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

[Aziz]

Hi again,

now the spice circuit simulation files of the last PC sound-card interface modules will be available. I have lowered the DC/DC converter clock frequency to make the simulation faster. The positive supply voltage will therefore be fed from a fixed 10V supply. The simulation takes quite long time.

To make a full spice simulation of all parts, change the DC/DC converter clock frequency to the desired operating frequency and feed the UC+ into the +5V regulator.

Set LTSpice into the "Alternate" engine. Whole parts of the presented modules are now the PCPI sound card interface module.

Aziz

[Aziz]

Complete Schematics of the last module!

Hi friends,

Although not all tests are finished yet, I will publish the schematics of the new board now. It consists of the following parts:
- Power supply, Reverse Battery Polarity Protection
- DC/DC Converter (synchronizeable)
- Clock Generation Modules
- Modulator Clock Divider (synchronizeable)
- Stereo Channel Modulator

The clock frequency of the DC/DC converter may differ to real circuit. So adjust the oscillator resistor and/or the capacitor for your requirements. The capacitor should be FKP2 or polystyrene one. A clock frequency between 30kHz to 100kHz should be ok. Resistors should be all metal film (1%).

The stereo channel modulator can be extended to modulate two channels more. In this case, the bipolar rectangle wave form modulation can not be used. But there is no need for this at the moment as I only want to use two independent channel processing. You can use NE5532/TL072 or compareable dual op-amps. There is also no need for JFET-input op-amp.

If you don't use the synchronisation of the modulator, then the polarity of the encoded channel signal cannot be determined except you feed a defined voltage to get the signal polarity.

Left channel of the sound-card output is reserved for cycle triggering and synchronisation (DC/DC converter and modulator). Right channel is for the modulator clock. The modulator clock will be either half or the fourth of the feeded clock frequency (jumper selection).

The new board makes the use of sound-card signal processing possible now. You only need to feed two DC signals into the modulator. The generated internal noise is acceptable and gives satisfactory signal decoding.

More to come..

Aziz

[Aziz]

Hi all,

the modulator is working really fine (like frozen scope output, very quiet). I will optimize the circuit for 6 kHz clock (modulator at 3 kHz) and 1 V rms to the sound card input. The triangular modulation response is frequency dependent. The higher the frequency, the lower the output. This is typical for integrators (the last stage is an integrator).

We can use the dual op-amps NE5532. The OP27 isn't worth for the marginal improvement.

If I have finished the optimization, the complete schematics with the spice simulation files will be available for educational purposes.

Aziz

[Aziz]

Here is the new board.. much quiter than the previous version and
allows µV detection.

[Aziz]

Hi all,

the new board with power supply, DC/DC generator, clock generator and stereo modulator is finished now.
It seems, that I have reduced further noise and power consumption. The whole tests are not finished yet and I would like to see the signals with the scope soon.

I took a voltage divider on the analog power supply lines with a potentiometer for the input voltage. It is interesting, that normal potentiometers produce greatly much noise than a 10-turn wire potentiometer. To test various single and dual op-amps, the board has three IC sockets per stage.
The OP27 op-amps show the improment but have more current draw. The NE5532 works also fine with less current draw. The clock signals seems to be much more stable compared to LM311.

To be continued...
Aziz

[hobbes_lives]

I just bought an 50 MHz oscilloscope on ebay. I hope, that is ok and no crap and will be delivered soon. This will now help to develop this project much faster.

Aziz

Good luck. Buying scopes on eBay is a roll of the dice. I have two scopes that I bought recently, one was supposed to be fully functional and arrived broken, the other I knew had issues but bought it to repair. Most scopes on eBay have issues, or end up getting damaged in shipment. I hope yours arrives in good shape and fully functional.

[Aziz]

I just bought an 50 MHz oscilloscope on ebay. I hope, that is ok and no crap and will be delivered soon. This will now help to develop this project much faster.

Aziz

[Aziz]

Hi Alexis,

thanks for your hint. Indeed, this little circuit works perfect. I had already similar solution with additional mosfet gate protection using a zener diode.


Project update:
The new DC/DC converter and clock generation module already finished and is working. I am focusing to the new stereo modulator and looking for an oscilloscope now.

Aziz

[Alexismex]

Hello aziz, good work ,
I show you my protect polarity circuit i use in my detector
work very well without loss
Have nice work
alexis

[Aziz]

Hi all,

the circuit design gets almost finished. It takes too much time to finish a complete simulation run.

I have not decided yet, which op-amp I will use. I also would like to test some different op-amps in a real circuit. So I might provide both single and dual op-amps. The OP27 should give very good performance. It also has more rail-to-rail level and is quiter (less noise) than the NE5532/NE5534.

I will support now only two sample channels (four former). With the two independent differential integrators, only two channels should be enough to process the signals. I will try to omit the S&H stages whenever possible. The new design will be much lighter and hopefully better.

I will start now with the new power supply, DC/DC converter, reference clock generator and the channel modulator. I also will provide a battery reverse polarity protection circuit.

Now it is time to fire up the soldering iron again. The new simulation models and schematics will be available, if I have tested the modules.

Aziz

[Aziz]

Hi friends,

it seems, that I have a much better solution for the LM311 (clock generation). This fu.ck.ing chip has no noise specs. Therefore it won't be used in my design anymore.

