A little help in building demining system

[Qiaozhi]

Here's a PM I received today from allahyarzadeh

I thought it would be better to reply to these questions in the Tech Forum, as I'm sure others would like to put forward suggestions for this worthwhile endeavour:

Hi,
sorry to bother you with my questions,
you are well known for your knowledge in metal detectors and coils,
I'm currently designing a humanitarian de-mining system funded by NGOs and UN charter ,and to be distributed with the people still living in the mine fields,as you may know there are lots of people dying nowdays in these fields.
I've designed a PI which is simple and some how very slow. I use it to mines with very low metal objects in them,which are as small as 3mm,
first of all I have problem with my device stability,and then comes speed and then comes ground effect,my device detects the soil itself when it comes near it,I have to upgrade it before complete it's production and distribution among ordinary people.
I use a 2mH coil with size of 20cm x 16cm Ellipse shape with stranded 0.8mm copper wire,
I Use Bu808 BJT transistor as switch ,30v power voltage to be switched on coil, and I use no damping resistor ,instead one 330 ohm (pulse rated non inductive ressistor) and six back to back diodes(1n4148 after them one 1k ohm resistor to ground and then it is my amplifier,one nasty old LM709 amplifier and then some S&H and other amplifiers until it reaches to ADC( a 18bit) .
in this configuration my 330 ohm resistor became very hot and the device need about 15 minutes to be completly stable,which is not good.
in analog amplification I have about 110db gain from main signal to detect small changes.

I use so many Technics to achieve more depth with increasing the back EMF peak voltage as high as 1000 volts and also in digital parts sampling from many areas and many pulses and then using some statistical Technics to omit noise and take out the real signal which has made my device too slow.

in this topology I had many high current consumption due to very high peak voltage and strong field.so I couldn't increase the frequency more than 100Hz.

Until I introduced with this forum and Guru's Post here I tried to use some mosfets with cascode we had a discussion here:
http://www.geotech1.com/forums/showt...879#post208879
http://www.geotech1.com/forums/showt...879#post208879

and I tried to use much less inductance for my coil and using a damping resistor ,using differential reading of signal ,
but this new one is really fast in responce but it's depth is not more than my previous design and stability is not ok again.

I wanted you to help me in this and answer some of my questions please ,
I know you are busy any of them you could answer please.

1- does more peak voltage guarantee me more depth?
2- what is optimal inductance of coil? I'm aware of low time constant of small gold nuggets,but I'm not going to find gold.I'm going to find small pieces of steel or ferrus metal, cr6 or sth.
3- how I should calculate the damping ressistor needed for damping the coil?what are the equations?
4- can I use blocking Fets instead of resistor and diodes,because I have many noise from there,
5- how I can achieve more than 120db gain for the main signal?(counting 24bit adc)
6- how I can undermine the ground effect which my device sees the earth as metal object when I near the coil to ground?
7- what frequency should I use? my previos device used 100 Hz which was very low for midpass filter and my 2nd circuit was 500hz ,what's the relation between frequency and coil and other specifications?
8- How I should build my coil? double d? monocoil? it is very better for me to recognise both right and left signal with double coils because before hitting the mines head with coil it could detect it.(it's some how vital)
9- do I need to shield my coil?with what? aluminium foils?
10 - I ordered a high voltage battery of 48v Li-ion battery with 5000mAh capacity,and I use this voltage to switch on my coil,to have more depth and more field, is it right?
11- I have a doubt about pulse signals I have one circuit with constant pulses and sample pulse, and in my second design I use to send some big pulses and some small pulses. What is the best timing and pulse?

----------------------------------------------------------------
My reply:

In summary;
You want to be able to detect targets as small as 3mm diameter. The target material will be small pieces of steel or ferrous metal, such as cr6 or sth. I assume cr6 refers to chromium plating, but what is sth?
There are problems with the current design involving slow response speed and ground effect (detects the soil).
The coil (20cm x 16cm) has 2mH inductance, and is wound with 0.8mm diameter stranded wire.
The TX coil transistor is a BJT, 30V supply voltage, and no damping resistor!
Flyback voltage reaches 1000V, and the pulse rate is 100pps.

