PI detection - an alternative approach?

[6666]

I've got a circuit ready and sent for some wire. One more question; Is the size of the coil important when I am looking for changes in the discharge curve using different metals as targets?

Are you going to publish your circuit ?

[Tec]

Hi,

Noticed this post.

My detector Spi-MX pinnacle, measures the time taken for curve to discharge.

It measures in at least 2 places (4 places) to calculate conductivity.

Disadvantage on uni-polar pulse is that shift due to earth magnetic field and noise can-not be cancelled , where as when sampling you can take a later sample and subtract.

The Spi-MX pinnacle avoids this problem by bi-polar pulsing (pulsing in one direction, then the other) adding these two signals cancels out noise and magnetic field signal..
i.e any signal arriving at the coil will drive a current in one direction. This will increase one of the pulses, and decrease the other pulse, so the effect cancels.


Advantage of measuring time is increased dynamic range, ie you can't max-out the measurement, like you can with a a sample.

Tec

[Qiaozhi]

Advantage of measuring time is increased dynamic range, ie you can't max-out the measurement, like you can with a a sample.

That's an interesting observation.

[Teleno]

My detector Spi-MX pinnacle, measures the time taken for curve to discharge.

Googling "SPI MX" shows no metal detectors. Any pointers?

[pebe]

Are you going to publish your circuit ?

I haven’t got a finished circuit yet. It is evolving in the light of comments I have been receiving, but initially it will be a PIC programmed to give just a variable pulse width at a variable frequency. Pulses will switch on the coil via a FET and the flyback will be clamped with diodes as normal. Then I can try various targets with each variable and get flyback curves. Using a storage scope they can be filed and I can then start comparing them to find what is or is not important. When I get meaningful results I can finalise the circuit and post it.

My thoughts at the moment are to make the circuit as simple as possible. I intend to use a 12F629 with an external 20MHz crystal oscillator. The internal timer counts every fourth pulse of the oscillator, so I can get a timer resolution of 200ns. Hopefully that will be good enough – if not I can use an external counter that will improve resolution by a factor of 2 or 4.

Each measurement will be taken when the flyback voltage has fallen to 0.3V, using an op-amp as a comparator. To combat noise, all recorded times will be the running average of the last 16 measurements.

In use, the program measures delay time every cycle. It records the time (with no target) every 2sec as being the time reference of ground effect with no signal, ie. the ‘norm’.

As the area is swept, and a target is found, the program continues reading till the target has passed, then after 10ms(?) the next reading updates the norm. Only then are the LED and tone activated. Once the target has been found the operator holds in a button to stop the norm being updated and can then go back to examine the target signal.

Indicators: An LED normally flashes twice a second. When the signal time is shorter than the norm the rate will increase to 10 times a second; for longer times it is on all the time.
The audio signal has not yet been decided. Changes to volume and pitch are both possible.

That’s about it so far, because so much depends on the initial tests.

[Thomas]

Googling "SPI MX" shows no metal detectors. Any pointers?

https://www.google.com/webhp?sourcei...0pi%20detector

[Teleno]

https://www.google.com/webhp?sourcei...0pi%20detector

Broken links, disappeared sites and vanished videos in YouTube. The "mystery" PI detector.

[Thomas]

You're right, Looks like a major fail to me.

[Qiaozhi]

The details can be found here -> www.tctec.net/spimx/

[Teleno]

The details can be found here -> www.tctec.net/spimx/

Thanks! Hardware is very conservative, a run-off-the-mill bridge coil driver with the usual R-D input combination.

[6666]

That’s about it so far, because so much depends on the initial tests.

Thanks Pebe