frequency depth relation diagram

[6666]

Tinkerer
any possibility of a higher resolution schematic please.

[Tinkerer]

RX test schematic

Here is the pdf. document.

Tinkerer

[Davor]

I like the way you employ SAH in your Rx.
Do you maybe have spice simulations of your TEM setup? It would be very interesting to match simulation with reality.

[Tinkerer]

I like the way you employ SAH in your Rx.
Do you maybe have spice simulations of your TEM setup? It would be very interesting to match simulation with reality.

No, I have no Spice simulation of the RX circuit. It is a work in progress, where I try all kinds of different kind of approaches on the bread board and then change the circuit accordingly.

Here are the test results:


Thickness test 07-16-2012

Coil diameter 45cm, not shielded.

All targets 25cm above the coil.

All targets diameter about 21mm

All the thicker copper pieces are cut from the same rod.

The #2 is cut out of an annealed copper sheet

The #3 is a copper coin


Sample #1- PCB, single sided .................. 200mV
Sample #2- 0.5mm copper disk .................. 500mV
Sample #3- 1mm copper disk .................. 500mV
Sample #4- 3.4mm copper .................. 620mV
Sample #5- 9.2mm copper .................. 1000mV
Sample #6- 20mm copper .................. 1250mV
Sample #7- 28.6mm copper .................. 1600mV
Sample #8- 40mm copper .................. 2000mV
Sample #9- US$ 5cent or nickel 1.8mm thick .... 400mv
Sample #10- Lead disk, 1.5mm thick ............ 400mV
Sample #11- Aluminum foil ................. 50mV

After that I will reprogram the PIC timer for a different timing schedule and repeat the test to see if there are notable differences.

The next test will then be with a TX pulse aimed at exciting higher response in the thicker targets.

Any suggestions are welcome.

Tinkerer

Hey, I am disappointed in you guys. Nobody took a guess as to which targets the first 2 measurements corresponded.

[Aziz]

Hi all,

I recently tested a standard PI damping transmitter and thought finally: "What a shame to waste so much power!". The induction balanced TEM mode gives so much more benefits.

Thanks for your measurement results Tinkerer.

Cheers,
Aziz

[Tinkerer]

Hi all,

I recently tested a standard PI damping transmitter and thought finally: "What a shame to waste so much power!". The induction balanced TEM mode gives so much more benefits.

Thanks for your measurement results Tinkerer.

Cheers,
Aziz

Hi Aziz,
And nobody should forget that you analyzed, posted and explained the method for everybody to use it.

Here is the sequel.

I increased the power by about 50% and reduced the pulse repetition rate by about 40%.

No changes were made in the RX circuit.

We can immediately see that the reduction in PRR, pulse repetition rate, has a very strong effect on the results.

Just increasing power brings little benefit. If changes are made, they must be made throughout the circuit, not only in the timing and power.

For a fast target response we have only about 10ms for the processing. With a high repetition rate, we can stack more samples during that time. This results in better S/N.

Thickness test 07-17-2012
About 50% more power and 60% PPR for the second column
Coil diameter 45cm, not shielded.

All targets 25cm above the coil.

All targets diameter about 21mm

All the thicker copper pieces are cut from the same rod.

The #2 is cut out of an annealed copper sheet

The #3 is a copper coin


Sample #1- PCB, single sided .................. 200mV .... 125mV
Sample #2- 0.5mm copper disk .................. 500mV .... 250mV
Sample #3- 1mm copper disk .................. 500mV .... 270mV
Sample #4- 3.4mm copper .................. 620mV .... 300mV
Sample #5- 9.2mm copper .................. 1000mV .... 500mV
Sample #6- 20mm copper .................. 1250mV .... 650mV
Sample #7- 28.6mm copper .................. 1600mV .... 800mV
Sample #8- 40mm copper .................. 2000mV .... 1000mV
Sample #9- US$ 5cent or nickel 1.8mm thick .... 400mv .... 225mV
Sample #10- Lead disk, 1.5mm thick ............ 400mV .... 200mV
Sample #11- Aluminum foil ................. 50mV .... 30mV

For the next test I will go the other way.
Higher cycle rate and lower power.

Let's see what the results are going to be.

Does anybody want to suggest RX circuit changes for correcting the difference in PRR?

Tinkerer

[Davor]

There is nothing wrong with your Rx, and having all other components super sexy as they are the only difference could be, say, MAX4066 that is a somewhat better switch than the garden variety 4066.

Faster repetition means more portions of the good stuff, so why not cranking it up all the way to the practical limit. Guess it will be somewhere around the range where it beats up posh PI detectors stick:

[Tinkerer]

There is nothing wrong with your Rx, and having all other components super sexy as they are the only difference could be, say, MAX4066 that is a somewhat better switch than the garden variety 4066.

