"Beyond" TGSL ....

[moodz]

and here is a more practical example of what I was talking about earlier.... The RX oscillator is the master oscillator ( not the TX ) .... The TX oscillator is kept FREQUENCY locked to the RX oscillator by the shift register ( rather a neat trick IMHO ) however the PHASE lock is done via the coupling between the induction balanced TX oscillator coil and synchronous oscillator RX coil.

So there are two paths for oscillator control .... an internal Frequency control loop and an external ( via coil / target coupling of phase ).

Thus very sensitive phase recovery of the RX signal is possible and the rest of the electronics pretty much works the same as other VLF type descriminators.

The purpose of this circuit is to greatly increase the sensivity and descrimination of weak noisy target phase info.

moodz

[Aziz]

Hi bmoore,

Aziz: your work is impressive and the progress on the project appeared great until you tested it in the real world. After all the fantastic progress you leave the impression of giving up too easy. Isn't there some way in software to sense or compensate for the ground issues. I'm new to this forum so forgive me if this comment is not presented correctly.

I haven't been able to make the other two experiment configurations due to lack of time and resource yet.

The main problem with single transmit frequency like in the VLF is the reduced ability to ground balance (it's not perfect). If you don't have a good ground balancing method, all the detector get's unuseable in the field. This applies particularly to the mono coil VLF experiment.

But the induction balanced coil configuration works in the following configuration:
Take the reference clock from the transmitting (TX) resonant LC tank for a synchronous demodulation of the receive (RX) signal. The high TX voltage must be reduced before it is fed into the sound card. The receive signal should be amplified if the signal amplitude is below the noise floor level. In a good 24-bit sound card, it's usually not necessary to amplify the RX signal (resonant LC tank has it's own gain).

The demodulation is a simple complex digital Lock-in amplifier. It delivers the phase shift between TX and RX coil voltage. If you normalize the TX voltage, then you get the root-mean-square (rms) value of the RX voltage.
The phase shift is a good discriminating feature, the rms RX voltage magnitude the signal strength. But ground also changes these values slightly.

The ground balance its not perfect as mentioned. There are more efficient ways to make a much better ground balancing. This is the reason, why I am not following the VLF technique anymore.

I am working on an efficient novel ground balancing method.

Aziz

[simonbaker]

Aziz: your work is impressive and the progress on the project appeared great until you tested it in the real world. After all the fantastic progress you leave the impression of giving up too easy. Isn't there some way in software to sense or compensate for the ground issues. I'm new to this forum so forgive me if this comment is not presented correctly.

I think there still may be life in a mono coil design if someone is willing to really beat to death the signal processing aspects. Or at least a "non-nulled" two coil design. It could be a nice long term project for a good software / estimation theorist type to try to outsmart the noise.

-SB

[bmoore]

Aziz: your work is impressive and the progress on the project appeared great until you tested it in the real world. After all the fantastic progress you leave the impression of giving up too easy. Isn't there some way in software to sense or compensate for the ground issues. I'm new to this forum so forgive me if this comment is not presented correctly.

[simonbaker]

Hi Simon,



yes, I have measured the phase shift between the internal generated reference clock (driving TX frequency) and the TX resonant tank.

Well, the simple IB configuration can deliver much more performance than 20 cm. I will try two different more configurations soon:
- standard IB configuration (coil coupling 0)
- shared coil coupling configuration (coil coupling > 0)

The latter one achieves a better coil stability as the RX coil can be smaller than the TX coil and can be mounted on the same center. The standard IB configuration is more critical to coil stability and is difficult to build.

This time a true synchronous demodulation will be performed. I.e., the TX tank will be used as a reference and will be measured against RX coil. This should deliver more reliable results in the field.

Aziz

Sounds interesting -- two coils, but no careful nulling, handle it in processing?

The potential problem I see with non-nulled coils is they may behave like a single coil, so you won't be able to detect a strong phase difference -- but I don't know, I'll look forward to your results.

-SB

[Aziz]

Hi Simon,

Well it was worth trying anyway. I guess not surprising the reaction of the coil is too complex to easily process.

Were you able to compute a phase shift in the coil by comparing the coil input signal to your driver signal?

I would think that a real target would produce a different phase shift than the ground.

Maybe if you only try to get 20 cm depth you can still distinguish targets from ground.

Regards,

-SB

yes, I have measured the phase shift between the internal generated reference clock (driving TX frequency) and the TX resonant tank.

Well, the simple IB configuration can deliver much more performance than 20 cm. I will try two different more configurations soon:
- standard IB configuration (coil coupling 0)
- shared coil coupling configuration (coil coupling > 0)

The latter one achieves a better coil stability as the RX coil can be smaller than the TX coil and can be mounted on the same center. The standard IB configuration is more critical to coil stability and is difficult to build.

