"Beyond" TGSL ....

[simonbaker]

The reason? purely out of curiosity and just for fun.

Now the real reason:
1) Getting rid of the frequency shift/phase shift/amplitude change of the driven TX coil voltage caused by ground and targets. As the reference in-phase and quadrature-phase signals were generated internally (digital), the RX coil is sensing the TX coil changes due to resonant tank changes. The mismatch between real-life signal data and digital part will be minimized. This should make the detector more stable. I should look for a regulated LC oscillator to keep the TX coil voltage amplitude constant. This should make the detector even more stable. Frequency/phase shifts doesn't matter as the TX reference will be used. So the lock-in amplifier locks onto the TX oscillator instead to the internally generated reference frequency.

I do not really need an external hardware even. It can also be externally driven by the sound card. The RX signal and TX coil voltage needs to be fed into the sound card. The 24-bit sound-card has enough dynamic range to allow the signals unamplified or active buffered.

The active circuit just increases the TX output power and amplifies the signals.

2) didn't do that before (new)

3) can I top the 1 EUR coin detection > 50 cm?

Really funny thing.
Aziz

It seems like for (1) the best thing would be to both drive and sample the TX signal and you could really see the effect of the ground and maybe take it out of the RX signal or whatever. I think that's what you were doing originally???

Anyway, whatever you cook up will be interesting fare for consumption....

-SB

[Aziz]

It seems like for (1) the best thing would be to both drive and sample the TX signal and you could really see the effect of the ground and maybe take it out of the RX signal or whatever. I think that's what you were doing originally???

Anyway, whatever you cook up will be interesting fare for consumption....

-SB

No, I did not do all variations.

But you rised some good ideas! Yes indeed, it could be very useful to cancel the ground effects totaly.

I have some food for the coming days.. thanks.

Aziz

[Aziz]

Wiring Finished

Hi all,

the test board is almost finished. The LC oscillator is running and observed at oscilloscope: perfect (low distortion) sine waves at 28 Vpp between coil ends. I did not have all the parts so I have taken similar parts.
Total power consumption: max. 12 mA (at 13.8 V input)
TX reference output: 1 Vp (2 Vpp)
The rest will be tested soon.

Aziz

[Nicko]

...However I must reinvent the wheel. A personality flaw, too late to change.

Anyone need some wheels?

What colour? (re:HHGTTG)

Cheers

[Aziz]

Schematics Changed Slightly

Hi all,

now the new board is more tested. But I am not happy with the hardware phase shifter at all. The phase shifting pot (GB pot) is changing the input impedance slightly and this changes the reference TX signal amplitude at the output slightly. Something I wouldn't want to have it. Otherwise, I have to reduce the TX oscillator signal output impedance, which is not good to increase the load of the LC oscillator for stability issue.

So let's get rid of the hardware phase shifter totally. Let's mod it.

I will replace it with a simple inverting amplifier, which also reduces the high input voltage of the TX coil (capacitive voltage divider gone, input impedance increased, gain < 1). The phase shifting (GB setting) will be done purely digital in the software.

Aziz

[simonbaker]

What colour? (re:HHGTTG)

Cheers

I don't know that reference, but I take it I'm not alone...

SB

[Qiaozhi]

I don't know that reference, but I take it I'm not alone...

SB

Hitchhiker's Guide to the Galaxy?

[Nicko]

I don't know that reference, but I take it I'm not alone...

http://en.wikipedia.org/wiki/List_of...Marketing_girl

[simonbaker]

http://en.wikipedia.org/wiki/List_of...Marketing_girl

I stand enlightened and entertained. The resemblance is uncanny...

-SB

[Aziz]

Promissed Spice Files and Schematics

Hi all,

this is the current version of the purely digital version of a VLF hardware implementation using a sound card and laptop (netbook). The software is not included (hell, it's still an experimental code and even not finished yet).

Small changes in the RX amplifier gain and reference gain should be done to meet the sound card input specification.

With the new Netbook computers and a 9V battery block for the circuit (drains < 10 mA), an one-day-detecting should be easily possible (using the internal sound card of the computer only).

The software should implement a complex digital lock-in amplifier, which extracts from the input signals phase and amplitude relationship of TX/RX coil configuration. Furthermore ground effects, which also affect the TX voltage, should be taken into account (information is there and is detectable).

I will work on the software as time allows.
Cheers,

Aziz

[Nicko]

Hi all,

this is the current version of the purely digital version of a VLF hardware implementation using a sound card and laptop (netbook). The software is not included (hell, it's still an experimental code and even not finished yet).
...

Can I make noises here about the possibility of using an ATmega/WinAVR/GCC/AVRstudio4 combo (all free) instead of a laptop with VisualC++ or whatever? Or are you running a Linux distro with GCC?

It would be easier to lug around than a laptop and the newer ATmega32s are very powerful. Further, they support pretty much all the features of GCC - its a cool toolchain. AVRs are, IMHO, much better at supporting C & C++ than PICs - I used to be a PIC aficionado, but switched sides about 3 years ago when WinAVR got so much better.

