoro Eldorado, because I have already made the Eldorado model several times in the past and I have always managed to do it well.
Another very important reason is because I currently have the original Tesoro coil.
In that way, I accelerated and simplified my work and got the time and comfort to focus only on the important things.
In addition to Edlorado, I took Bandido and transferred some solutions to the Eldorado platform.
It's basically a simple design. It has two channels. GEB and DISC channel. But both channels are actually part of the pinpoint at Eldorado.
Only the channels are controlled by the synchrodemodulator from Eldorado.
Both GEB and DISC work very "cleanly" and accurately.
But I decided to separate the channels, not to mix them, but to try and work on only one, respectively. To be able to try in particular how one channel will behave and then the other.
And that's why the schematic is simplified, it has only one common audio level and with just use of jumpers to activate or deactivate one of the channels.
When I specifically test a GEB channel, to check how only it will behave. And also when it comes to the Disc channel.
The video shows the testing of the GEB channel, which has a more complex task and is therefore more important.
I didn't record the same with the DISC channel, because there is no need, but I tested it and it works perfectly well as well.

The schematic:

[ivconic]

Coil specs are:
TX=5.84mH / 17.9 ohms
RX=6.19mH / 19.5 ohms
Freq= 9.3xx Khz
Power rails = +5v / -3.5v
TX amplitude = 6Vpp (!!)

[ivconic]

If you watch the video carefully, you will notice that it behaves very well. There are two pins on the pcb, which when I connect with a small screwdriver; "Retune" happens and the sound is muted and calmed down.
It's a "retuned" state. You will see the measured voltages in the diagram even in that state. Voltage is 0 when retuned. This condition does not last long, depending on the trimmer settings, "threshold" and "sense"; it may take longer or shorter.
And that's what we've mentioned so far that the main drawback of non-motion detectors. That certain drift. The most vital part of the problem is the solution to that problem.
There are many ways to skin a cat. But I am now looking for a completely analog solution.
If I connect those two pins on the pcb permanently, then it is in "Auto" mode, and then it is no longer non-motion but becomes slow-motion. And that is something that is not my goal.
My goal is to remain completely non-motion, and that drift is occasionally "compensated", that is, that there is additional electronics that will monitor the drift even when it exceeds a certain value; to briefly return it to 0v.
It is also desirable that there is a certain "window", adjustable in width, so that it has a voltage range a little wider, out of which the "auto-tune" will work and return the signal to 0v.
But now I come to the crucial problem!
And in the case of drift, but also in the case of useful and realistic detection of metals; that voltage will start from 0v to some value (end values are indicated on the diagram); and in that case,
a simple "auto-tune" solution would work and return to 0v signal in the case of drift, but unfortunately also in the case of real detection.
In case of drift things are clear and everything is ok. The behavior will calm down and everything is great.
But in the case of real detection, the "auto-tune" will constantly return to 0v, while the detection signal still exists and may increase, so there will be sudden jumps in voltage, from 0v to x.xxV,
which will produce "chopping" in the sound, introduce a certain delay in behavior and probably produce more side effects ...
How to solve it?
Thereby; I will always repeat; I don't want motion or slow motion behavior but full non -motion behavior.
If I were to add Atmega now and write code about it, things would probably get a little easier in solving that problem. But I intentionally want to achieve all this without hiring a processor first.
Through practice, it has been shown to me many times that if the matter is resolved well in the analog domain, that there are no flaws, then later the transition to the digital domain is much easier
but also opens many more possibilities. That's why I have a habit of always trying to make the whole device in the analog domain to perfection first.
An additional problem is the fact that very low voltages appear here, after which an audio indication can appear, and it is desirable that the audio circuit works at these low voltages, because then the detection range (depth) is much greater.
Part of that problem can be solved and compensated by fine-tuning the threshold and sense trimmers.
But the main part of the problem is precisely in the subtle difference in voltages that occurs when drift occurs.
So now it is necessary to design a very delicate and efficient auto - tune circuit here.
The auto tune button will still remain a mandatory part of the control on the detector panel.
But with a good auto tune circuit, the frequent need to press that button will be reduced.
Let's say I remember when I had Minelab Explorer SE, it had an adjustable threshold, and when it gets a little stronger, by moving the probe over the earth's surface, changes in the treshold are clearly heard,
but the threshold always returns to the default value soon after.
Something similar is needed here.
Something like SAT, but this is not functionally the same thing, but it should look exactly like SAT.
For days I have been reviewing various lessons on Youtube about the use of opamps, comparators, various configurations with them ...
I understood and understood all that. But I was "stuck" around understanding what to choose as the best solution.
A particular problem for me is the fact that these are very small voltages and very small changes in voltage.
So I got into a situation; or to practically try all possible options, which would lead to terrible prolongation and loss of time, many new pcbs for testing, many trial & error situations, etc.

[ivconic]

Usually; when I get stuck somewhere around a solution: I don't hesitate to ask someone who knows things better than me.
So i asked Carl. He suggested me this.


"...I would focus on making the retune circuit better. The LF442 has a max
Ib of 20pA so with 470nF that is 42.5uV/s. You can get opamps with down
to 20fA max (LMP7721) so that gets you 42.5nV/s, or 100uV in 39 minutes.
A bigger cap will also help. Also pay attention to cap leakage, use the
lowest leakage you can find (this might limit the max cap size).

