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

[Aziz]

Hi all,

I have made a good progress to achieve the "World's Quietest Metal Detector Front-End Amplifier"(c)(r)(tm).
Currently at approx. 0.5 nV/sqrt(Hz) @ gain=206.
I will squeeze out more noise.

Cheers,
42^12

[sled]

I'm currentyl experimenting with an USB based metal detector too, not for a final in-the-field metal detector though - more as a development platform for DSP algorithms although one could use it in the field with a raspberry pi or any other credit card sized computer with an usb 2.0 interface.

My idea is to use a stereo codec for data acquisition/transmission. The idea is not really new since the Minelab Terra series also uses a stereo codec (the Wolfson WM8731 audio codec to be exact, datasheet will pop up on google).

First I tried to experiment with exactly this chip and an I2S data bus to an STM32F4 (ARM Cortex-M4) microcontroller. The good thing is that it comes with a whopping 24Bit, 96kHz for both ADC and DAC so you can produce a TX signal, RX offset nulling with the two DACs and you get the RX signal and TX feedback signal with the two ADCs. The biggest problem was the coding, it's really a PITA to experiment with algorithms on an embedded system....

So my new solution is to use a cheap USB sound card with a good chipset. My choice was a Delock 61961 usb sound adapter for roughly 15 USD, but there are many other manufacturers that use this chip.

Why this one? Because it uses the VIA VT1620A chipset:

• 2-channel ADC with 90dB SNR, 16/24-bit supporting 44.1/48/96/192kHz sample rates
• 2-channel DAC with 95dB SNR, 16/24-bit supporting 44.1/48/96/192kHz sample rates

and 92dB dynamic range, I attached the data sheet.

What am I going to do with it? Well, first I'll build a small adapter board that has two audio jacks and plugs into the usb sound adapter, on the adapter board I'll put an ADC buffer opamp and a low power audio amplifier like the MC34119.

Then I will need to build a coil adapter, I have a Garrett ACE250 and Tesoro Cibola coil - does anybody know where I can buy a socket for those cable plugs? My current solution is to rip apart a PC power supply cable with a MOLEX connector, since the pins of the MOLEX connector fit most of the coils.

On the software side I'll try to keep it cross-plattform by using C++11, JACK Audio Server and Qt4 as GUI. The idea is to write abstract interfaces for the hardware driver (like generic signal sources, with the specific device logic encapsulated). Then I'll build a signal processing pipeline where different filters/processors can be plugged in. A control and plotting GUI is also planned but it will be completely independent of the DSP part.

It's a lot of work, and trying to build a perfect metal detector is like a never-ending project, therefore I really want to go in small steps and continously publishing some results with the source code of course. My first "feature" will be a simple USB based LC meter to measure out the coil and finding the resonance frequency so that one can null out RX and TX coils. I estimate about 1-2 Weeks to do that since I still have to work and study beside my hobby projects Also the adapter board with the ADC buffer op-amp and audio amplifier has to be designed.

Any ideas, tips are appreciated, especially about designing the ADC buffer op-amp since I need to find a suitable gain to not overdrive the ADC input and still having good sensitivity for small signals...

I also have some other fancy ideas, like putting a Accelerometer/Gyro in the coil to measure the sweep speed and adjusting parameters on-the-fly for ground balance, sample size - but it's still far away.


Simon

[sled]

It seems like GNU Radio could be a candidate as DSP library instead of reinventing the wheel.

[Aziz]

It seems like GNU Radio could be a candidate as DSP library instead of reinventing the wheel.

Hi Simon,

all sounding well & interesting and I'm looking forward to support you. I'll publish a new and improved detector controller schematics for USB sound card application soon (in April or so). A good starting point for the DSP is a FFT algorithm (the core DSP algorithm for us). You can find it on the internet almost everywhere. All open-source.

But if you are going to publish your source codes, oh boy!, it would be the total end of Greedlab! *LOL*
But you may warn everybody with the "Total Thermo Nuclear Melt-down"(c)(r)(tm)-threat. Even ML for unfair & greedy behaviour.

I have to finish the ultra-low-noise amplifier stuff until I can continue the new detector controller project (Class-E TEM & TEM V2.0 transmitter with KISS & cheap design).
So stay tuned..

Cheers,
Aziz

[sled]

Hi Simon,

all sounding well & interesting and I'm looking forward to support you. I'll publish a new and improved detector controller schematics for USB sound card application soon (in April or so). A good starting point for the DSP is a FFT algorithm (the core DSP algorithm for us). You can find it on the internet almost everywhere. All open-source.

cool thanks I followed your thread since a year or so, I like how you dive into details My goal is to build a development platform and move the result on to an embedded device later on. For the algorithm part, I've already played around in MATLAB with some fake data. The FFT is certainly a good starting point for a multi frequency device, at the moment I'm focusing on a simple single frequency device that uses I/Q demodulation and I'm experimenting with the hilbert transform to achieve a 90deg phase shift for the synchronous detection. All is still experimental, and the first things I develop will be some calibration tools for the soundcard (discrepancy between ADC/DAC sampling) and then some calibration tools for the coil (finding the resonance frequency, imbalance etc.).

But if you are going to publish your source codes, oh boy!, it would be the total end of Greedlab! *LOL*
But you may warn everybody with the "Total Thermo Nuclear Melt-down"(c)(r)(tm)-threat. Even ML for unfair & greedy behaviour.

