Competition!

[ivconic]

... OTOH, Auto-Mation-Assist's AGD development is a recent example of success, and there have been others like it (from Moodz, Deemon, Tinkerer, Teemo, etc)...

Define success?!
Is it something like this:
"...Hey guys, I drew a 1x1 mile schematic, here guys I drew a pcb with 56,000 invisible components... and it all works great for me! Take my word for it, I swear it is!..."
or is it success this:
"... Here's the beginning of the idea, part of the schematics... we're working on it... the results are promising... it's just not over... here it is... but now I'm going to change the story a bit and write about something else... "
or...
Or is success when you make a couple of detectors and they all work, you publish a schematic, a pcb, several versions of the pcb, xxxxx of them appear who start doing that work, there is a flood of questions, a discussion, a bunch of video clips, everyone gets another good idea along the way, add, subtract, correct... and that story spreads globally... one day, 15 years later, you see that in some tribe in Africa, a local healer is making a pcb according to your designs and starting mass production of those detectors.
In the meantime, you forgot that you ever worked on that detector, and then enthusiasts from Turkey, Afghanistan, Rwanda, Nepal, Germany, the Netherlands, the United Arab Emirates, Niger, South Korea... contact you and start flooding you with questions about something you don't remember for a long time.
Then you realize that detector has long since started to live independently, without you, and that it has its own life and path.
Well, every couple of years "he" makes you happy with a postcard from different parts of the world.

[ivconic]

Or is it success when your email inbox every once in a while shows something like this:

[Op04]

When AGD by Auto posted here, I watch the progress and see the unusual thin sample width 2-18us that time I was in a middle PI project, so I start to use that sample width to see how the result come. It did work for me! I believe AGD project will definitely works, I really hope to the author willing to make it more simple i.e choose cheaper chips so now we have a new PI with auto ground balance in analog way which is something new, I still can't believe why the experts here seem too quiet on that thread

[ivconic]

When AGD by Auto posted here, I watch the progress and see the unusual thin sample width 2-18us that time I was in a middle PI project, so I start to use that sample width to see how the result come. It did work for me! I believe AGD project will definitely works, I really hope to the author willing to make it more simple i.e choose cheaper chips so now we have a new PI with auto ground balance in analog way which is something new, I still can't believe why the experts here seem too quiet on that thread

All I would like to see is a shorter video demonstrating the basic functions and behavior on common targets. Then I'd probably decide to spend time on it, adjust the component list to those components that I can find easily, and then set out to have some fun during the dull winter days.

[Op04]

All I would like to see is a shorter video demonstrating the basic functions and behavior on common targets. Then I'd probably decide to spend time on it, adjust the component list to those components that I can find easily, and then set out to have some fun during the dull winter days.

Unfortunately Sir, for now I can't build the whole project coz it will need a serious attention and enough fund too and I still not have time to make through hole version for AGD and till now I still tryin to understand each section part of it

[ivconic]

Unfortunately Sir, for now I can't build the whole project coz it will need a serious attention and enough fund too and I still not have time to make through hole version for AGD and till now I still tryin to understand each section part of it

You said it!

[Op04]

You said it!

Ok...well we must not give up, may God help us all. Perhaps somewhere in the future, hope we can build any project we want.

[profesor]

It was a pity. I was hopeful about this title.
1- first a good coil design theory.
2-signal analysis and recommendations.
3- evaluation period and result.
I thought the topic would progress this way.
Then everyone would design their own circuit.

Joseph J. Rogowski.
He made two very good entries.
thanks.

[lucifer]

I would like to share another point of view about metal detecting.

1. Search location,
2. target types sought,
3. target time constants sought,
4. coil size,
5. detection method being used,
6. optimization methods to better find desired targets.

While detecting on the beaches in Ocean county and Atlantic county NJ I had to find ways to optimize my detecting effort to find more targets more easily.
1. Choose beaches that are densely populated
2. Choose a coil size that best finds desired targets
3. Use circuits that optimize the signal to noise ratio
4. Observe when the weather conditions result in sand and coins are being thrown up onto the beach
5. Use a half inch square wire mesh to make a 1 foot diameter by 1 foot tall sifting basket to quickly retrieve targets from sand
6. Place a magnet in a small plastic bag with a string attached to throw on the beach when I came upon black sand to see if it is organic or magnetic sand.

