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

[Aziz]

Although I do think you'll be in for a nasty surprise when you take your super-duper detector outside, and find that it doesn't live up to expectation.

Anyway ... keep trying ... this is a lesson to us all.

Indeed, this is a good lesson for us all. Particularly for me.

Now, I bound the TX coil very tight and the result is quite stable. The drift of manually balancing the coil is stabilized within a minute. Also the temperature of the coil is getting normal within a minute (due to touching the coils with warm hands).

The sound card itself seems didn't produce any drifts. It is producing only some noise in his amplifier and digitizing stage. When the RX coil is connected to line-input, it is registering only 0.8 to 1 µV rms noise without a resonant capacitor. With the resonant capacitor, the noise is increased (up to 10 times). This is quite normal behaviour - it is amplifiying also noise on resonant frequency.

I will try to drive the TX coil with some CMOS inverters and getting the power H-Bridge coil drive unnecessary. It isn't necessary to drive the TX coil with excess of power. This produces more drifts and noise. Also, +5 V should be enough for this and taking power directly from USB port.

Guys,

you all will be very surprized, when I am going to show you, that the laptop detector will also work very well with only conntected coil without any electronics. If somebody has a laptop, you only need the search coil and the detector software to get a very impressive, powerfull and easy to use MD.
With a very simple electronics module, the performance is increased and will out-perform any similar detector.

I hope, TheWizard will go on with his development on a portable one. I will focus only with the laptop variant. On the coming weeks, I will start with the software coding to get this beast easy useable. I will redesign all the software code part and doing some very tricky techniques.

Take care and have a good day,
Aziz

[Qiaozhi]

Hello,

I have found some of the drifting problems:

Because I am using an experimental DD coil configuration (not fixed together), some strange effects (drift, unstability, etc.) arise. They are really very interesting to observe when measuring very small signal changes.

1. When the DD coil is adjusted to a minimum coupling, it takes some time to get mechanical stabilized.
a) The coils are bending due to gravity force (some µm?).
b) The coils are sliding a bit (gravity and unstable position)
The effects are very small, but have a big factor for very small signal changes.

2. When the TX coil is operated with some power, the generated heat has effects on the result.
a) TX Coil gets slightly warmer -> changing resistance.
b) Driving transistors gets warmer -> changing resistance, changing coil current
c) Other resistors on the electronics for the TX coil driver gets warmer -> ...
d) Coil temperature changes geometrics of the coil -> changing the balance

3. Power supply
When more power is consumpted from the batteries, the battery voltage decreases. This has effects on not stabilized power supplies.

4. Other
a) Seismic effects: Not touching the coil ground, going very slowly on the ground, not speaking (air pressure change, wind)
b) Changing air pressure or wind (closed windows)
c) Capacitive effects on unshielded coils.

Indeed, one can build a very sensitive seismic detector! So the coil must be built very very tight.

Now I will try some other experiments.
Aziz

So what does all of this tell you? ... that you should be measuring relative, not absolute changes in the signal, and tracking the signal drift over time.

Anyway, in the real world, these effects are to be expected, and will no doubt scupper any of your attempts to measure to a resolution of nanovolts. I would recommend that you do some field-based testing before proceeding with this project. Metal detecting in the comfortable environment of the lab is somewhat different to the world outside.

By the way, I'm not being negative towards your efforts. Although I do think you'll be in for a nasty surprise when you take your super-duper detector outside, and find that it doesn't live up to expectation.

Anyway ... keep trying ... this is a lesson to us all.

[Aziz]

Hello,

I have found some of the drifting problems:

Because I am using an experimental DD coil configuration (not fixed together), some strange effects (drift, unstability, etc.) arise. They are really very interesting to observe when measuring very small signal changes.

1. When the DD coil is adjusted to a minimum coupling, it takes some time to get mechanical stabilized.
a) The coils are bending due to gravity force (some µm?).
b) The coils are sliding a bit (gravity and unstable position)
The effects are very small, but have a big factor for very small signal changes.

2. When the TX coil is operated with some power, the generated heat has effects on the result.
a) TX Coil gets slightly warmer -> changing resistance.
b) Driving transistors gets warmer -> changing resistance, changing coil current
c) Other resistors on the electronics for the TX coil driver gets warmer -> ...
d) Coil temperature changes geometrics of the coil -> changing the balance

3. Power supply
When more power is consumpted from the batteries, the battery voltage decreases. This has effects on not stabilized power supplies.

4. Other
a) Seismic effects: Not touching the coil ground, going very slowly on the ground, not speaking (air pressure change, wind)
b) Changing air pressure or wind (closed windows)
c) Capacitive effects on unshielded coils.

Indeed, one can build a very sensitive seismic detector! So the coil must be built very very tight.

Now I will try some other experiments.
Aziz

[Aziz]

Hi to all,

I have tested my coil switching H-Bridge with clock synchronisation with the laptop.

New air-test detection record on laptop detector:
1 Euro coin at 40 cm!

The laptop detector has less demands on coils (they do not to be perfectly balanced at all).

TheWizard,

now the synchronisation is perfect. By increasing the coil current, the detection range can be extended. I have limited the coil current through a 68 Ohms series resistance. There is more potential available yet.
I can observe phase changes parallel to detection changes.

By the way, the receive signal is not amplified by an external amplifier. Only the line-input with protection diodes is connected to coil. Also the receive coil is not yet driven in resonant mode. So operating in resonant mode will increase the performance and stability (reducing noise).