Biasing with the half digital power supply for the clock generation wasn't a good idea former. Any digital power supply loads had an effect to the timings and caused timing jitter (very very critical). I could oberserve this problem during independent 1 Hz LED blinking of the microcontroller. I have changed this into a conventional inverting comparator with an usual and specified op-amp (probably dual op-amp NE5532 or two single op-amp OP27) with the common mode biasing. It needs bipolar power supply and therefore the DC/DC converter must have it's own clock. The synchronization of the DC/DC converter will be done like the previous solution already posted earlier. The synchronisation will be done by the cycle start trigger pulse (left channel of the sound card output).

The right output channel of the sound card will only be used for the modulator. The digital clock signal have to be divided by two to make a sure 50% duty-cycle clock. So the modulation will be performed on the half clock frequency.

The clock generation module will use both analog power supply (+5/-5V) and digital power supply (-5V). The timing signals will be much more precise now.

The modulator uses only one JFET per channel now. It also supports bipolar rectangle wave output to reduce one op-amp stage (needs more harmonics to decode the signal). It also uses much less op-amps to reduce more noise.

I need to built the clock generation, DC/DC converter, power supply and the modulator once more. In this case, all the functional parts will be placed on a single board.

I did not continued with the simplified version. I think, the microcontrolled one will be much better. So I have to continue with both options.


Aziz

[Aziz]

Hi all,

I am currently working on the next revision of the microcontrolled PI board. Found some critical problems, which needs redesign.

I will reduce the complexity of the circuit and the number of required stages and op-amps further.

Following sections will be revised:
- Power supply, DC/DC converter (with own clock oscillator, synchronizeable)
- clock generation modules (more precise and stable timings, less timing jitter)
- channel modulator (only one sample per channel coding, less complexity but lower noise)
- will implement two independend channel differential integrators

Some single-supply circuits (like clock generation) will be changed into bipolar-supply mode. The LM311 comparator will be thrown out of the window and the part of the circuit will be implemented with more precise op-amps to reduce clock timing jitter.

So there is much potential to reduce more noise.
Aziz

[Aziz]

Hi simonbaker,

How difficult is it to make USB interface?

Using sound card is very nifty but now it seems so much trouble to modulate, time stretch, etc. just because of limitations of sound card.

It seems in the end the effort would be better spent working with on-board ADC and circuitry to read the S&H samples.

But before give up on sound card -- need some real data. Can you compute the spectrum of the RX pulse so we understand more about it before designing circuits?

Regards,

-SB

I suppose, you would need the full USB bandwidth to pass the huge ADC data (real-time sampling). If you decide to pass fewer samples, then the 1.5 Mbit would be by far enough. The latter one can be made with ATtiny45, ATtiny2313, ATmega with few parts only (already posted some links on this). Or you can use the FTD chips for this purpose. The USB isn't a big issue and quite trivial.

But it is very difficult to achieve the real 24-bit resolution by the on-board ADC. Why doing this by yourself? Let the fu.ck.ing industry make this for you. At a much cheaper price and without any risc. That' the reason for, why I insist on the sound card solution.

The 24-bit or more resolution would be just waste of power, if the signal data is bad or noisy. We should focus on this item too.

It seems, we can even work with only two sample channels. This would reduce the circuit complexity much. Also the modulator would become much lighter and getting more quiet (less noise). Each channel would be passed on the stereo line (left and right). The S&H stages would become obsolete, if we use a modified two channel differential integrator with hold feature. It will anyway hold the signal for long time and will be sure detected with the slow ADC.

But I will try first to make the direct and continious sampling possible (without the modulator and the complex circuit). This possibility gives the huge potential for masking out undesired noise sources (no missing time domain and the noise will be present at any time in the data). And I like smart and light solutions.

Today, I was able to detect the unique frequency spectrum fingerprint of the target response . Which represents the different time constant of the target . This makes save discrimination of the PI technique possible. Quite surprizing result on the slow decaying coil with less target eddy current stimulation. I will build a two channel pre-amplifier to make more experiments on this. And it is quite non-critical with DD coils now. I very likely won't have the much ground effects like in the VLF's. I just need to amplifiy the low receive signal on the DD receive coil to see more signal response.

It is a pitty, that I have very limitted resources (No, I still won't work for the german government anymore. Never ever! The damn vultures are cruising over me. ) . I even can't see yet, how the damping process is going on. The sound card is not a good scope. I should try to get an oscilloscope to make faster progress. And I probably should rob a german bank to afford expensive parts. It is feared that, that they themselves are bankrupt too.

Therefore, I should search for much effective, easy and cheap solutions.
Sound card and laptop solution is still top priority to me. I need the huge number crunching possibility.

More results on sampling data representations and findings, as soon as I have more results on this.

Aziz

[simonbaker]

How difficult is it to make USB interface?

Using sound card is very nifty but now it seems so much trouble to modulate, time stretch, etc. just because of limitations of sound card.

It seems in the end the effort would be better spent working with on-board ADC and circuitry to read the S&H samples.

But before give up on sound card -- need some real data. Can you compute the spectrum of the RX pulse so we understand more about it before designing circuits?

Regards,

-SB

[Aziz]

Hi all,

I had a look into some analog multiplier IC's. Tooooo much expensive $$$$!
It is due to super matched transistor pairs in the Gilbert's cell core. A discrete implementation would even cost much more.

The low cost AD633 needs more supply voltage (+-8V). The analog multipliers generally need more operation current (the current is the result of the math). That's quite critical. They also have generally bad noise characteristics.

The one, I would like to try it out was the AD834. But it is difficult to get this chip. It is also quite expensive.

---

I will try the DD coil configuration next time with direct sampling on a slow decaying coil voltage.

Aziz