Beach detecting detector

[urge38]

I cant help but wonder, that with all the research that fisher did, with there AQ, if it was possible to do a decent job of making a PI with true pi depth AND discriminate like a VLF then fisher or even minelab would have done this by now, ive watched the only two videos I could find on the Voodoo and I have to say, i dont think it does not do a good job of iron rejection of large iron targets

[Orbit]

https://www.google.com/url?sa=t&sour...kQT6dUevYgMFHC

[Orbit]

Taking a sample during the tx pulse is one of the methods of discrimination with a dd coil, or the hybrid approach used by Minelab ! ...

[Teleno]

There are 2 ways to discriminate with a PI. First, sample during the TX pulse where you have access to a reactive signal. This requires the use of an IB coil and is how Minelab does iron reject on their PIs. It is more akin to VLF and not as deep as the PI side so iron ID is not available for the full PI depth.

The second way is to analyze the flyback decay. A 2-sample analysis (TDI, ImpulseAQ) is how viscous ground is rejected (a form of discrimination) and results in low/high conductivity ranges which is pseudo-discrimination. If you want better discrimination then take more samples. Most coins & rings have fairly well-behaved exponential decays while iron targets do not. It's possible with 3 or 4 samples to do a decent job of iron reject. However, later samples occur at a weaker signal level so discrimination still won't work at maximum depth.

Edit: The Voodoo is a hybrid, not strictly a PI. There are several ways to make a hybrid PI/VLF but generally the VLF side retains the depth limitations of VLF compared to the PI side. Technically, sampling during the TX pulse (above) is a hybrid approach.

I believe there's a third way. Using a square Tx current, sampling after the rising edge gives a pure dI/dT eddy signal because magnetization hasn't occurred yet. Sampling after the falling edge gives a sum of the eddy signal plus the magnetization decay.

[Altra]

I believe there's a third way. Using a square Tx current, sampling after the rising edge gives a pure dI/dT eddy signal because magnetization hasn't occurred yet. Sampling after the falling edge gives a sum of the eddy signal plus the magnetization decay.

Could you post a simple diagram of your current wave form and sampling point?

Thanks

[Carl-NC]

I believe there's a third way. Using a square Tx current, sampling after the rising edge gives a pure dI/dT eddy signal because magnetization hasn't occurred yet. Sampling after the falling edge gives a sum of the eddy signal plus the magnetization decay.

Good point. It still requires an IB coil but would overcome the VLF-like depth limitations. Unlike the GPZ approach, you would want independent bipolar pulses, like this:

[yatahaze323]

Hey Carl, are there any current PI detectors that you know about that utilize the square wave to discriminate iron?

[Tinkerer]

Good point. It still requires an IB coil but would overcome the VLF-like depth limitations. Unlike the GPZ approach, you would want independent bipolar pulses, like this:



[ATTACH]57975[/ATTACH]

I am not sure what is meant with "magnetization"

Starting at the 0 current level, the magnetic domains are lined up with the Earth magnetic field.
Then we apply current up to the high level. The existing magnetic alignment in permeable targets is changed to the new field alignment. The change of the magnetic field induces eddy currents in the conductors.
Reached the high level, the current stops to change.
The magnetic field becomes static. Existing eddy currents decay. The magnetic domains settle in the new static magnetic field alignment.

The same happens at every change of current.

The amount of change from the magneto static situation depends of the field vectors and field strength of the Earth Magnetic field and the field vectors and the field strength of the coil TX field.

This is my understanding of the process.
I would be happy if somebody could explain to me where I am wrong.

[Carl-NC]

I'm assuming that viscous magnetic material will behave differently during the "charge" event than during the "discharge" event. Whereas eddy exponentials do not.

[Teleno]

There are 2 ways to discriminate with a PI. First, sample during the TX pulse where you have access to a reactive signal. This requires the use of an IB coil and is how Minelab does iron reject on their PIs. It is more akin to VLF and not as deep as the PI side so iron ID is not available for the full PI depth.

The second way is to analyze the flyback decay. A 2-sample analysis (TDI, ImpulseAQ) is how viscous ground is rejected (a form of discrimination) and results in low/high conductivity ranges which is pseudo-discrimination. If you want better discrimination then take more samples. Most coins & rings have fairly well-behaved exponential decays while iron targets do not. It's possible with 3 or 4 samples to do a decent job of iron reject. However, later samples occur at a weaker signal level so discrimination still won't work at maximum depth.

Edit: The Voodoo is a hybrid, not strictly a PI. There are several ways to make a hybrid PI/VLF but generally the VLF side retains the depth limitations of VLF compared to the PI side. Technically, sampling during the TX pulse (above) is a hybrid approach.

I've implemented the second method with interesting results.

The logarithm of the ratio between two consecutive samples allows at least an educated guess of what a target can be. The distance is 10 cm. The objects I'm using are the most common finds for everybody I guess.

The method works best when the max dimension of the targets is parallel to the coil plane.

[Qiaozhi]

It would be interesting to see how this method performs off the bench and in the real world.

[Teleno]

It would be interesting to see how this method performs off the bench and in the real world.

I took the detector to the beach yesterday and my subjective impression is around 70% correct. But then again I only found mostly nails, sometimes id'd as caps or coins. Foil is correctly 100% of the time. Caps get close to that.

I don't know if that's better than VLFs for this sort of guesses cause I've only tried PI's.

[Teleno]

I've implemented the second method with interesting results.

The logarithm of the ratio between two consecutive samples allows at least an educated guess of what a target can be. The distance is 10 cm. The objects I'm using are the most common finds for everybody I guess.

The method works best when the max dimension of the targets is parallel to the coil plane.

Actually assuming a simple exponential the target's tau can be calculated easily as where is the time between consecutive samples s1, s2.

In my prototype is 32 us minimum because of the rather longish conversion time of the Atmega328P ADC. This means that s2 is rather weak. Shorter times would give a more accurate reading.

In any case there's an overlap between the taus of coins and nails especially when their longer axis is perpendicular to the coil.

[Carl-NC]

As I mentioned before, 2 samples are sufficient for identifying well-behaved eddy responses, such as with rings, most coins, and even foil. Nails behave differently because of the combined eddy & magnetic responses, so 2 samples are not enough to distinguish them as non-pure-eddy. With 2 samples they will always look like some coin or another.