Once I understood how PI metal detectors work, I began to appreciate why the TX pulse lasts about 3 coil time constants. The total coil circuit resistance along with the coil inductance will govern the shape of the TX pulse with the current rising to almost horizontal when the TX pulse turns off. This ensures that any currents induced into targets at the initial current turn on have died out at about 3 time constants when the TX pulse turns off and the intended stimulus of the target occurs and then in a mono coil the TX coil becomes the RX coil. The time between the TX turn off and the RX turn on is called delay. The delay is based on how quickly the TX oscillations can be damped so the coil can be in the RX mode. The more capacitance that the coil sees, the lower value the damping resistance needs to be to quickly damp these oscillations. However, during the delay time, the induced eddy currents in the target are decaying and may not be there when the RX coil turn on. This is why I wrote the Fastcoil.pdf article. I set up my workbench with a signal generator and an oscilloscope to measure coil seen capacitance from coil winding capacitance, coil to shield capacitance and any other coil seen capacitance that I could minimize to reduce the delay. Coil to shield capacitance using a single layer of Scotch24 mesh is less than using a solid foil shield.

Being competitive with metal detecting requires an interest in selecting the right detector type and coil size for the desired targets, detecting location and environmental conditions. It also depends on how deeply you are willing to go to optimize your metal detector, what test equipment you have access to, and how deeply you are willing to dive into tinkering with your working metal detector or building one from scratch.

Learn about a concept called dielectric constant of a materiel. Lower is better as it loads the coil with less seen capacitance. Teflon insulation is about 2 while PVC is about 6. Remember, making a hand made coil could take some time but you can apply a low dielectric constant spiral wrap to secure the coil winding as well as provide a good spacer between the coil windings and the Scotch24 shield.

Abstract concepts used for a practical purpose become less abstract once you experience the results and visualize how the pieces finally all fit together.

I came to Geotech1, I researched, I wrote an article and I found some targets.

Joseph J. Rogowski

Joseph, your posts are always dead-on. You mention a lot of valid points that I completely agree with. How do you measure the coil-to-shield capacitance? Do you just add the shield and measure the coil self-resonant frequency again?
Regarding the TX pulse length of 3 (or more) time constants likely we can use that in opposite as well. I mean if less than 3 time constants pulse length is used, it might be still useful. Sensitivity to slowly decaying targets (like iron) will be reduced but fast targets won't be affected as much. That could be best used if coil is driven by square wave current and one is more interested in fast targets.

[MDfanatic]

It was a pity. I was hopeful about this title.
1- first a good coil design theory.
2-signal analysis and recommendations.
3- evaluation period and result.
I thought the topic would progress this way.
Then everyone would design their own circuit.

Joseph J. Rogowski.
He made two very good entries.
thanks.

From what I've seen and read, we have some brilliant minds on the forum. There's no doubt in that.

The point is, who has the knowledge, time and patience to design a circuit from scratch and is willing to start a new, collaborative project?
That is the real problem and mainly because these kind of projects are of no interest to someone who can make a sh1tload of money for the same time invested on a paid project.

Let's face reality, whoever has the knowledge to build something better than what's already offered (DIY or not) wouldn't waste the time to give it to everyone for free.
Only a handful of examples were succesful (Hammerhead, Voodoo project, Felezjoo PI afaik) and all of them had an altruist project leader who had the skills, time and patience (and huge cohones) to complete the project.

If I had the knowledge I'd do it myself but unfortunately my skills are that of an infant

[bbsailor]

Joseph, your posts are always dead-on. You mention a lot of valid points that I completely agree with. How do you measure the coil-to-shield capacitance? Do you just add the shield and measure the coil self-resonant frequency again?
Regarding the TX pulse length of 3 (or more) time constants likely we can use that in opposite as well. I mean if less than 3 time constants pulse length is used, it might be still useful. Sensitivity to slowly decaying targets (like iron) will be reduced but fast targets won't be affected as much. That could be best used if coil is driven by square wave current and one is more interested in fast targets.

Lucifer,

Read my fastcoil.pdf article by searching for the word fastcoil.pdf. In that article I explain the concept of distributed coil capacitance. Only about one quarter of the coil to shield capacitance affects the lowering of the coil self resonance with the shield added. The coil to shield capacitance is measured with an LCR meter with one test lead attached to the mounted coil shield and the other test lead going to one of the coil wires. Measuring the coil resonance with a signal generator and oscilloscope will give you a more accurate indication of your desired measurement values.

