TineFPGA-GOLD Metal detector

[SaltyDog]

Saltydog, interesting project,
Have you managed to test the discrimination in real ground conditions?

cheers

Mdtoday

Hi, the project is in no condition to put in a box just yet ... still in prototype stage. Will box up when I am happy with the speed improvements I will make ..

In the meantime, I have brought the field inside to my workshop ... i.e buckets of soil/salty sand,black-sand, hot-rocks ... performs well in all ........... "so far"

[SaltyDog]

Why not replicate the Al foil 1x 2x 3x 4x tests. See what your fpga does with those slopes.

I have managed to find some foil.

One layer of foil is detected in the gold band, anything above 1 layer .. i.e 2 layers and up, go into the correct "Aluminium" band ..
I have tried up to 20 layers ..

Interestingly, a rolled up ball of foil about 1/2" in diameter, ends up in the Stainless steel slot(LED1)
.. the good thing is that apart from a single layer of foil, all the rest are distinguishable from gold/silver ..

[SaltyDog]

Here's a question for you knowledgeable guys:
Does anyone know where I can get a waterproof IP68, array of 8 super-bright LED's from .. just thinking about when I put this in a box.

I had a look around, but could only find the strip LED's that are aimed at lighting, they are not individually addressable ..

[dbanner]

I think that with an optimal front end and some super smart gymnastics on the fpga side of things, you may end up with a usable machine.

Over the years I've been looking at Mr. Green's experiments along with other testing, my conclusion is that this is a fickle game when it comes to a true discrimating pulse machine (VDI)

So far you are encouraged by some good results, so I am hoping that you can end up with success.

Looking forward to some videos down the road.

Good luck and stay safe.

[SaltyDog]

I think that with an optimal front end and some super smart gymnastics on the fpga side of things, you may end up with a usable machine.

With all due respect, it is usable as is ... I just want it even better ...

[SaltyDog]

I suspect the lack of response to the Earth's magnetic field is due to the reduced sensitivity of your current configuration. That is, direct sampling of the preamp output using a 10-bit ADC. A typical PI will have around 80dB more gain than you are using. This seems to be confirmed by the gold ring detection at only 6 inches, and the inability to detect a horseshoe magnet unless it's next to the coil.

Actually, I can detect my ring at 200mm at the moment, if I wind the gain right up, I get 300mm but I also get 50hz hum detection at that point. I will be adding in a notch filter to the FPGA to deal with that issue.
(However where I metal detect ... there are no pylons ...)

The horseshoe magnet was 2 foot from the coil ..

PS: I suspect it is the way I have used the filters (instead of integration) that has removed the EF ..

[Carl-NC]

Chart with stacked nickels and stacked quarters. Four quarters decay faster than 1 quarter at start of decay and slower latter in the decay with a PI. What happens when IDing stacked coins with a VLF?

Can a target thick enough to be effected by skin effect decay straight line on a linear X log Y chart?

Ironically, most White's VDI machines will ID 2 stacked quarters the same as a half dollar, and 4 quarters the same as a silver dollar. Other detectors will give somewhat similar results.

All targets are subject to skin effect. It's a more dominant effect on tau than the raw conductivity of the metal.

Here's a question for you knowledgeable guys:
Does anyone know where I can get a waterproof IP68, array of 8 super-bright LED's from .. just thinking about when I put this in a box.

I had a look around, but could only find the strip LED's that are aimed at lighting, they are not individually addressable ..

I would use an enclosure with a clear faceplate and put them under a bezel or label. Or use a single RGB LED and color-code the ID. However, if you plan to dive with it colors will look different at depth.

PS: I suspect it is the way I have used the filters (instead of integration) that has removed the EF ..

Yes, EF can be removed with high-pass filtering.

[SaltyDog]

I would use an enclosure with a clear faceplate and put them under a bezel or label. Or use a single RGB LED and color-code the ID. However, if you plan to dive with it colors will look different at depth.

I have a IP68 enclosure already, so I will see if I can somehow seal in a clear window into it.
I prefer the row of LED's, it gives more information about the target than a RGB LED ..

[green]

Ironically, most White's VDI machines will ID 2 stacked quarters the same as a half dollar, and 4 quarters the same as a silver dollar. Other detectors will give somewhat similar results.

All targets are subject to skin effect. It's a more dominant effect on tau than the raw conductivity of the metal. .

Log amplifier failed on my tester awhile back. Replaced it and did some testing to see if the repair worked. The target current should have decayed to zero by the end of the 2500us Tx. Tx off(maybe 1us pulse)then excites the target. Should have just skin current at start of decay. Higher resistance, steeper slope, lower TC which charts show. At start of decay, 1 quarter has a higher TC than the other targets. Would be interested where in the decay curve SaltyDog is calculating the slope, assuming he is IDing targets based on decay slope.

[waltr]

SaltyDog,
Have you seen this thread on Minipulse preamp improvment?
https://www.geotech1.com/forums/show...tly-improvable

Looks good for obtaining the fast recovery from the TX pulse.

[SaltyDog]

Hi,

Yes, I had seen that thread. I all trialing some other improvements with the preamp at the moment, working with another forum member on that part.

With some improvements, I have the main part of the discharge curve happening in 5us now .. you will notice there is no preamp saturation ..
I am still waiting on parts to finish some other idea's ..

Attached is preamp output, the blue trace is the TX pulse and finishes in 2us.

[Elliot]

Decay charts with some different materials.

(third chart)1x1in targets cut from aluminum can, lead sheet and copper clad board have almost the same decay rate. Big difference between 1x1in aluminum foil and aluminum bar targets. Big difference between 1x1in copper clad and piece cut from copper tubing.

Second chart is an example of decay slopes with Carl's 1x1in aluminum foil targets(to 16 layers, didn't try 32 layers)

Hi Green,

How are you producing those TC charts? i.e. what test gear configuration are you using?

[green]

Hi Green,

How are you producing those TC charts? i.e. what test gear configuration are you using?

https://www.geotech1.com/forums/show...567#post253567 Using Excel to get multiple targets on the same chart.

I first charted the decay slopes linear X log Y by recording amplifier out with a mono coil. A no target recording and a target recording, subtracted the no target recording from the target recording with Excel and charted. Wanted to try a log amplifier so I could see log amplitude decay on the scope. Either way gives similar results.

[SaltyDog]

Update: Just completed the port of my Verilog code and CPU code to the TinyFPGA-BX board ...

Next task is to finish design of pre-amp front-end and MOSFET drive ...
After that I will create a final PCB using Kicad ..

Stay tuned ..

[ricalbertvig]

I am attentive to your answer--everything goes very well