Help needed for optimizing

[Tinkerer]

attached is a input signal conditioning circuit.

The scope shots show the signal as it arrives at the board.
The feedline is a twisted pair, plus ground from the center tapped RX coil.
Signal with target and without target.

A distinctive phase shift, as well as a change in amplitude can be seen.

The scope settings are 5uS/div and 50mV/div.

As is, the sensitivity gives 70cm for a 1 Euro coin. (In air, inside the house) The discrimination is outstanding.

Many things can be improved and I would much appreciate help for the improvements.

#1 Question: Impedance matching. The Twisted pair of the CAT5 cable has a nominal impedance of 100 Ohm.
A home made twisted cable with all 3 legs, that is, the twisted pair and the ground, all 3 twisted together, gives a cleaner signal of greater amplitude. Obviously, the ideal cable could improve the S/N considerably.
How should the ideal cable be?

SUGGESTIONS?

#2 Question: the signal conditioning circuit.
I am sure it could be improved.
I wonder how I should chose the capacitors of the filters?
The output should match the input of an ADC. Is this level shifter OK? Probably I should use an AGC amplifier, because of the large dynamic range needed.
SUGGESTIONS?

Tinkerer

[Aziz]

Field Testing?

Hi Tinkerer,

have you been field testing yet? If yes, how is the detector responding to the ground?
Did you test it against some mineral rocks? I.e. iron stone, ferrite, magnetite, etc.

Just for the curiosity, take a piece of granite and look at the RX signal.
I have a piece of it from the Austrian mountains, which is reacting like a ferrite.

1 EUR at 70 cm is really outstanding sensitivity. I am quite surprised about it.

Aziz

[Tinkerer]

Aziz,

thanks for the answer.

I have not done any field testing yet. The circuit is on the breadboard and I am trying to get any major errors and kinks out of it before I make a PCB.

The response to various stones is normal, that is, stones that contain iron or are slightly magnetic, like fired clay pottery, show their magnetic properties like iron.
Iron shows a signal that is of opposite polarity than gold, copper aluminium etc.

There is still a problem with certain very thin, large aluminum sheets or gold leaf. It detects a 3.5"x3.5" gold leaf of nanometer thickness at 8", but it recognizes it as iron.

There are still many problems to be solved.

Tinkerer

[simonbaker]

It seems you are steadily making progress. Good going!

-SB

[Aziz]

Aziz,
...
There is still a problem with certain very thin, large aluminum sheets or gold leaf. It detects a 3.5"x3.5" gold leaf of nanometer thickness at 8", but it recognizes it as iron.
...
Tinkerer

There it is (again): The skin effect.

Whereas the high magnetic susceptibility of iron force the eddy currents to flow on the very thin surface of the target, the very thin other targets behave of course same. No matter of their susceptibility in this case because, there is no other physical conducting path.

We can not discriminate 100% accurate. Ok, we have one problem less.

Please, go field testing. Forget the new PCB. Just do the field testing with the existing prototype. And get some hot rocks/granite for testing.

And take care to the geo-magnetic field effect (static magnetic field). Just tilt the coil and observe the behaviour of the detector and describe it here for further improvements. And rise up/down the coil from the ground.

Aziz

[Qiaozhi]

As is, the sensitivity gives 70cm for a 1 Euro coin. (In air, inside the house)

What size coil are you using?

[Tinkerer]

Qiaozhi
The outer TX coil winding diameter is 480mm. The design is a hybrid, neither PI nor VLF, but a mixture of both.
The TX could be classified as a multi frequency or extreme wide spectrum.

Aziz,

I agree that it is the skin effect that causes this problem. I believe I can solve it by increasing the amount of samples at different times. The TX frequency spectrum is spread over a certain amount of time (maybe we could call it CHIRP?, kind of?), therefore, sampling at different times is equivalent of different frequencies.

I definitely can not go outside with the RX front end on a stick-in protoboard. What I want to do is make several small PCB's that I can screw on an acrylic board for initial outside tests.
I have a Power supply PCB that works, even if not efficient.
The TX PCB also works fine.

The PIC is on another PCB.

Now I want to make the PCB of the circuit shown. On this PCB I need some options so that I can try different opamps.

Moving the coil around inside as far as the 2 meter cable allows, causes no problem in spite of the re-bar grid in the floor and all sorts of metal parts around. The big 3 foot steel blades of the overhead fan obviously generate a lot of noise when it runs.
The circuit "ground-balances" itself in tracking mode as far as the dynamic range allows.

However, this dynamic range is not yet wide enough.

This is a very initial stage of the design. It does look very promising, but many aspects still need to be improved. At the speed I am going, mostly trial and error, with my minimal electronics knowledge, it may take a long time to finish.

I am willing to share the design in exchange of help to finish the design in a short time.

Tinkerer

[Aziz]

Hi Tinkerer,

that's too early time to optimize the circuit. You need to go field testing. You will then find more problems to deal with.

Get the boards mounted together and look, what the ground and rocks say.

Aziz

[Tinkerer]

It seems you are steadily making progress. Good going!

-SB

Well, the goal is a detector that is capable of detecting a US$ 25c, at 1 meter depth. Still ways to go.
I think I am about halve way there.
The circuits are still very crude. There are many improvements possible, so the chances look good for reaching the goal one day.

By the way, I like your coil making. It makes it possible to produce very repeatable results.

Tinkerer

[Tinkerer]

Hi Tinkerer,

that's too early time to optimize the circuit. You need to go field testing. You will then find more problems to deal with.

Get the boards mounted together and look, what the ground and rocks say.

