Concept Design

[ivconic]

[Carl-NC]

Sorry the pace is slow, my time is a bit scarce right now. There are two major issues that need some thought before designing a PCB. First is to investigate dynamic range requirements. For this, we need to know what is the largest expected signal we will see. There are two possibilities: hot ground, and a large nearby target. I am not too concerned about target overload, if it happens you just raise the coil. So I'm mostly concerned about hot ground, and I want to make sure the AFE does not overload when hunting the hottest expected ground. I think that, realistically, this will probably need to be figured out in testing because I don't even know what is the susceptibility of "really hot ground." It might be prudent to put an attenuator between the preamp stages. Normally you consider the maximum signal first because that sets your AFE gain, then do the noise analysis on the actual gain. I did the noise analysis first because I wanted to get a feel for ADC resolution. I have that now, and increasing or decreasing the gain isn't going to make a huge difference.

The other issue is more critical, and that is the matter of TX slope adjustment. This is akin to the voltage-vs-current phase error in VLF detectors that affects ground balance. For autotrack mode I'm not 100% convinced it has to be fixed in hardware; like VLF, you can deal with it in software. For manual GB not so much. I have not had time to investigate this. I know a couple of people have mentioned possible solutions and I need to track down those posts and review.

Just wanted to give an update on where I am at. Anyone else is welcomed to jump in on these issues and offer their input.

[ivconic]

Sorry the pace is slow, my time is a bit scarce right now. There are two major issues that need some thought before designing a PCB. First is to investigate dynamic range requirements. For this, we need to know what is the largest expected signal we will see. There are two possibilities: hot ground, and a large nearby target. I am not too concerned about target overload, if it happens you just raise the coil. So I'm mostly concerned about hot ground, and I want to make sure the AFE does not overload when hunting the hottest expected ground. I think that, realistically, this will probably need to be figured out in testing because I don't even know what is the susceptibility of "really hot ground." It might be prudent to put an attenuator between the preamp stages. Normally you consider the maximum signal first because that sets your AFE gain, then do the noise analysis on the actual gain. I did the noise analysis first because I wanted to get a feel for ADC resolution. I have that now, and increasing or decreasing the gain isn't going to make a huge difference.

The other issue is more critical, and that is the matter of TX slope adjustment. This is akin to the voltage-vs-current phase error in VLF detectors that affects ground balance. For autotrack mode I'm not 100% convinced it has to be fixed in hardware; like VLF, you can deal with it in software. For manual GB not so much. I have not had time to investigate this. I know a couple of people have mentioned possible solutions and I need to track down those posts and review.

Just wanted to give an update on where I am at. Anyone else is welcomed to jump in on these issues and offer their input.

1) RX AGC. Triggers only above defined level. Maybe adjustable.
(RX AGC can actually controll the TX power too)
2) Simplified manual GB + as fast as possible auto-track, switchable between.
...
How hot can be the soil in Africa and South America? (Tony would know this)
Is it Australian hot or more/less than that?

[ivconic]

I think I mentioned this on a forum a long time ago.
I think it was at the time when I was fascinated with Musketeer models (the memory of the first Musketeer trips my "switch" at the memory of this story).
When we say that soil is "hot"... I always wonder; is it as hot at night as it is during the day?
This should be investigated. Because it was shown and proved to me that there are differences. On my soil.
The sun's activities directly and radically affect the extent to which the soil will be "hot". It is an "external factor".
The "internal factor" is the soil composition itself.
But I noticed that the "external factor" enhances the effect of the "internal factor".
I first noticed this in a significant way right around the time of my first Musketeer Colt with smallest coil, ~18cm i think.
During the day on the sunny side, I couldn't find anything with it.
The same day the sun went down, walking the same path; suddenly I started making great discoveries!
Ok, this is a VLF I/B detector and here we are talking about a PI detector. There will certainly be a difference.
Why did I remember this? Well, Africa and South America are mostly in the tropical belt. With pronounced sunny activities during the day.
Has anyone ever thought of working those areas with a detector overnight?
Australia?
And people shall find it more pleasant and bearable to work when it's not hellishly hot under the sun.

[Willy Bayot]

The other issue is more critical, and that is the matter of TX slope adjustment. This is akin to the voltage-vs-current phase error in VLF detectors that affects ground balance. For autotrack mode I'm not 100% convinced it has to be fixed in hardware; like VLF, you can deal with it in software. For manual GB not so much. I have not had time to investigate this. I know a couple of people have mentioned possible solutions and I need to track down those posts and review.

Just wanted to give an update on where I am at. Anyone else is welcomed to jump in on these issues and offer their input.

The DYNAMIC XMIT current slope adjustment is indeed critical but it is even more critical to adjust it so that the signal offset is minimized (and kept constant) at the end of the receive chain, otherwise, there is not enough room to amplifiy without saturation in difficult grounds. If we do not find a good solution, forget about the CC concept!!
Sorry to insist but the electronic experts on this forum should seriously plunge their hands into this problem before going further on.