The TX pulse length is based on the coil inductance divided by the total TX circuit resistance (R) which includes coil R, MOSFET coil driver on resistance, coax cable R and any other R that adds to the coil R that influences the TX pulse current rise graph. After 3 TX time constants the current raises to about 95 percent of max and looks almost horizontal and then turns off to drop and create the main stimulus to the targets under the coil. The main reason why the TX pulse length is about 3 coil time constants is to ensure that any stimulus currents that may occur in any target under the coil from the initial early part of the TX pulse which is more vertical than horizontal, will have died out after about 3 coil time constants.

Target time constants are different from coil time constants. Targets that have low time constants like small gold nuggets require a coil with a low delay so there is still some target eddy currents after the coil delay when the TX pulse turns off, the TX oscillations have been fully damped and the RX mode of the coil turns on and there is still some decaying target eddy currents to be detected.

I hope I have answered your question?

Joseph J. Rogowski

[Carl-NC]

So I think the real challenge is this: Who is willing to lead a project?

No one, apparently. And that is the problem.

Ivica, you are the most vocal proponent of "doing something." Is there anything you would like to see designed, and are willing to lead the effort? That doesn't mean you have to know how to do everything, or be able to propose all the solutions. I think if you were to guide the project, people would participate. I certainly would.

[ivconic]

I am honored by the proposal.
But as always I have a short story to tell!


2013/4/5 I worked as "Quality Assurance Manager" in a software company from Switzerland. I was not a "boss" or a "leader", I had several people "above" me and several "below" me.
As a solid developer (for some areas) my task was to control the other developers and coordinate their work.
However, I had a couple of "bosses" that I had to listen to. I noticed then that it's terribly frustrating when your boss is someone who thinks he knows more than you, but actually knows much less!
Although I was well paid; work was very stressful because of that.
Taught by that experience; I'm SURE that most people who would have something to say constructively on the project will be constantly frustrated by my ignorance!
And that is not a healthy working climate.
So it is with great regret that I have to decline this honor and such a role.
Because it's clear that a few of you here really know a lot more than I do. It's stupid for me to "manage" anything.
But... that's why I can say what I would like the project to be and how to do it!
From my previous posts (who had the nerve to follow and read everything) it is clear what my DIY possibilities are.
I will only be interested in such a project - one that I can do right away in my workshop with my tools and materials. Everything else will be pure abstraction for me.
So here's a shorter list:
1) Through hole version and SMD 1206 version are fine.
2) Non-motion metal detector (PI or VLF I/B) with GEB, not mandatory a Disc, but advanced and improved auto zero cicruit is most desirable.
3) If PI, abilitiy to sample delay 10uS and bit less than that.
4) If VLF I/B, non-motion with advanced auto zero.
5) If MCU; STM32 familly at most. Atmega 328P is "thin" but can do a lot of side tasks.
6) Avoid expensive and "privileged to have" components. Use only common components.
Example: NE5534, NE5532, OP39, LME... LF familly, there are indeed nice specs opamps in those old series.
More advanced and richer experts can easilly replace those with newer and more expensive ones in their works.
7) Taking into account all; starting with "worse case scenario", use "worse" material for concept and those who can; will use it to transfer the same idea into most advanced high tech version for themselves.
Like I said before; those do not lose anything, in fact they gain in such case. Because what's work good with "bad" and "old" material; will work even better with "good" and "new" material.
...
Ok, that's my wish. I would like such apporach. Because I will be able to join and participate.

[ivconic]