Firstly, what is the decay constant for a 3mm diameter ferrous target?

Using the equation:

Where the conductivity of iron (Fe) is:

Hence:

= 2.95us

Therefore the eddy currents in the target will have completely decayed away after 5 * 2.95us = 14.75us

It has been recommended in some papers that you should aim for a main sample delay of one-tenth of the target object. In this case it appears we would ideally require a main sample delay of ~1.5us, which is extremely aggressive. Clearly this is not achievable with a 2mH coil inductance, a large coil current, and a low TX pulse rate of 100pps. I find it difficult to understand how it is possible to detect such a small target with the current setup. Unless it has something to do with not having a damping resistor. In which case, the RX waveform will be a decaying sinewave, and the detector is not really a PI after all, but more like some early Tandy and Heathkit designs that utilized this approach. This would also explain the problems with ground effect, as it's basically a TR-type detector without a GB channel.

The best way to determine the damping resistor value is to use a damping resistor tool. This consists of a 10k trimmer in series with a 200R resistor, and both in parallel with a 2k resistor. This gives an adjustment range from 181R to 1k672.
Monitor the preamp output, and adjust the trimmer for critical damping or slight undamping. Measure the final value, and replace with a fixed resistor.

Regarding TX output power - please read the attached document: "Coil Voltage vs Depth - from Tec".
Also, you should be aware that such a small target does not require a huge coil current to become saturated, and extra depth is subject to the sixth power rule.

I suspect the "ground effect" you are experiencing is either due to coil-to-soil capacitance, or a simple TR effect due to the damping resistor being missing. Once you get the coil critically damped, you could add a second sample pulse to achieve ground balance, but this means you will lose some depth, and create a movable target hole in the response. It might be worth exploring an alternative technique put forward by Teleno (reading from post #69 onwards) -> http://www.geotech1.com/forums/showt...-transmit-time

Personally I think you should reduce the coil inductance considerably, to at least 300uH, increase the TX pulse rate to as high as possible (considering the limitations imposed by software processing overhead), concentrate on reducing noise in the RX channel, and decide on a suitable method for ground balance that does not reduce depth or miss targets. You will also need to shield the coil in order to use early sampling. Suggested reading -> http://www.geotech1.com/pages/metdet...s/FastCoil.pdf

Anyone who wants to chip in, please do so.

[ODM]

I feel like pointing out the "elephant in the room" ... An affordable detector is probably highly sought after, but I cannot suggest that it should be someone's first foray into designing a metal detector. Humanitarian demining is an admirable cause but I would have to discourage against such ideas, this is the kind of work that should be given (and left) to experienced engineers, not amateurs. Reliability in short and long term, design for manufacture, confidence of detection, self-testing features, all of this needs to be considered.

Also, demining is a broad and serious topic. There are also other hazards left over in warzone such as unexploded ordnance that may be significantly deeper than a buried landmine, so many types of equipment will be necessary. Metal detectors with little or no training to use them, or other protective gear and probing equipment, and lack of organization to work together in a coherent manner, can be even more hazardous than landmines alone if they give people a false sense of security to wander into known landmine hazard areas!

If this is an internationally backed project or even a government level one, it should have the funds to actually provide decent field proven detectors, even NATO surplus or something, if they can also provide protective equipment and training and organize mine action plans. But it's understandable to feel frustration over not having the funds, training or equipment to deal with the problem. Metal detectors for demining can be expensive and there should be more affordable options. But the detectors alone are not a large part of such actions budget when training and protection and salaries are accounted.

[Funfinder]

Carl V. Nelson - (Metal) Mine Detection
http://techdigest.jhuapl.edu/td/td2501/nelson.pdf


You can read here that it might be the best idea to use a non-motion detector
that is able to find cavity if you wanna find those ultra low metal plastic "boxes".
But this only works if the soil where this stuff is buried is mineralized enough.
On the other hand a PI will not find such small 3gram or 3mm metal pieces at all there.