This time a true synchronous demodulation will be performed. I.e., the TX tank will be used as a reference and will be measured against RX coil. This should deliver more reliable results in the field.

Aziz

[simonbaker]

Hi all,

I have been field testing with the latest design. As suspected, it didn't work well. It was total useless in the BFO mode (phase change detection).

The ground detunes the resonant tank heavily. Also the geo-magnetic field has an affect to the response signal during tilting the coil. Varying the coil height above the ground is also critical.

So this proposal is definitely not the way to go.

The best way to go is as mentioned:
Synchronously demodulating the RX signal, where the reference clock comes from the TX part.

I have to find an elegant way to make an easy & robust PI design for the Laptop version. I know, the other PI project I am still following (the PI controller) would give best results. But's its complexity is high. I have to find a very easy method.

Cheers,
Aziz

Well it was worth trying anyway. I guess not surprising the reaction of the coil is too complex to easily process.

Were you able to compute a phase shift in the coil by comparing the coil input signal to your driver signal?

I would think that a real target would produce a different phase shift than the ground.

Maybe if you only try to get 20 cm depth you can still distinguish targets from ground.

Regards,

-SB

[Aziz]

Field Testing of the MONO coil VLF

Hi all,

I have been field testing with the latest design. As suspected, it didn't work well. It was total useless in the BFO mode (phase change detection).

The ground detunes the resonant tank heavily. Also the geo-magnetic field has an affect to the response signal during tilting the coil. Varying the coil height above the ground is also critical.

So this proposal is definitely not the way to go.

The best way to go is as mentioned:
Synchronously demodulating the RX signal, where the reference clock comes from the TX part.

I have to find an elegant way to make an easy & robust PI design for the Laptop version. I know, the other PI project I am still following (the PI controller) would give best results. But's its complexity is high. I have to find a very easy method.

Cheers,
Aziz

[Aziz]

Hi all,

pitty, the simple circuit is so much sensitive, that it needs itself a shielding case (parts not placed in the coil yet). It very likely won't work well in the field due to the physical fact, that materials in the ground detune the resonant circuit. If the ground content has a big variation, then it's not worth all the effort.

A small piece of granite even gives a response . Sensing the detuning of the resonant circuit has obviously some disadvantages.
Another interesting fact is, if the resistive response dominates over reactive response, then discrimination won't work well! This can be observed with an iron plate and it's orientation to the coil.

A much better approach is still to take the reference clock from the TX LC tank and compare this against to the RX LC tank (like TGSL). So both LC tanks are affected in the same manner due to ground effect. But it requires an IB coil.

Nevertheless, true MONO coils can be used in a PI configuration. And I should focus to this section again.


Simonbaker:
The drifting and noise spikes from the sound card are causing false signals. But only, if I put the gain to the maximum position (noise spikes). The drifting is linear dependent of the gain. I suspect, the sound card output amplifier driver stage is responsible for this.

Mikebg:

The next step is design of wideband (PI) metal detector with soundcard. It should receive spectral information for target.

I have it already. Yes, it delivers a good spectral information of the target and a limited discrimination is possible. It currently works with IB coils only.

I should think of making a very simple PI (MONO coil configuration) using a sound card. Because it has many many advantages over VLF designs.

Aziz

[mikebg]

PI with soundcard

The next step is design of wideband (PI) metal detector with soundcard. It should receive spectral information for target. At pulse induction the MONOCOIL is damped by two 30ohm resistors inside computer. They appear in series as 60 ohm damping resistor as shown in the circuit below. The coil resistance 'r' makes additional damping.
The future metal detector should operate in two modes:
1. SEARCH mode operating in narrow band (no more than 16Hz) with sine induction to detect if there is target.
2. After receiving target signal, the software switches to wideband IDENTIFY mode with pulse induction to recognise target or to show in display spectral characteristic of target for visual identification by operator.

[simonbaker]

Hi all,

this is my latest simple, true discriminating Digital Laptop VLF detector. I have tested the circuit. Unfortunately, I have found some instability of my sound card output line (drifting and uncorrelated noise spikes at high output power level). So the performance is limited due to this fact a bit.

I have changed my experimental software to run on the positive phase shifts only (ferrites, iron, nickel cause negative phase shift -> lowering the resonance frequency). So it is running in the BFO mode only. I hope to make some real field tests soon.

1 EUR coin detection distance (air) with a 2.2 mH MONO coil at 19.5 kHz: 33 cm (not bad to start with)
Phase noise: better than +/- 0.001 degree.

Amplitude is thrown away due to sound card drifting problems (I have tested this with a closed loop cable, output line -> input line).

You will find the spice file below.

Aziz

I think that is excellent. I hope you can make a video of a test.