Of course, with such "broad" things, one can never compile a precise wish list from the start.
I would add just one more suggestion.
That the same concept is developed in parallel both in the analog version and in the "digital" version, or. without and with the use of a microprocessor.
I even think that it is a bigger challenge to make a good project without the use of a microprocessor or at least with as little involvement of a microprocessor as possible.
Let's say, we will save a lot and gain a lot if we use the processor only for the generation of pulses and control signals (depending on whether it is PI or VLF I/B),
possibly for the generation of sound (vco and not just simple sound) and possibly displaying some information and parameters on the LCD .
And the essential signal processing should still remain in the analog domain.
Ok, it seems a bit like we're going back in time, because for years we've been moving towards greater integration and trying to have the processor replace as much analog hardware as possible.
But then appetites grow, on the example of AMX the problem is clear, we will want the highest possible resolution in signal processing, as many "heavy" filters as possible...
in the end we will realize that it is a vicious circle, we will not be able to do all the tasks so well without special dedicated integrated circuits, then we will
chase the best ADC circuit...remember the great story on the AMX thread!
One thing leads to another, a good ADC requires an even better mcu, a good mcu has its own complicated story, lots of tiny voltages, lots of regulators, pages and pages of code.
That story is a bottomless well! We're going to drown very soon!
And if we use the processor only for simple things, the challenge will be to design the rest with analog circuits, but everything will have flow, meaning and great chances of success.
And if we use the processor for simple tasks, then we have a much larger choice of processors. Anyone will be able to port the code to their favorite processor.
Or to write the code for his processor himself.
The processor is fundamentally good because it provides very precise clocks and pulse trains.
At the same time, several small jobs can be done simultaneously (if one core, in fact with a gap of only a few uS).
And thus replace a large part of the circuit.
But it will not be burdened with an excessive number of samples that need to be passed through very heavy filters.
Software filters are powerful, but they take up a lot of resources and slow everything down.
Hardware filters are simpler and much easier to implement.
Finally, the most important thing, and please don't get this wrong.
With a fundamentally good detector, we don't need to know 100% whether the "colored" metal is actually copper, silver, bronze or gold.
Because even if we want to, we will never be able to do it 100% accurate.
With an essentially good detector, it only matters that we have indications of whether the detected metal is "colored" or ferrous.
If it is "colored", we are interested in the further division, whether it is aluminum, aluminum waste, various foils, etc.
All other "colored" metals are desirable and it is definitely recommended to dig on such a signal.
So the division can be simpler and the Disc scale much simpler.
There is no need to develop a "all-knowing" Disc because in practice it has been shown that even the most perfect Disc circuit is no more reliable than an ordinary one.
Because targets are never alone and almost never "clean". There is always some "disturbance" next to the desired target that will affect the phase shift and thus deceive the Disc circuit.
In theory it's always better, but in practice all that great effort to make the best Disc is actually totally useless.
I have had a Deus for 12 years, had an ML Explorer, maybe a dozen other "digital" and "smart" detectors.
NONE could clearly discriminate most suspicious targets. I always had to dig all such targets.
At Deus, I only pay attention to whether the VDI number is above 55-60 or not. If it's above 55-60 then I'm digging. If not, I don't dig it.
Then why would we take a "million" samples and use some powerful ADC? In addition, we will be forced to control such an ADC with some wicked processor?
If we already know in advance that even the most perfect Disc will never be accurate enough!
The problem is not in the detector, the problem is in the medium we search with the detector. And that is the soil. The most unpredictable medium possible.
And it affects the phase so drastically that big differences can happen on every inch of the surface.
So if the Disk will require too many resources; it is completely unnecessary and useless to spend such resources on Disk.
A lot of words and sentences, I apologize but it is very difficult for me to explain the point.
Essentially, we don't need a disc, which will never be good enough anyway and will "eat" a lot of resources in the digital version or many components in the analog version.
Essentially we only need to know if it is iron or a "non-ferrous" metal.And if it is possible to do it in an easy way, to recognize aluminum.
The GEB is much more important. And if it is non-motion (my big wish) that it has a very advanced and fast auto zero circuit.
All this written is a good guideline when starting the design.
What resources and material should be spent on, because it makes sense, and what resources and material should not be spent on, because there is no sense and no benefit, no matter how advanced and ingenious it looks from the electronic side.

[lucifer]

Lucifer,

Read my fastcoil.pdf article by searching for the word fastcoil.pdf. In that article I explain the concept of distributed coil capacitance. Only about one quarter of the coil to shield capacitance affects the lowering of the coil self resonance with the shield added. The coil to shield capacitance is measured with an LCR meter with one test lead attached to the mounted coil shield and the other test lead going to one of the coil wires. Measuring the coil resonance with a signal generator and oscilloscope will give you a more accurate indication of your desired measurement values.


I hope I have answered your question?

Joseph J. Rogowski

Thank you. I have the article on my computer for at least ten years and I've read it a few times but somehow I've missed or didn't recall that part.