[ivconic]

I'll cut your work and hassle in short: forget about the CC concept!

[ionut_mtb]

I think I mentioned this on a forum a long time ago.
I think it was at the time when I was fascinated with Musketeer models (the memory of the first Musketeer trips my "switch" at the memory of this story).
When we say that soil is "hot"... I always wonder; is it as hot at night as it is during the day?
This should be investigated. Because it was shown and proved to me that there are differences. On my soil.
The sun's activities directly and radically affect the extent to which the soil will be "hot". It is an "external factor".
The "internal factor" is the soil composition itself.
But I noticed that the "external factor" enhances the effect of the "internal factor".
I first noticed this in a significant way right around the time of my first Musketeer Colt with smallest coil, ~18cm i think.
During the day on the sunny side, I couldn't find anything with it.
The same day the sun went down, walking the same path; suddenly I started making great discoveries!
Ok, this is a VLF I/B detector and here we are talking about a PI detector. There will certainly be a difference.
Why did I remember this? Well, Africa and South America are mostly in the tropical belt. With pronounced sunny activities during the day.
Has anyone ever thought of working those areas with a detector overnight?
Australia?
And people shall find it more pleasant and bearable to work when it's not hellishly hot under the sun.

I think black coils have disadvantages when using in sunny days, heat upsets the balance, produce noice. Using white coil is much better.(eg: Eric, Fisher aqua, minelab use white coils)

I’ve never thought about your statement regarding hot ground (higher temperature/heat). For sure it is a difference in detecting day/night. Like it is when detecting dry/wet soil. I remember once when i took home some bricks samples from an old settlement and started to wash/clean with brush. It was summer, temperatures were very high. When splashing with water the bricks started to sizzle/babbling because were so dry!

[Tinkerer]

I'll cut your work and hassle in short: forget about the CC concept!

OK, I have some consolation for you:
It is not continuous current, nor constant current, because it goes from + current to 0 current and then to - current.
So I call it a square current wave TX. But it is not really square either, it is trapezoidal. Just like Dave Johnson stipulated 20 years ago.
So after 20 years we finally got there.
We really progress at a snail's pace.

[ivconic]

I think black coils have disadvantages when using in sunny days, heat upsets the balance, produce noice. Using white coil is much better.(eg: Eric, Fisher aqua, minelab use white coils)

I’ve never thought about your statement regarding hot ground (higher temperature/heat). For sure it is a difference in detecting day/night. Like it is when detecting dry/wet soil. I remember once when i took home some bricks samples from an old settlement and started to wash/clean with brush. It was summer, temperatures were very high. When splashing with water the bricks started to sizzle/babbling because were so dry!

Makes sense but was not the actual case. Musketeer acted calm all the time, I did not notice any additional noise or false signals.
Simply it was not able to detect coins at the same spots where couple hours later detected more than dozen.
I clearly noticed the difference walking on the same path. I entered the shade because the sun was setting and more and more of the path was in the shade.
So the difference was not caused by changes to the coil but by changes to the soil. (See the rhymes!!!)

[ivconic]

OK, I have some consolation for you:
It is not continuous current, nor constant current, because it goes from + current to 0 current and then to - current.
So I call it a square current wave TX. But it is not really square either, it is trapezoidal. Just like Dave Johnson stipulated 20 years ago.
So after 20 years we finally got there.
We really progress at a snail's pace.

Willy's post from above makes more sense now. RX frontend will be delicate more than I expected.
I'll be honest and say what I expected from this idea.
I was expecting something really inventive. Something new.
Something no one has done before (maybe in White's they just scratched the surface of that idea).
I was expecting an "ultra weak" TX with a consumption of a few mA to a few tens of mA at most.
And I expected an RX with an ideal S/N ratio.
I've always been more attracted to that kind of thinking.
According to me (I may be wrong); all the gains can be realized in a very delicate RX design, where
first the most attention will be paid to the reduction of noise and offset.
Of course, this is from the point of view of someone who dealt more with VLF I/B detectors and close to none with PI detectors.
Tony remember what we talked about xx years ago. The idea then was to successfully detect a hypodermic needle (similar to those on modern mines).
That was a real challenge. One of the prerequisites was as "weak" TX as possible.
(Without sharp edges, as similar as possible to a sine, to avoid triggering one type of mine with such a signal.)
If we here would adopt such an approach for this project as well; I think the project would be even more interesting.