Now, I have to solve some drifts (power supply not stabilized). Then, the detector will be quite superb.

Regards,
Aziz

[Aziz]

Hi TheWizard,

I am watching the new Intel Atom and Intel Centrino Atom architecture. It is getting very interesting due to power consumption. And the laptop industry has increased his efforts towards ultra mobility demands.
We can expect in few years more compact and light weight laptops, PDA's etc. We should be prepared for this: coding in C/C++ for a flexible architecture design.
I am not worry about the battery life at the moment. This will be increased in the future.

Aziz

[TheWizard]

yes - you can remove output filter from audio-card and get signal from PWM

I use small full-bridge driver(SMD, 1A) to protect MCU... I say before,
it is like "class D"

After tests with coil I can change full to half bridge...

"......So keeping the MD as simple as possible......."
PC is power machine, but is not convenient at field... but have many possibility
You can test with EEE laptop

[Aziz]

Hi TheWizard,

I got now reconstructed the digital reference clock from the AC line-output of the sound card. The coil can now be driven up to 24 kHz. I do not need power amplifier for the transmitting signal anymore. So the TX coil can be driven directly with the power supply via switching H-bridge (four transistors). The digital signal is reconstructed with an opamp Schmitt-Trigger circuit with proper level shifting to switch transistors or to drive logic circuit (CMOS level).

I want to share the stereo line-output of the sound card for reference clock generation for TX coil (left channel) and for audio signal output (right channel) and make do the metal detection with only one sound card. So keeping the MD as simple as possible.

Now, I am looking forward to test the lock-in amp with this mode.

Take care,
Aziz

[Lake]

Here is MicroChip application note about implementing Lock-in:

Implementing Digital Lock-In Amplifiers Using the dsPIC DSC

http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en532447

analog lock in amplifier

Hi All,

Take a look at the Analog Lock in Amplifier description.

Think you may have seen this before.

Look closely at the block diagram.

Has anyone seen anything that looks like this?

Amplifier, demodulators 0 and 90, low pass filter.

http://www.signalrecovery.com/_AppsNotes/TN1002.pdf

http://www.signalrecovery.com/AppsNotesDownload.htm

Anyway take a look around.

[Aziz]

Hi

It is good...
Now I waiting coil and then will test Transmiter x1
now is x4 and work good and with x8 - xN is PWM period for sine 12500

Then(with good coil) I will test and manual phase shifting and auto tune for Ground Balance...

We have to use also the first derivations of the lock-in amp outputs. This compensates drifts in the circuit and also change of ground characteristics. First derivation is very good for auto-tune!

Aziz

[TheWizard]

Hi

It is good...
Now I waiting coil and then will test Transmiter x1
now is x4 and work good and with x8 - xN is PWM period for sine 12500

Then(with good coil) I will test and manual phase shifting and auto tune for Ground Balance...

[Aziz]

Hi TheWizard,

I just tried an external crystal oscillator (11.9041 kHz) for driving the coil directly with series LC resonant tank. I got round about 170 Vpp on the coil (not correct due to 1 kHz cut-off frequency of the multimeter, so the voltage must be much higher). The switching transistors in the bridge got quite hot (2 x BD139/BD140). I was wondering about the transistors (max. Uce 100 V ). They are still alive.

I observed a constant drift of the lock-in amp phase output due to not accurately synchonized with the crystal osc. It is quite difficult to synchronize to an external oscillator. So it is better to use the reference clock directly from the µC/DSP/PC (lock-in amplifier) itself.

Next I will try to synchronize directly with the line-output (reference clock) to drive the coil.

Take care,
Aziz

[TheWizard]

I wish I would have enough time to "play" with many things and trust me when I say, lock in amplifiers are not in the first ten.
As I said before, I don't work with narrow band systems except for resistivity measurements and I have no need to implement lock in amplifiers for that.

Can I know who is "the first ten" ?

[simonbaker]

I just have to say this is excellent what you guys are accomplishing. I wish I had kept up with my signal processing theory so I could start doing some of these experiments and better follow your discussion.

What you are doing with laptops is the exactly right approach for developing metal detectors, whether you use all analog, DSP, or laptop computer. You are understanding the information which will guide all designs you may make.

When I was playing with metal detector circuits (very simple), I had thermal and 1/f noise problems in op amps, it was very low frequency just like target and ground signal when sweeping the coil. I had an idea to use a quad op amp and measure the thermal noise from one of the unused amps (grounded input) hoping that it would be correlated with the other op amps that were amplifying the signal. But I never had time to try it, just wondering if that might help.

I had a question about the high-Q TX coil design -- when you put it near the ground, wouldn't the ground make much lower Q? Maybe it doesn't matter.

I will try to understand this discussion, I think it is very important to any MD design. Thanks for sharing all the ideas and results!

-S

[Aziz]

Hi gwzd,

I made some numerical tests with the input data now. The noise is typically on the least 4-6 significant bits of the 24 bit input data. I tested also some software components. And they are working correct so far.
Using the noisy least significant data bits over some period time (time constant), the accuracy can be increased despite of the noise.
This is what I am using to detect very small signal changes.

Even, when the signal change is somehow correlated over time (due to drift), the software detector can handle this really very elegant.

I wish, your guys could see the real-time operation. It's difficult to demonstrate this without a camera.
Aziz