Finally, consider a differential retune: instead of connecting the (+)
input to ground, connect it to a same (470nF) cap to ground. When you
retune, short the cap as well; the drift on the 2nd cap will cancel the
drift on the main cap, limited by Ib matching and cap leakage matching..."

[ivconic]

Then we agreed that this might be very good topic for forum...
And here it is.
...
It all started with the idea of making some kind of terrain mapping device. But it ideally coincided with my long-held desire to make a real non-motion.
The plan is as follows; to make a good non-motion, completely analog design first, to make it work the best it can. And only then to translate such a design into a digital domain and add a processor, write code and add various options.
In stages, this project should meet more requirements. One of the phases can remain as a good analog non-motion detector.
But one man with his limited knowledge and resources; can't do much. More people united can do much more. So this is a good topic for the forum.

[ivconic]

Now that I have done this up to this stage and tested and experimented; I would have a few remarks and objections to point out.
Choosing a Tesoro coil with these inductances is not the "happiest" choice. I worked with these coils and various diy detectors for years and I realized one thing.
It is very difficult to get more "power" in a conventional way with a coil like this. Not impossible, but the solution would be forced with a lot of compromises.
So in the next revision I plan to change the type of the coil.
I have one small and very handy coil for the Quest X5 detector. It has very low inductances and will represent a very small "load" for the TX stage. But that immediately means changing the TX stage too.
It is always good to have an original and proven coil. And this coil for the Quest X5 proved to be very powerful even though it is small in size.
What is most beautiful in the whole story; is that it is a coil with specifications the same or similar with a large number of coils that can be easily found on the market today.
So it will be easy for most of those interested to get such or similar coil, instead of struggling and making it themselves.
Another objection; negative voltage generation. The classic Tesoro "chopper" is not able to generate negative voltage to reach the level of positive rail.
This was acceptable at higher supply voltages. But here we will try to make a modern design that will be powered by low voltages.
And the disproportion between the rails itself cannot be called symmetry.
Therefore, it is necessary to change the method of obtaining negative voltage.
Of course, the first thing that comes to mind is the already proven use of the 7660 or something similar.
As for the sound, I left that for last. Until my channels work perfectly; no need to waste time solving audio generation and audio output.
But I will say in advance.
My intention is to hear both channels separately, at the same time, to share the same audio output, and if it will be complicated, then let each channel have its own separate audio output, no problem.
Various logical actions can be added to the behavior of the sound, XOR, OR, AND ... this will ultimately be defined by the previous solutions in the previous stages.

[ivconic]

Okay, I think I've written a lot. I think I wrote everything that matters. Now it is up to you to get involved in the topic and give your opinions, suggestions, solutions, ideas ...

[gurdal]

Sir ivica you are so perfect.
I wish you success.
I'm following with excitement
Gurdal

[ivconic]

PCB from video was first one, with lot of catastrophic mistakes (as usual).
So i tried to fix all mistakes and here it is.
There is lot of emtpy space in the pcb, there was addtional opamp, gain stage, which was totally mistaken so i deleted it.
What you see on video is now present on this pcb, hopefully there are no more mistakes.

[gurdal]

thank you so much master

[ArchibaldSTM]

I have a question about the potentiometer DISC, --- will it not affect the TX frequency when tuning? On one of the forums there was information that this is what happens in Bandido. Thanks!

[Carl-NC]

I have a question about the potentiometer DISC, --- will it not affect the TX frequency when tuning? On one of the forums there was information that this is what happens in Bandido. Thanks!

Yes, slightly, but it should not matter as everything is synchronous.

[ArchibaldSTM]

Thanks Carl!

[ivconic]

"...I would focus on making the retune circuit better. The LF442 has a max
Ib of 20pA so with 470nF that is 42.5uV/s. You can get opamps with down
to 20fA max (LMP7721) so that gets you 42.5nV/s, or 100uV in 39 minutes.
A bigger cap will also help. Also pay attention to cap leakage, use the
lowest leakage you can find (this might limit the max cap size).

Finally, consider a differential retune: instead of connecting the (+)
input to ground, connect it to a same (470nF) cap to ground. When you
retune, short the cap as well; the drift on the 2nd cap will cancel the
drift on the main cap, limited by Ib matching and cap leakage matching..."

It would be good to find another opamp with similar specifications, older generation (if any), which would be easy to find in local stores.
At the moment, I can try the second part of the suggestion in practice. I'll have to add another 4066.
I will have two surplus switches left, which is good, I will need them later when I solve the audio.
And it would look like this:

[eduardo1979]

Reset

Hello.
First of all I want to congratulate Ivconic for their brilliant ideas and for bringing this forum to life. I found Ivconic's idea to build a non-movement VLF metal detector quite interesting and it aroused some interest in being able to contribute my grain of sand.
I think I have found a solution, but it would be necessary to try. I have added a small reset circuit that gave me very good results with very good stability. This is the Whites 3900 reset circuit, which I used in some old projects of mine that I did a long time ago.
It would be interesting that together we can improve this interesting project.