What do you mean by output line "drifting" ?

-SB

[Aziz]

Hi all,

this is my latest simple, true discriminating Digital Laptop VLF detector. I have tested the circuit. Unfortunately, I have found some instability of my sound card output line (drifting and uncorrelated noise spikes at high output power level). So the performance is limited due to this fact a bit.

I have changed my experimental software to run on the positive phase shifts only (ferrites, iron, nickel cause negative phase shift -> lowering the resonance frequency). So it is running in the BFO mode only. I hope to make some real field tests soon.

1 EUR coin detection distance (air) with a 2.2 mH MONO coil at 19.5 kHz: 33 cm (not bad to start with)
Phase noise: better than +/- 0.001 degree.

Amplitude is thrown away due to sound card drifting problems (I have tested this with a closed loop cable, output line -> input line).

You will find the spice file below.

Aziz

[simonbaker]

Hi all,

I have a digital VLF version, which achieves 9.9/10 KISS rating now.
Watch out, it uses only a MONO coil, no external battery required and all parts placed in the search coil:
just netbook + sound card + coil

The total parts count is reduced too. It can be extended into a fully differential version (will have double TX power). An external amplifier is not required. The sound card delivers enough power for this.

The TX LC resonant tank changes it's amplitude and phase (frequency). It's a combination of an off-resonance and a BFO type detector. As the TX coil voltage will be high, a simple resistive voltage divider will be used before it is sent to the input line of the sound card. These two parameters can be detected very precise using a digital Lock-in amplifier. I am quite surprised, that just the phase shift gives a very sensitive and robust discriminating signal. No more coil adjustments necessary. It's quite robust. This one is the simplest digital VLF, which I have ever designed.

I will start wiring the new circuit
Stay tuned...
Aziz

I'm very interested in this, Aziz, because I have been thinking about the possibilities of exactly such a mono coil design. If it works even reasonably well, it could be the basis for an easy-to-build MD -- except for the laptop -- but maybe if your algorithms can be streamlined they could fit into a microprocessor design.

I had an idea to reduce the need for fast ADC sampling for traditional IB TX-RX digital designs. The idea is to use a small capacitor to integrate the RX signal over 1/2 cycle and sample it only at the end of the 1/2 cycle, then discharge the cap with a JFet and start again. So only need to sample at 14.5 kHz instead of 500 kHz or whatever.

Basically it is similar to reducing the value of C12 in the TGSL and digitally sampling it every half-cycle instead of allowing C12 to integrate over the MD response time constant. The response time constant will be implemented by a digital filter in the software.

-SB

[mikebg]

Improved version of the synchronous oscillator

http://www.wipo.int/pctdb/images4/PCT-PAGES/2001/152001/01026213/01026213.pdf

Guess, whose invention is it? Same guy.

But this version has a -48 dB noise rejection!!!
Aziz

Fortunately, this link did not work, so I tried to figure out who is this guy. Vasil Uzunoglu is name which sounds familiar to me because I have often heard among the designers of ham radio. Ideas of Uzunoglu for synchronous oscillators are used in many RX projects and KISS designs made by amateurs. I am left with a misleading impression that Uzunoglu is a radio amateur designer from Greece. So I asked colleagues from Russia, using ideas, what they know about Uzunoglu. As a response I received as DJVU file an application for US patent pending and a reprimand, it is not polite to drive others to seek web information for me. When I searched for "Vasil Uzunoglu" and "V. Uzunoglu", I found really interesting things. There is even a sketch of UFO made by him, which he saw in 1966.

Most interesting is the content of his last US patent 6,667,666. It is written in an unusual style for a patent. The style is as if this is an article intended for publication in ham radio magazine. These articles are written in a style textbook. Inside the patent there are circuits repeated as well as the attached example.

Aziz, if you have the document from the above link, please check whether there is a substantial difference to the US patent 6,667,666.

[Aziz]

It's getting much simpler

Hi all,

I have a digital VLF version, which achieves 9.9/10 KISS rating now.
Watch out, it uses only a MONO coil, no external battery required and all parts placed in the search coil:
just netbook + sound card + coil

The total parts count is reduced too. It can be extended into a fully differential version (will have double TX power). An external amplifier is not required. The sound card delivers enough power for this.

The TX LC resonant tank changes it's amplitude and phase (frequency). It's a combination of an off-resonance and a BFO type detector. As the TX coil voltage will be high, a simple resistive voltage divider will be used before it is sent to the input line of the sound card. These two parameters can be detected very precise using a digital Lock-in amplifier. I am quite surprised, that just the phase shift gives a very sensitive and robust discriminating signal. No more coil adjustments necessary. It's quite robust. This one is the simplest digital VLF, which I have ever designed.

I will start wiring the new circuit
Stay tuned...
Aziz