For detectors that can not find any cavity a pretty small coil is a must !!! 10cm-20cm (the deeper buried, the larger the coil)!

Be careful with new war-zones !!! They might be full of radioactive ordnance and dangerous dust!!!
You might get lung-cancer there! Don't go there without X-ray-meter!
And it's important to be informed what else stuff was in use at this region.

[Hammerhead]

Egypt is considered as the most mine-affected in the Western Sahara because of German and English in World War II and now refuses to England and Germany, giving Egypt maps of the minefields and Malk because many people die

[Davor]

There are a few unknowns here.
From previous communication with allahyarzadeh I was unable to grasp what area he is going to detect, so we have no idea what kind of mineralisation he has. In case of seriously viscous soils nothing from the above will help anyway. But we continue to have no idea about it.
AFAIK, the current crop uf NATO uxo detectors, Vallon, use coils with somewhat over 1mH inductance (Eric Foster has one), mostly because small amounts of readily detectable metals with long-ish tau in the fuze (aluminium or brass). In my backyard such a mine is an infamous "meat-pâté", or a balls-breaker, coming from the intended results that are primarily to maim the legs and appendices.
I'm not too familiar with the technology, but NATO's choice of a slow-ish machine over others speaks volumes about the vast majority of anti-personnel landmines.
I am, however familiar with a construction of an anti-tank land mine that employs a metal plate to focus the blast. That metal plate is in fact pressed steel. I'd expect these to show more readily on a magnetometer than a metal detector. However It borders with humanitarian side of demining business, these are more of a military matter. They are destroyed by some other means than merely detecting them, as more often than not those are booby trapped.
In any case a regular Joe, regardless of nobility of his intentions, has no business roaming minefields, with or without a detector. People get crippled there. Or die.

Odd as it might seem, large portions of Croatia were demined thanks to the maps provided from "the other side". After the mines' military function expires, there is no excuse in this world or next not to exchange the maps and prevent further losses.

[Qiaozhi]

AFAIK, the current crop uf NATO uxo detectors, Vallon, use coils with somewhat over 1mH inductance (Eric Foster has one), mostly because small amounts of readily detectable metals with long-ish tau in the fuze (aluminium or brass)

Yes - I'd heard that was the case; which sort of implies that the intended target has a longer time constant than 3us.

According to allahyarzadeh, this effort is being funded by NTOs, so essentially this is a privately funded venture.http://www.investopedia.com/ask/answ...ganization.asp
ODN is correct about the "elephant in the room", but I guess these people are fairly desperate to rid the area of land mines, and perhaps (maybe due to being a remote location, or some other reason) they're not getting help in this matter.

Perhaps allahyarzadeh could post a schematic of his current design here in this thread, which would make it a lot easier to discuss the feasibility of this approach.

[ODM]

Incorporating a microcontroller is a very good start to the design reliability-wise, affording for self testing and monitoring, and also suggests they have contacts with manufacturing and supply chains. Producing detectors for demining use definitely needs a proven chain of parts supply, so ebay etc. are off the table as a starting point. So are other often-faked items, like generic microcontroller boards. A western supply house for components, but the pcbs themselves can be made in china. Producing the boards locally is an option as long as there are senior workers to guide the assembly process.

Something that is rarely done for hobby electronics is artificial aging, temperature testing and shake testing. These can be "sort-of done" by themselves for each production apparatus and I believe any university with electronic engineering would be glad to do an EMC run for a few sample detectors for them, if only for publicity. We can probably collectively guide the effort here but I sincerely believe they would be better off contacting a proven manufacturer for guidance regarding design and testing. They might even make an offer for existing detectors.

I would be interested to see what kind of action plan they do have and what parties they are working with to train deminers and supply them with equipment. Proper protection gear is very much needed and there are plenty of unscrupulous parties out there selling both bogus detectors, and protection equipment, that isn't up to the task. ( for examples https://en.wikipedia.org/wiki/ADE_651 )

[Carl-NC]

While I agree with the concerns about demining safety, I'm guessing that allahyarzadeh isn't going to be selling his design directly to the villagers. Training and safety is someone else's responsibility.

