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

[Aziz]

Hi barry,

Aziz. JTAG Ice atmel programmers are cheap on eBay. Well worth the investment.

Thanks! Indeed, they are quite cheap. But they do not support the ATtiny2313.
No, I should buy 15-20 samples of ATtiny2313 instead of a new programmer.

Anyway, I would not make the same mistake again. I have found another sample at home. So 4 samples should be enough at the moment.

Aziz

[Götz von Berlichingen]

I dont use AVRs (yet), so dont kill me if my Information is wrong.
A simple parallel AVR Programmer (and more) can be found at :
http://elm-chan.org/works/avrx/report_e.html

[Aziz]

Hi all,

I have to revise the channel modulator. I cannot achieve the desired ADC resolution in the demodulator otherwise. Anyway, this will be done later when there is more time. The presented channel modulator will not work optimal and has reduced resolution. I will very likely change the modulation type. Back to the black board.
Aziz

[Aziz]

Again channel modulator resolution:

I will lose only two bits of max. theoretical resolution (24 bits). So this would end to an effectively 22-bit ADC conversion. Of course, there is no free lunch. The 4 channels will reduce the total bandwith of the signal line by log2(4) = 2 bits.
In the demodulator, I will loose 1-2 bits due to the inefficient modulation type (cost of easiness of the hardware).

Performing a boxcar integrator will compensate the losses of the ADC resolution. So there are still more bits for detection left.

I will see, whether the presented modulator will show the expected results practically. If not, an another one will be developped.

Now the projects still goes forward.
Next task: Micro controller I/O line mapping and checking for possible problems. If there aren't, I will continue with the PGA, gated integrator and S&H stages.

Aziz

[Aziz]

Some more facts about the channel modulator:

I was interested in the capabilities of the channel modulator. I have increased the modulation frequency up to 24 kHz. No problem at all. The precision synch generation works perfect.
Due to the synchronisation of the modulator with the cycle period (reset logic of the binary counter), some restrictions on the pulses per second (PPS) arises:

fclk_synch = 24 kHz, the DC/DC converter runs with 24 kHz (no problem)

Max. PPS, max. Channels
====================
24kHz/16 = 1500, full 4 channels
24kHz/8 = 3000, 3 channels, channel 1 will not work (reset occurs)
24kHz/4 = 6000, 2 channels, channel 1+2 will not work (reset occurs)

For a sufficient target discrimination, 2 or 3 channels would be enough allowing 3000-6000 PPS.
SNR increase on 3000 PPS would be (compared to 100 PPS response time): sqrt(3000/100) = sqrt(30) = factor 5.5
SNR increase on 6000 PPS: sqrt(6000/100) = sqrt(60) = factor 7.7

SNR increase on 1500 PPS: sqrt(1500/100) = sqrt(15) = factor 3.9

So the SNR increase will increase the ADC's resolution and compensate the lost resolution bits. The higher modulation frequency is more quiet for signal acquisition (lower noises).
But I tend to have all 4 channels for investigations. So max. 1500 PPS is just enough for me.


Aziz

[Aziz]

Hi all,

there is room for further improvements later by splitting the 4 channels into two 2-channel stereo channel modulator. I can then achieve 1 bit more resolution and double the PPS rate. In this case, I have to overlay the command signalling to the sample channel signalling. The laptop software needs to know GB push botton events and operation mode changes. This will disturb the operation a little bit only on occured events. But for later experiments and improvements, it can be realized with ease.

1 bit more resolution increases the SNR by a factor of sqrt(2^1) = 1.41.
Now 1500 PPS compared to 100 PPS would increase the resolution by:
log2(1500/100) = log2(15) = 3.9 bits.
for 3000 PPS: log2(30) = 4.9 bits.
for 6000 PPS: log2(60) = 5.9 bits.

We achieve and exceed the 24-bit resolution then. The 100 PPS response time is a comparison reference and means, we would have a compareable detector with 100 pulses per second and would have on a coil movement of 1 m/s a measurement on every cm distance. It it just a time response figure of the detector and it is a resonable reference value. There is no free lunch: if you want to have a fast response detector, the SNR will be limited. If you "oversample" by reducing the responce time (slower response), you can increase the SNR and thus the sensitivity.