Aziz

I agree, only field testing will show how deep the high frequency waves reach into the ground. I have a feeling that the problem with the skin effect on very thin foil targets will solve itself this way. Not that I expect to find a lot of gold leaf buried out there
Now, another interesting experiment would be to see how deep the individual frequencies penetrate in different ground. But for that I will need Moodz's FPGA with the built-in oscilloscope, so that I can walk around with it.
Which frequency corresponds to which target response?
To find that out, I need to have a frontend that makes it possible to look at the responses on a very wide bandwidth.
But to get that wide bandwidth to the frontend I need a cable that carries this signal without changing it a lot, or at least I have to know what the changes are.

Tinkerer

[Aziz]

Hi Tinkerer,

yes, the FPGA board is quite useful instrumentation platform. Well, I have my netbook/laptop to see, how the ground, rocks and targets response.

I am onto making a new simple PI box, which is dedicated to field tests only. I tend to use simple MONO coils and simple PI architecture this time.

BTW, you don't need an ultra fast ADC. My 96 kHz sound card is capable of to deliver enough information. Some of the 24-bit codec chips could be driven at 192 kHz and should be well enough with the FPGA board.

And don't be disappointed when making field tests. It should be a new challenge to solve all the arising problems.

Here are some of the problems:
- static magnetic field (remanent ground/rock magnetisation)
- geo-magnetic field
- remanent magnetisation
- magnetisation relaxation
- ground conductivity (salt/wet grounds)
- ground susceptibility
- skin effect
- EMI

Don't focus on the discrimination problem too much. This is the one of less critical problems.

Aziz

[Qiaozhi]

Qiaozhi
The outer TX coil winding diameter is 480mm.

Hi Tinkerer,

that's too early time to optimize the circuit. You need to go field testing. You will then find more problems to deal with.

Get the boards mounted together and look, what the ground and rocks say.

Aziz

I agree.

Tinkerer - you should try the current setup outside before attempting to optimize the design further. There's nothing like the real world poking its nose into your "ultimate" design and giving it the finger!

Well, the goal is a detector that is capable of detecting a US$ 25c, at 1 meter depth

Blimey ... I hope you like digging!

[moodz]

Good work ....the impedance of the cable is one thing ... but what is the impedance of the coil and your front end input over the bandwidth of interest. An LT spice sim could be written to determine this but theory and practice are different .... the method I have used in the past is using a Return Loss bridge over a swept frequency range ...

http://www.qsl.net/n9zia/rlb/index.html

The AD8307 log detector is works from DC to over 500Mhz ... you would also need a variable oscillator / or a wideband source / or the working circuit ( conditions apply.)
ReturnLossBridge1.pdf

This circuit is for RF but could easily mod for low freq to DC operation.

[Tinkerer]

I agree.

Tinkerer - you should try the current setup outside before attempting to optimize the design further. There's nothing like the real world poking its nose into your "ultimate" design and giving it the finger!


Blimey ... I hope you like digging!

I think I used the wrong word. (optimizing) what I meant, is correcting the most obvious mistakes and add small improvements to the crude circuit.

Where I am, it takes considerable effort and lots of time to make a PCB. To go through all that and then to find out that because of some simple mistake or basic flaw, the PCB delivers only 20% of the design's capability makes one want to pull the hair out. (where I have not much left anyway)

the digging?? I will leave this for others. Maybe a good secondary market could develop, for a miniature Back-hoe, to dig the deep targets.

Due to the depth capability, a good FE discrimination is important. Imagine digging a 10 foot hole, only to find an old rusty pick? But how did it get there in the first place?

Tinkerer

[Tinkerer]

Good work ....the impedance of the cable is one thing ... but what is the impedance of the coil and your front end input over the bandwidth of interest. An LT spice sim could be written to determine this but theory and practice are different .... the method I have used in the past is using a Return Loss bridge over a swept frequency range ...

http://www.qsl.net/n9zia/rlb/index.html

The AD8307 log detector is works from DC to over 500Mhz ... you would also need a variable oscillator / or a wideband source / or the working circuit ( conditions apply.)
[ATTACH]13201[/ATTACH]

This circuit is for RF but could easily mod for low freq to DC operation.

Moodz, thanks for the links. How is your FPGA coming along? I am waiting anxiously.

With the differential input, I have noticed some quirks that I am trying to understand. It seems to be very important for the leads to and from the coils to be done in a certain way. Any deviation from the right way seems to generate notable differences at the input of the preamp.

I have been working on this by trial and error, but I am sure there are a few simple thumbs rules that could help me greatly in understanding helping to find a reasonable solution quickly.

For example, the many frequencies involved.

The rate of change in the response signal, from no target to full target, varies from about 3Hz to 20Hz.
The smallest targets have a frequency, or maybe I should say delta? of about 5uS. I noticed that the bandwidth of the Preamp needs to be about 200kHz, below that, the signal of small targets, shows notable attenuation.

Now, my synchronous sampling can vary between 10Khz to 1KHz.

Somehow, these frequencies interact. I don't know how, but I notice notable differences in the results caused by the interactions.

Then, in the house, there is the overpowering 60Hz mains noise. the system here is 120V, a single phase and Ground, whereby Ground is really ground, the power lines only have phase, no ground or neutral cable. This makes it that If I touch the scope probe, I get a 60Hz sine wave of somewhere 35 to 70V, depending on the humidity.

These 60Hz and the harmonics thereof, stick their ugly head out everywhere and grin at me. Any little piece of wire gives them a chance to wriggle themselves into my circuit.

Differential input helps a lot, but the common mode rejection is only as good as the matching of the inputs etc.

Some simple basic advise on these subjects would be most helpful.

Ah, by the way, in spite of all that, I get an easy 70cm in air on a 1 Euro coin. I know it can be improved.

Tinkerer