A lot of humanitarian demining involves older mines that are easier to detect, and often in reasonable soil. The White's XLT was selected by one organization for this. I don't know enough about these mines or the detecting conditions, but I'd guess that a fairly simple design might work just fine. I do know the general requirements for such a detector are 1) very cheap 2) uses common batteries (AA) and 3) is super-easy to use.

Detecting modern minimum metal mines is a pretty tall challenge. The Garrett Recon comes with a test piece which is a small stainless firing pin embedded in the end of a plastic rod that represents the depth of the pin (about 10-12cm I recall); the TDI was barely unable to detect it. Most high-end expensive demining units are intended for minimum metal mines.

From the description of the mines and the test target, it sounds like allahyarzadeh is trying to shoot for the hard design. If so, then I'd go for short pulses and a high pulse rate. Also bipolar pulsing to avoid magnetic triggers. However, I'd also say this kind of project is beyond the scope of a beginner.

[ivconic]

While I agree with the concerns about demining safety, I'm guessing that allahyarzadeh isn't going to be selling his design directly to the villagers. Training and safety is someone else's responsibility.
A lot of humanitarian demining involves older mines that are easier to detect, and often in reasonable soil. The White's XLT was selected by one organization for this. I don't know enough about these mines or the detecting conditions, but I'd guess that a fairly simple design might work just fine. I do know the general requirements for such a detector are 1) very cheap 2) uses common batteries (AA) and 3) is super-easy to use.
Detecting modern minimum metal mines is a pretty tall challenge. The Garrett Recon comes with a test piece which is a small stainless firing pin embedded in the end of a plastic rod that represents the depth of the pin (about 10-12cm I recall); the TDI was barely unable to detect it. Most high-end expensive demining units are intended for minimum metal mines.
From the description of the mines and the test target, it sounds like allahyarzadeh is trying to shoot for the hard design. If so, then I'd go for short pulses and a high pulse rate. Also bipolar pulsing to avoid magnetic triggers. However, I'd also say this kind of project is beyond the scope of a beginner.

To be able to compete at all for funds from NGOs and UN charter; a complete project must be applied along with training&safety plans.
It's very serious problematic and story must be complete. I seriously doubt that just a piece of electronics will do enough to get money from funds.

"...Detecting modern minimum metal mines is a pretty tall challenge..."

You may say that again! It's the "holly grail" of commercial projects now days.

"...the TDI was barely unable to detect it..."

Not only the TDI; most of others too.
Even the Deus too...

"...I'd go for short pulses and a high pulse rate..."

That's what i thought, but than i saw "2mH coil" !!!?

"...this kind of project is beyond the scope of a beginner..."

Quite beyond! Unfortunately.
But i would really like to see AF108 tested more detailed on stainless firing pin kinda targets.
I don't have experiences with it at all, but schematic is looking quite interesting.

[Koala]

Modern mines. If you become too good at detecting them want the manufacturers of them change their design to detect the detector.

[green]

I measured the time constant of a 1/4 20 x 3/4 inch stainless bolt to be 2.2usec. Does anyone know how big the stainless pin is?

[Koala]

https://en.wikipedia.org/wiki/Minimum_metal_mine

[Qiaozhi]

... and some more:
https://en.wikipedia.org/wiki/List_of_land_mines

[Koala]

No metal or less than 1 gram. Wow that's going to be tough to impossible. Going to that much trouble I bet the 1 gram is both an alloy that's hard to detect and in an orientation and depth that makes it doubly difficult.

[ODM]

Here's some sources for what the sorta-state-of-art equipment looks like
MetalDetectors_Catalogue2005.pdf

Landmines generally present these kinds of targets
DSTO-TN-0649 Landmine detection test targets.pdf

Some testing methods and results for above detectors
Technical_Note_JRC_Bloodworth.pdf

This one's too big to attach on the forum but found an online copy, it contains some interesting material related to designing proper ground balancing.
http://www.gichd.org/fileadmin/pdf/L...-Terrascan.pdf