Faster repetition means more portions of the good stuff, so why not cranking it up all the way to the practical limit. Guess it will be somewhere around the range where it beats up posh PI detectors stick:

Davor,

now it is your turn to calculate the partial signal amplitude caused by the skin effect.

After your explanations, I will explain where the additional gain with increased PPR, comes from in that circuit.

The idea is to use the ADG412 analog switch. But I have a CD4066 at hand.

Below are the results with the PRR cranked up to 10,000.
The power reduced to 50% of first test. We can see that the power was not the important part.

Thickness test 07-17-2012
POWER REDUCED TO 50% OF FIRST TEST – PRR INCREASED TO 10,000
RESULTS IN THIRD COLUMN
Coil diameter 45cm, not shielded.

All targets 25cm above the coil.

All targets diameter about 21mm

All the thicker copper pieces are cut from the same rod.

The #2 is cut out of an annealed copper sheet

The #3 is a copper coin

Sample #1- PCB, single sided .................. 200mV ...125mV ...225mV
Sample #2- 0.5mm copper disk .................. 500mV ...250mV ...500mV
Sample #3- 1mm copper disk .................. 500mV ...270mV ...575mV
Sample #4- 3.4mm copper .................. 620mV ...300mV ...600mV
Sample #5- 9.2mm copper .................. 1000mV ...500mV ...900mV
Sample #6- 20mm copper .................. 1250mV ...650mV ...1250mV
Sample #7- 28.6mm copper .................. 1600mV ...800mV ...1600mV
Sample #8- 40mm copper .................. 2000mV ...1000mV ...2000mV
Sample #9- US$ 5cent or nickel 1.8mm thick .. 400mv ...225mV ...400mV
Sample #10- Lead disk, 1.5mm thick ............ 400mV ...200mV ...350mV
Sample #11- Aluminum foil ................. 50mV ...30mV ...75mV

There are still things to be optimised in the circuit, but it looks promising.

Any suggestions, anybody?

Tinkerer

[Davor]

To calculate what? You've done everything already

By your measurements thicker objects have somewhat thicker skin

Actually, I guess more of the skin is bathing in TEM field and that could explain the difference. I'd say that PCB and tin foil have their saying in this.

Perhaps a 2 tone VLF would be able to estimate target thickness.

BTW, I'm not sure how happy AD8055 is with unity gain operation. I'd expect some ringing in the configuration at hand.

[Tinkerer]

To calculate what? You've done everything already

By your measurements thicker objects have somewhat thicker skin

Actually, I guess more of the skin is bathing in TEM field and that could explain the difference. I'd say that PCB and tin foil have their saying in this.

Perhaps a 2 tone VLF would be able to estimate target thickness.

BTW, I'm not sure how happy AD8055 is with unity gain operation. I'd expect some ringing in the configuration at hand.

Thanks for the feedback.

We see a difference between the copper, nickel and lead, due to the difference in conductivity.

But the thickness very obviously makes the bulk of the response.

What about the frequency? How do you convert frequency into step response? Because it is the di/dt of the switch OFF that generates the eddy currents.

The AD8055 is specified to drive a 100 Ohm load at Gain+1.
We don't need the high speed though, maybe we can find a better buffer with high drive capability.

We are driving a high capacitance load. I have not yet looked for the best buffer to do that.

Like usually, I start with what I have in the parts box and when things work more or less, I start optimizing. Lots of Datasheets to read and many Datasheets of many, many pages.

Below, follow some explanations, Maybe you or somebody could correct them?

The second column shows the results with 50% more power and 4000PRR
Sample times are not changed.
With the sample times equal, the duty cycle is where the gain comes from.
At 10,000 cycles and 7us sample time, we have a duty cycle of about 7%
At 6,500 cycles and 7us sample time we have a duty cycle of about 4.5%
At 4,000 cycles and 7us sample time we have a duty cycle of about 2.7%
So we could say, considering the difference in duty cycle when we look at the power, we get a power correction factor of:
Power: this is an approximate number, the meter is not calibrated.
10,000 cycles at 200mA = 1
6 500 cycles at 400mA =*1.55, duty cycle correction factor.
4,000 cycles at 600mA = *2.6, duty cycle correction factor.
So we can see that the power does make a difference, but the duty cycle makes a bigger difference.
Now, let’s look at another power factor difference.
The field density. With an equal coil diameter, the field density is higher so it should excite the same eddy currents in a target of 3 times smaller surface area.
This does not show with this test, because of the same surface area presented to the coil for all targets.
Still another factor to look at, is the di/dt. Is it the same for all 3 tests?

Ideas? suggestions?