Cheers

[Aziz]

Hi Nicko,

Can I make noises here about the possibility of using an ATmega/WinAVR/GCC/AVRstudio4 combo (all free) instead of a laptop with VisualC++ or whatever? Or are you running a Linux distro with GCC?

It would be easier to lug around than a laptop and the newer ATmega32s are very powerful. Further, they support pretty much all the features of GCC - its a cool toolchain. AVRs are, IMHO, much better at supporting C & C++ than PICs - I used to be a PIC aficionado, but switched sides about 3 years ago when WinAVR got so much better.

Cheers

you can forget any micro controllers or even DSP's! They aren't powerfull enough. The Atom processor in the Netbook for instance delivers enough number crunshing support to make a very easy but powerfull digital VLF detector. The presented solution is the best and easiest way of trying a digital VLF detector.

I am using just the Microsoft Visual C/C++ for the implementation. No Linux, no µC board, no problems at all. And it uses the most used platforms. There are some open source projects in the internet regarding implementing the sound card. They can be used to start with (Linux or Windows, what you prefer).

I promise you all, the final implementation will beat any VLF detector in the world. You all haven't seen the power of digital lock-in amplifier and it's noise immunity at all.

The latest ideas will solve the arised problems of my previous experiments.
Aziz

[Aziz]

Just some comments:

If the sound card is not capable of sampling at 96 kHz, the oscillator frequency should be lowered increasing either the coil inductivity or the resonant tank capacitors (CRX/CTX). If possible, a 24 bit sound card should be used. But it should run with 48 kHz/16 bit sound cards as well (reduced performance).

Now all what you need is the software and the coil + active circuit board shown above.

Aziz

[WM6]

Just some comments:

If the sound card is not capable of sampling at 96 kHz, the oscillator frequency should be lowered increasing either the coil inductivity or the resonant tank capacitors (CRX/CTX). If possible, a 24 bit sound card should be used. But it should run with 48 kHz/16 bit sound cards as well (reduced performance).

Now all what you need is the software and the coil + active circuit board shown above.

Aziz

Hi Aziz,

sorry if I miss something.

Soundcard usually have one signal input (Line In). You lead two signalls ("Lout" Signal S and "Rout" Reference R) to soundcard Line Input.

Presume that those Line In have to be stereo?

So "Lout" and "Rout" mean left and right channel of Line In?

How important is corect connection in other words: can both channell be exchanged by each other or not?

[Aziz]

Hints For A Digital VLF Detector Software

Some requirements and comments on the digital laptop VLF detector software:

- Make it frame based: a bunch of samples will be acquired and then processed. It is convenient to use power of two length frames (for FFT transforms). 512 or 1024 is a good starting value.

- Store the frame buffers in a larger buffer for a larger time constant algorithms (filter, lock-in amplifier)

- Use a Fast Fourier Transform (FFT) algorithm. You will need it as a phase shifter, filter, etc. It's not efficient but quite simple to do it this way.

- Make adjustable time constant parameter for the digital lock-in amplifier. Increase in time constant increases the signal-to-noise ratio and hence the sensitivity. A high time constant parameter lowers the detector response (pin-pointing with high sensitivity).

- Make an auto-ground-balance (tracking) algorithm.

- Make a motion/non-motion detection processing

How to phase shift a time-domain signal? (particularly for the quadrature-phase TX reference)

R(I) (reference in-phase samples vector)
- R(I) -> Complex FFT -> Power spectrum (complex I/Q output)
- Power spectrum -> Amplitude and phase spectrum
- Modifiy amplitude spectrum for filtering
- Add +90 degree or PI/4 (grad/rad) in the Phase vector spectrum to all components except the DC component (index=0)
- Amplitude and phase spectrum -> Power spectrum (back conversion)
- Power spectrum -> complex invers FFT (back transformation)-> R(Q) (90 degree phase shifted and filtered)

All what we did is a pure re-sampling of the time-domain signal but it's phase shifted and optionally filtered.

For the digital lock-in amplifier, you have then:
Signal S (RX), coming from sound card, optionally filtered
Reference R(I) (TX), coming from sound card, optionally filtered
Reference R(Q) (90° phase shifted R(I), computed)

Lock-in:
S*R(I) (mixer) -> low-pass filter -> I (channel X)
S*R(Q) (mixer) -> low-pass filter -> Q (channel Y)
S, R() is a vector of time constant length

From the complex I/Q channel, the amplitude M (rms value) and phase P can be calculated easily:

M = sqrt(Q*Q + I*I) (rms magnitude)
P = atan(Q/I) (Phase in radiants -> make it degree if you want)
P = atan2(Q, I) (better c math function for corrent radiant treatment)

Ground balance, coil, circuit dependent phase shift corrections? Well, correct phase P before processing it.

After processing M and P, you have lot's of informations for a digital sound generator. Put it to the sound card output to hear your detector screaming.

That's it. This is just a simple way to start with. Later, advanced techniques can be used to process also the TX response for a better ground balance.

Cheers,

Aziz