The higher PPS rates will be used to increase the SNR by taking a reasonable response time for the detector. Best sensitivity can be achieved for pin-pointing: Lets take 10 PPS:
log2(1500/10) = log2(150) = 7.2 bits more. If we have a 20 bit ADC conversion (due to conversion losses), so we could effectively achieve a 27 bits ADC conversion.

Who can achieve such a sensitivity at the moment?


Aziz

[Aziz]

I think, an additional pin-pointing push button makes really sense. I will have one more I/O line for this and total 16 I/O lines. So I need an external 20 MHz crystal oscillator for the micro to meet the specification. The XTAL2 pin can be used as a general purpose I/O line then.

Aziz

[Aziz]

I got an idea for the pin-pointing button:

By pressing the GB button longer (2 sec), the pin-pointing could be activated. By pressing GB button again longer, the normal mode could be set. A distinct sound in the laptop signals the mode of operation changes. Shorter GB button press will do ground balance.

So, the reserve I/O line could be kept.

Aziz

[Aziz]

Hello friends,

I will revise my channel modulator with precision timings and higher resolution. It will be a full stereo channel modulator (two channels per line). The binary counter will be dropped and replaced by a JK flip-flop. So it delivers precise synchronous logic levels (both q, /q). The JK flip-flops will be switched to a two bit synchronous binary counter with reset capability (synchronisation).

This will give the ultimate high resolution data conversion possibility. A fallback into a pure two channel PI could be also realized. In this case, I will have the highest possible resolution (each channel has its own ADC).

Aziz

[Aziz]

Hello friends,

here is the improved stereo channel modulator. The modulator shown below is for the left channel. Same applies to the right channel with the corresponding signals s3 and s4.
The modulator logic is now improved and should give more precise timings. The modulator clock is buffered twice. The other one feeds the DC/DC converter. I will build this improved modulator next time.

Aziz

[Aziz]

Hi all,

all these efforts are for the interesting question:
"Can the stereo 24-bit codec chips (ADC's) be used in a PI?"

Today, there are superb 24-bit 192 kHz multi-channel codec chips available. They were produced in mass and are quite cheap (used in HD sound-cards). A super high number crunshing DSP (like the blackfin) could do then all the software parts in realtime, which does the laptop at the moment for the interesting question. In this case, the modulator would not be necessary and the DC signal can be fed directly to the ADC channels.

This is a pure long term sight. But it is going to be interesting.

Aziz

[Aziz]

Hi all,

I have replaced the Schmitt-trigger CD40106BP with MC14106BCP. So the 220 pF capacitor in the mono-flop (cycle trigger pulse) runs stable. If you want to use the CD40106, you have to increase the capacitor to 330 pF or 470 pF. The cycle trigger pulse with isn't important and can be increased slightly. The micro will be triggered on the raising edge (from -5V to 0V).

I have started with the improved stereo channel modulator and will be finished in few days. I had to built the clock generation part once more on a second board.

Aziz

[tatarlar]

tam şema

Arkadaşım bu şemanın son ve tam hali varmı

[Aziz]

Merhaba,

Arkadaşım bu şemanın son ve tam hali varmı

son ve tam hali yoktur. Bekleme bunu, bu proje uzun sürebilir. Sadece "experimental version". Semasi tam olsa bile, bu detektör degildir. "PC software'sis" bu zaten calisamaz. En büyük parcasi'da software oluyor. Bu proje ileri giderse (daha belli degildir), her detektörü yaninda utandiracaktir. Buna cok eminim.

Aziz

[Aziz]

Hi all,

the new stereo channel modulator for total four channel signal acquisition is working now.

I took lots of TL072 (dual op-amp) instead of the TL074 (quad op-amp) for the case, that I want to use a pure two channel PI. Then I can remove some parts from the board (two TL072, four BF245) and can use the full ADC resolution. The modulated stereo signal will be overlayed with the control signalling from the micro later (laptop have to know mode changes and push button events).

Bad news: My on-board internal HD sound-card isn't working on the right channel. I didn't use this till now. Must be a mechanical problem (too much plug-in/plug-out stress I guess). Fortunatelly, I have another two external USB HD sound-cards and the development will not be delayed. I also tested a very cheap sound-card (48 kHz/16-Bit) and is really not convenient to break the limits. But is in principle working with reduced resolution.

I will do some more tests on the new channel modulator.

Aziz