Let's made a PC-base metal detector with usb interface !!!

[Davor]

in digital ? – NO !!!

Why not? You may always go for successive approximation.

[Aziz]

Oooops, I've forgotton to use the +5V USB port for the controller power supply. You can share the power source of your Tablet/Netbook of course and 5 V/100 mA is fairly enough power for everything. So you wouldn't need an external battery in this case.

(The sound card output power source using the transformator and bridge rectifier thing... shame.. shame.. *LOL*, no, you can't get enough free USB ports: GPS, external sound card, ... )

I have to buy some USB connectors & cables next time..

Cheers,
Aziz

[Aziz]

But there is - in zero crossing. Besides, you do not need a nice sine, a step function is just fine.

Yep, it isn't necessary to have perfect sine.

[Aziz]

FFT? -> there is no correlation, there is no synchronization with TX
(games on a netbook -> http://www.md4u.ru/viewtopic.php?f=1...0%BA&start=150 )

Yep, (complex) FFT is a good start. You would need it with the Hilbert transformation approximation either (the discrete Hilbert transformation).
But you can do it without Hilbert transformation as well.

All we need is:
sampled RX (this is the RX signal on the left channel)
sampled TX (this is the TX reference on the right channel)
TX-I (=TX (in phase), this is the TX reference on the right channel)
TX-Q (quadrature of TX reference, 90 ° phase shifted TX reference, must be extracted from TX)
Magnitude of TX (demodulation required)
Demodulating RX with TX-I/TX-Q using complex Lock-in amp
-> you get magnitude and phase of RX referred to TX reference
Note, that RX demodulation is dependent on magnitude of TX reference. You have to compensate all the effects happening to the amplitude of TX reference (as we don't have a regulated TX oscillator, we have to do it in software now ).

This is just one of the more possible ways. So there are more ways to arrive Rome in Italy.

Just focus on the following basic topics:
- (complex) FFT basics
- 2d-vector math (note: I & Q pair or Im/Re pair can be considered as a 2d-vector)
- 2d-matrix math (how to rotate a 2d-vector by 90°? -> hey, you can do a phase shift in the complex domain)
- window function basics (avoiding spectral leakage)
- Lock-in amplifier basics (demodulating RX with TX-I & TX-Q reference)
- filter basics (can be easily performed in the FFT domain)

All the information there is buried in RX and TX reference signal. The RX signal is synchronised to the TX reference however. There is enough information to solve all the issues. That's the reason, why the simple oscillator does not need any regulation.

Regarding the link: Sorry, I don't speak Russian. I wish, I could understand it.

Cheers,
Aziz

[Sergey_P]

in zero crossing

in digital ? – NO !!!

[WM6]

Another link from Ru site:

http://www.md4u.ru/download/file.php?id=12268&sid=c411963e5478eee07a9e325628b39633

[Davor]

there is no synchronization with TX

But there is - in zero crossing. Besides, you do not need a nice sine, a step function is just fine.

[WM6]

FFT? -> there is no correlation, there is no synchronization with TX
(games on a netbook -> http://www.md4u.ru/viewtopic.php?f=1...0%BA&start=150 )

Thanks Sergey:

http://www.microsofttranslator.com/bv.aspx?from=ru&to=en&a=http%3A%2F%2Fwww.md4u.ru%2Fviewtopic.php%3Ff%3D11%26t%3D3772%26hilit%3D%25D0%25BD%25D0%25B5%25D1%2582%25D0%25B1%25D1%2583%25D0%25BA%26start%3D150

[Sergey_P]

generated digitally

FFT? -> there is no correlation, there is no synchronization with TX
(games on a netbook -> http://www.md4u.ru/viewtopic.php?f=1...0%BA&start=150 )

[Aziz]

That's the reason for, why the complete hardware part gets so much trivial and simple.

But the detector can outperform most of the competitors with ease. Even without having an RX amplifier.


Comeon guys, it's the right time to enter the DSP technology. You have a hardware solution, which is completely free (public domain) and you can use a PC platform to develop the complete VLF/LF detector technology in the software.

Aziz

[Aziz]

is a special case...

Where pi/2 in the oscillator?

Not required.

The quadrature TX reference is generated digitally.

[Sergey_P]

you can use the Hilbert transform to create the pi/2 (90 °) phase

is a special case...

Where pi/2 in the oscillator?

[Aziz]

How to perform the Hilbert transform, using the oscillator?

Hilbert transform? WTF, who is Hilbert? What are you talking about? Oh man!, it's really beyond my scope.


Of course, you can use the Hilbert transform to create the pi/2 (90 °) phase shifted TX reference (to get quadrature channel out of TX reference) and can do the digital I/Q-Lockin amplifier demodulation finally.
x(t) -> H(x(t))
sin(t) -> -cos(t)
cos(t) -> sin(t)

That's fine for lock-in amplifier demodulation.

But there is more than one way to skin a cat.

Cheers,
Aziz

[Sergey_P]

(stereo line)

How to perform the Hilbert transform, using the oscillator?

[Aziz]

Actually you could run it as a real 4 quadrant Rx and greatly expand your choice of coils. Differential coils should give superior ground and noise cancellation.

4 quadrant Rx? I don't know, what you mean by that.

I do have only two ADC channels (stereo line), which one is reserved for TX reference and the other one for RX signal channel.
Aziz