Tinkerer

[Davor]

di/dt converts to voltage by virtue of coil inductance. di/dt will not be exactly the same because of the target coupling to the Tx coil which changes a Tx coil inductance. At 25cm distance you can say this coupling is negligible. Strictly speaking - yes, but practically - no. Smaller targets will have smaller coupling, and my guess is that their voltage response would go by their surface area.

What about the frequency? How do you convert frequency into step response? Because it is the di/dt of the switch OFF that generates the eddy currents

You don't. It is the voltage that matters, and in PI voltage is a byproduct of di/dt at switchoff. Even with VLF and it's phase shifts - these shifts are just byproducts of delays. A step response gives you a wide frequency range response. There is only a small SNAFU with single sample that prevents you from doing any frequency analysis, and instead you are constrained to ~ 1/2T center frequency. In your case and 7us it will be about ~70kHz. More samples would yield more frequencies, but only in case they are properly weighted. This is something I intend to play with after my IGSLs start making money

I think sampling can be done by a several daisy chain sampled channels with gains adjusted for an optimum weighting function, and thus enabling some FFT-like analysis of the samples.

[6666]

Tinkerer thanks for the PDF

[Funfinder]

Thanx for all the new info but it seems this thread had been hijacked!
Please let's move back to frequency depth relation, thx.

There has to be a relation between frequency and penetration depth otherwise someone simply could use his modified FM-transceiver for metaldetection.

And of course power too, otherwise radar-stations wouldn't need such high energy-consumption.

What I see is that the ground works also as a reflector of the waves and not just as a "consumer". The higher the frequency the more energy gets reflected.

That's why finally an above the coil reflector must be created to reduce the wasted energy that radiates into direction sky.

I doubt that there is a possibility to "saturate" the soil by radiation.
The ground is no closed capacitor - it is just a medium comparable with a shallow lake with alot dust particles that only contains sunlight in the 5meters below the surface and not any deeper.

But if the sun would shine with the doubled brightness of course the sea would be illuminated deeper.


Yes, and then there is the timing, important for the reception circuit and search speed.

1Hz = 1 positive plus one negative amplitude per second
and has a wavelenght in no relation to the coils-diameter

So what I think is that 2 different coil and detection circuits
are needed. the question is: where is the border-line?

I think the border is around a 5 x 5cm object (no matter how thick) at 50cm.

a 15cm coil should be still able to reach this and perhaps
even at 100kHz so it also will detect extremly small stuff.

a 100cm coil should be able to reach the 5x5cm object at
50cm, too, but as its smallest limit and at even 500Hz.
But this coil has to go extremly deep because of the
very low frequency and the huge diameter.


All those medium sized coils with medium frequency
are just a lousy compromise!

[Midas]

Thanx for all the new info but it seems this thread had been hijacked!
Please let's move back to frequency depth relation, thx.

There has to be a relation between frequency and penetration depth otherwise someone simply could use his modified FM-transceiver for metaldetection.

And of course power too, otherwise radar-stations wouldn't need such high energy-consumption.

What I see is that the ground works also as a reflector of the waves and not just as a "consumer". The higher the frequency the more energy gets reflected.

That's why finally an above the coil reflector must be created to reduce the wasted energy that radiates into direction sky.

I doubt that there is a possibility to "saturate" the soil by radiation.
The ground is no closed capacitor - it is just a medium comparable with a shallow lake with alot dust particles that only contains sunlight in the 5meters below the surface and not any deeper.

But if the sun would shine with the doubled brightness of course the sea would be illuminated deeper.


Yes, and then there is the timing, important for the reception circuit and search speed.

1Hz = 1 positive plus one negative amplitude per second
and has a wavelenght in no relation to the coils-diameter

So what I think is that 2 different coil and detection circuits
are needed. the question is: where is the border-line?

I think the border is around a 5 x 5cm object (no matter how thick) at 50cm.

a 15cm coil should be still able to reach this and perhaps
even at 100kHz so it also will detect extremly small stuff.

a 100cm coil should be able to reach the 5x5cm object at
50cm, too, but as its smallest limit and at even 500Hz.
But this coil has to go extremly deep because of the
very low frequency and the huge diameter.


All those medium sized coils with medium frequency
are just a lousy compromise!

Heres an experiment to remove some of the variables for you Funfinder. Bury a coil at some depth with wires leading to the surface, then into an O-scope. Stick another coil on the surface and attach it to a frequency generator (ideally one that regulates its power usage to the same level regardless of frequency, though perhaps not essential). Run it through a range of frequencies and record the amplitude on the buried receive coil. Then repeat the experiment with the coils spaced at the same distance, but this time with an air gap instead of soil. Compare the two data sets to see how much reduction in transmission the soil has at all the different frequencies. Then of course report your results here.

Regards
Midas

[Skippy]

Would this experiment be better if the buried transmitter were target-sized? It would not be hard to make a 30mm diameter coil with suitable inductance, like 1mH.