No worries, I'm not going to spam your thread I'm just not ready to start a new thread because too little material to show yet, just ideas

I have to finish the ultra-low-noise amplifier stuff until I can continue the new detector controller project (Class-E TEM & TEM V2.0 transmitter with KISS & cheap design).
So stay tuned..

Yep, looking forward to that - I'm primarily a programmer and still somewhat new to electronics, especially analog circuitry. A low-gain, low-noise amplifier is needed in the front of the ADC. Didn't you build some sample and hold ADC somewhile back using an array of capacitors? what happened to this one?

[Aziz]

...
Yep, looking forward to that - I'm primarily a programmer and still somewhat new to electronics, especially analog circuitry. A low-gain, low-noise amplifier is needed in the front of the ADC. Didn't you build some sample and hold ADC somewhile back using an array of capacitors? what happened to this one?

Yep, I done that capacitor arrays. But you don't need it.

Single Frequency:
You can use the oscillator detector controllers I have made here recently (last year I think). The reference signal from the transmitter (TX) coil can be used to I/Q demodulate the receive (RX) signal (lock-in amplifier).
I = TX reference (right input channel)
Q = 90 degree phase shifted TX reference (Hilbert transform or if you like even FFT/invers FFT)
RX signal (left input channel)

Multi-Frequency:
A complete schematics of the detector controller with an ultra-low noise amplifier will be published soon. I'm just testing some BJT transistors and the amplifier. Be patient please or use the older projects here.

Cheers,
Aziz,
Software Engineer too

[Aziz]

Ok guys,

it seems, that I can have a 0.4 nV/sqrt(Hz) to 0.5 nV/sqrt(Hz) voltage noise density amplifier front-end.
(The 2N 3904 is a good candidate so far.)

Amplifier input protection will be done with two diodes only. There will not be a series input resistor. Only the resistor of the RX coil & cable (I'll try to keep it below 5 Ohms).

There will be one BJT for the task.
There will be a spare for the first BJT.
There will be another spare for the spare BJT.
There will be another spare for the last spare BJT.
If one breaks, there are enough spare BJT's.

*LOL*
Cheers,
/\sif

[Aziz]

I've been told by one of my schizophrenic ID's, that the ultimate hyper end of Greedlab is coming... *LOL*

Oh boys!, the "Ultimate Thermo Nuclear Melt-down"(c)(r)(tm) is coming... "rette wer sich kann!" *LOL*

Aziz,
^sif
/\sif
4212

666

BTW, who is 666?

[WM6]

BTW, who is 666?

Probably 3x BD600 in paralel (applied in low noise ampl)?

Aziz, be careful, don't destroy Rosenheim.

[cnckecgo]

QED dedector pcb schema

[PATCHES JUNIOR]

Aziz, why not let everyone here know your secret idea. Was this not you posting this on another high tech forum? If not I am sorry, sort of.

my ultra ******* hyper ultimate #@&*$#@ ultra low noise AC amplifier does outperform Candy detectors with ease. It is destined for a wideband CW detector with the "Top-Hat"(c)(r)(tm) IB AI coil. And I'm going to use the novel "Class-E-TEM/TEM V2.0"(c)(r)(tm) transmitter. And yes, the WBGB(c)(r)(tm) will be incorporated as well.
But hey, it's all beyond the scope of ML engineers.. /\sif

I took out a couple cuss words that are very popularly used by you.

[Aziz]

Aziz, why not let everyone here know your secret idea. Was this not you posting this on another high tech forum? If not I am sorry, sort of.

I took out a couple cuss words that are very popularly used by you.

Hey PJ,

I think it was one of my schizophrenic ID's. Maybe ^sif or /\sif. Maybe 4212. Don't know who.
But hell, the ID "666" is talking to me: "Press the red button Aziz! Comeon boy!, press it. Just do it! Let's make the thermo nuclear melt-down! Vaporize everything. Press the f***** red button!" *LOL*

Nope!, I must preserve Greedlab!
^sif

[PATCHES JUNIOR]

Hey PJ,

I think it was one of my schizophrenic ID's. Maybe ^sif or /\sif. Maybe 4212. Don't know who.
But hell, the ID "666" is talking to me: "Press the red button Aziz! Comeon boy!, press it. Just do it! Let's make the thermo nuclear melt-down! Vaporize everything. Press the f***** red button!" *LOL*

Nope!, I must preserve Greedlab!
^sif

That is the responce I would expect from someone who has no idea on how to build a great detector. Here we all thought you might be fairly smart. But I guess you live off of sims and have not actually build a detector that is any good. So sad.

Best thing for you is to get away from your supreme super hero and get out on your own. The longer you worship him, the more your work will be discredited. I hope you see that he is bringing you down. You seemed like a decent bloke at first.

[Davor]

This forum is awesome. But! This forum would be the best place under the Sun if only there were no substance-free trolls with agenda like PJ to spoil it now and then.

[PATCHES JUNIOR]

This forum is awesome. But! This forum would be the best place under the Sun if only there were no substance-free trolls with agenda like PJ to spoil it now and then.

Davor, so you are saying that Aziz with all his blue pills, pushing the button, nuclear meltdown, saying he built this detector when he only did some sim, cussing on forums with bad language, is OK with you? I bet you are a faithful follower of Marshal also.