[Tinkerer]

Willy's post from above makes more sense now. RX frontend will be delicate more than I expected.
I'll be honest and say what I expected from this idea.
I was expecting something really inventive. Something new.
Something no one has done before (maybe in White's they just scratched the surface of that idea).
I was expecting an "ultra weak" TX with a consumption of a few mA to a few tens of mA at most.
And I expected an RX with an ideal S/N ratio.
I've always been more attracted to that kind of thinking.
According to me (I may be wrong); all the gains can be realized in a very delicate RX design, where
first the most attention will be paid to the reduction of noise and offset.
Of course, this is from the point of view of someone who dealt more with VLF I/B detectors and close to none with PI detectors.
Tony remember what we talked about xx years ago. The idea then was to successfully detect a hypodermic needle (similar to those on modern mines).
That was a real challenge. One of the prerequisites was as "weak" TX as possible.
(Without sharp edges, as similar as possible to a sine, to avoid triggering one type of mine with such a signal.)
If we here would adopt such an approach for this project as well; I think the project would be even more interesting.

I aim for a TX of 400mA, about 22Ampere turns/us for the AMX. I like your idea of maybe 40mA for the next project. Another challenge.

[ivconic]

Sorry to say this... but it is nothing inventive and new, then.
It's just Copy&Paste from what others already did much better than we wil ever be able to do.

[moodz]

The DYNAMIC XMIT current slope adjustment is indeed critical but it is even more critical to adjust it so that the signal offset is minimized (and kept constant) at the end of the receive chain, otherwise, there is not enough room to amplifiy without saturation in difficult grounds. If we do not find a good solution, forget about the CC concept!!
Sorry to insist but the electronic experts on this forum should seriously plunge their hands into this problem before going further on.

I dont think we should get caught up in slope adjustments ... its just not that critical. The ML GPZ7000 uses "CC" but they also use a Double D coil so they can balance out any ramp issues .. ( who said the ramp will always be linear ?? )
The required feedback loops to maintain CC would always be lagging the actual required correction this would introduce lag errors in actual feedback vs the required feedback which would produce artifacts resulting in increased noise or target falsing.
We cant fix problems by creating new ones that may not be problems anyway.
What sort of TX RX coil are we proposing anyway btw ?
moodz

[Tinkerer]

I dont think we should get caught up in slope adjustments ... its just not that critical. The ML GPZ7000 uses "CC" but they also use a Double D coil so they can balance out any ramp issues .. ( who said the ramp will always be linear ?? )
The required feedback loops to maintain CC would always be lagging the actual required correction this would introduce lag errors in actual feedback vs the required feedback which would produce artifacts resulting in increased noise or target falsing.
We cant fix problems by creating new ones that may not be problems anyway.
What sort of TX RX coil are we proposing anyway btw ?
moodz

I propose a DD coil configuration of about 8" overall diameter.
TX coil 70 turns, AWG22 fine strand tinned insulated wire, 400mA+400mA at about 200kHz - 250kHz resonant frequency. I prefer low Amps and higher turns.
RX coil of 300uH, 25 turns, AWG30 Kynar wire.
Graphite paint for shielding.
This seems to be a reasonable compromise, but I expect to build a few different coils to test the basic ideas.

What do you propose?
It would be interesting to execute yours and mine coils, to see what works best.

[Tinkerer]

Sorry the pace is slow, my time is a bit scarce right now. There are two major issues that need some thought before designing a PCB. First is to investigate dynamic range requirements. For this, we need to know what is the largest expected signal we will see. There are two possibilities: hot ground, and a large nearby target. I am not too concerned about target overload, if it happens you just raise the coil. So I'm mostly concerned about hot ground, and I want to make sure the AFE does not overload when hunting the hottest expected ground. I think that, realistically, this will probably need to be figured out in testing because I don't even know what is the susceptibility of "really hot ground." It might be prudent to put an attenuator between the preamp stages. Normally you consider the maximum signal first because that sets your AFE gain, then do the noise analysis on the actual gain. I did the noise analysis first because I wanted to get a feel for ADC resolution. I have that now, and increasing or decreasing the gain isn't going to make a huge difference.

The other issue is more critical, and that is the matter of TX slope adjustment. This is akin to the voltage-vs-current phase error in VLF detectors that affects ground balance. For autotrack mode I'm not 100% convinced it has to be fixed in hardware; like VLF, you can deal with it in software. For manual GB not so much. I have not had time to investigate this. I know a couple of people have mentioned possible solutions and I need to track down those posts and review.

Just wanted to give an update on where I am at. Anyone else is welcomed to jump in on these issues and offer their input.

I have a question about the RX coil:
For a differential input, we can use a center tapped RX coil or a floating RX coil. I simulated both and do not get much difference.
How does that look with the noise analysis?
About the “hot ground”:
I remember in Venezuela, there were areas where I had to sweep the coil at a height of about 12” above the ground not to get the detector saturated.
Some of the ground was magnetic, but there also was a certain amount of salt at some places. There was a lot of Manganese in the rocks and ground. Manganese oxidates brown or black, looks a lot like iron oxide. There is also a lot of bauxite.
So, all in all, without having any actual numbers, I feel we need to have a relatively large dynamic range. Using +/- 5V and an attenuator will help, while still using a gain of maybe 500 to 1000 on the AFE.