Ground Balance Theory

[Davor]

I think I've got a candidate for a viscous soil "target" spice model. It can be "convinced" to behave, but tuning it will be quite interesting. I simply replaced a LR target's "R" with a modified behavioural U/I combination, otherwise handy for creating voltage controlled potentiometers, but in a way that resistance is proportional to the absolute value of a current passing through it. Hence the resistance gets lower with time.

If the "V=f(..." term is reduced to V={R}*I(Vsens1) the Vsens1 and B1 combination would perform as a resistor of R ohms.

So far I'm off any comparable result but it is promising. It does have a linear response in a log/log scale.

I'd be very happy if this could be tuned for everyday use, but I have zero experience with viscous soils. If someone gets lucky tuning this, please share with rest of us.

[Ferric Toes]

I think I've got a candidate for a viscous soil "target" spice model. It can be "convinced" to behave, but tuning it will be quite interesting. I simply replaced a LR target's "R" with a modified behavioural U/I combination, otherwise handy for creating voltage controlled potentiometers, but in a way that resistance is proportional to the absolute value of a current passing through it. Hence the resistance gets lower with time.

If the "V=f(..." term is reduced to V={R}*I(Vsens1) the Vsens1 and B1 combination would perform as a resistor of R ohms.

So far I'm off any comparable result but it is promising. It does have a linear response in a log/log scale.

I'd be very happy if this could be tuned for everyday use, but I have zero experience with viscous soils. If someone gets lucky tuning this, please share with rest of us.

I've never got around to using Spice, although I use other simulators from time to time. You say that your simulation has a linear response in a log/log scale, and presumably the settings can be adjusted to alter the slope? What we are looking for is a negative slope of 1, which is the theoretical value for non interacting SPM particles. In practice the slope can vary a bit either way from 0.9 - 1.1. This I have measured for a wide range of materials from different sources on my viscosity meter. Other sources of measurement report a wider spread, but this I believe to be due to non ideal instruments by which the measurements were done. The viscosity meter I have is in use in the geophysics dept of three different universities and two geophysical survey companies and none have yet reported errors in these figures. Some exotic rocks or particle assemblages may well fall outside the -+10%, but the measurements so far have been for soils, sands, rocks, and fired clays that would normally be encountered in the hobby of metal detecting. What is encountered in the field appears to be small changes in slope and large changes in amplitude.

Eric.

[Tepco]

Coil was probably 6in D at most. Vertical scale probably 3 decades. This is raw data straight off the preamp with no processing.

Eric.

This may explain discrepancy in results. My coil was large, significantly larger compared to target size, made of solid conductor (proper Litz wire unavailable at a time) also different scaling, so coil response most likely overshadowed the desired response. Have to recheck this in a different setup.

[Davor]

I've never got around to using Spice, although I use other simulators from time to time. You say that your simulation has a linear response in a log/log scale, and presumably the settings can be adjusted to alter the slope? What we are looking for is a negative slope of 1, which is the theoretical value for non interacting SPM particles. In practice the slope can vary a bit either way from 0.9 - 1.1. This I have measured for a wide range of materials from different sources on my viscosity meter. Other sources of measurement report a wider spread, but this I believe to be due to non ideal instruments by which the measurements were done. The viscosity meter I have is in use in the geophysics dept of three different universities and two geophysical survey companies and none have yet reported errors in these figures. Some exotic rocks or particle assemblages may well fall outside the -+10%, but the measurements so far have been for soils, sands, rocks, and fired clays that would normally be encountered in the hobby of metal detecting. What is encountered in the field appears to be small changes in slope and large changes in amplitude.

Eric.

I gave this simulation target another go to learn more about the slope and stuff. The variable resistance approach was a real deal, only the abs (I...) part needed some extra boost to behave. I must put it on power of 6 or more to approach a slope of -1. With the original value it is closer to the slope of -2. I instantly learned a lot about the viscous soil. It is so incredibly dependent on the coupling, hence a monocoil is perhaps the stupidest setup you can have on such soils.

I'm not certain if this model will ever be a verbatim representation of viscous soil, as it is obvious it can't go below a slope of -1, but perhaps it can be useful as a learning tool.

[Ferric Toes]

I gave this simulation target another go to learn more about the slope and stuff. The variable resistance approach was a real deal, only the abs (I...) part needed some extra boost to behave. I must put it on power of 6 or more to approach a slope of -1. With the original value it is closer to the slope of -2. I instantly learned a lot about the viscous soil. It is so incredibly dependent on the coupling, hence a monocoil is perhaps the stupidest setup you can have on such soils.

I'm not certain if this model will ever be a verbatim representation of viscous soil, as it is obvious it can't go below a slope of -1, but perhaps it can be useful as a learning tool.

Yet, in the testing I have done in Australia, a mono coil works well, and because it is simple with no response anomalies it would still be my choice for a coil. Having said that, I would now opt for the bifilar centre tapped mono because of its additional benefits. Last time I was in Oz, I compared a mono TX/RX with a concentric coplanar TX plus RX and found no benefit in the latter except that the amplitude of the ground response is less when the coil is close to the surface. Once the coil is above the surface by a distance equal to the difference of the radii, the ground signal is the same as of an equivalent mono. Of course the dynamic range of the receiver chain has to accommodate the large and rapid increase in amplitude as a mono coil approaches the ground surface, but with care in design that has not been a problem.

Eric.

[Davor]

I guess the most benefit from decoupling Tx and soil/Rx will be obtained with some IB setup. I often wondered what is the source of praises of DD coils on "mineralised" soil and it makes sense now. I'd say the best bet with bad soils is a proper coil choice.

I'll surely try LF CW detection. I expect it to behave just fine on bad soils.

[Ferric Toes]

I guess the most benefit from decoupling Tx and soil/Rx will be obtained with some IB setup. I often wondered what is the source of praises of DD coils on "mineralised" soil and it makes sense now. I'd say the best bet with bad soils is a proper coil choice.

I'll surely try LF CW detection. I expect it to behave just fine on bad soils.

For PI, DD coils do run quieter on bad soils if the GB arrangement in the detector is not optimal, i.e. if there are non-linearities prior to subtraction in the RX circuitry at high signal levels. DD, or another form of balanced coil, is also essential for measurements in the ON time. With DD there is also a loss in detection range as the individual coils are smaller than the same size mono. You can of course go up in size using, say a 18in DD to get the same range as a 14in mono. Horses for courses, as the saying goes.

Eric.

[Ferric Toes]

Soil Electromagnetic Properties and Metal Detector Performance.

Here is a a very good paper for both FD and TD enthusiasts. Aziz and Mikebg should enjoy it. Looks like all the theoretical information is already out there, so all we need is some hardware.

http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA509654

Thanks Doug. By the way, your link is HTML and no diagrams.

Eric.

[Aziz]

Here is a a very good paper for both FD and TD enthusiasts. Aziz and Mikebg should enjoy it. Looks like all the theoretical information is already out there, so all we need is some hardware.

http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA509654

Thanks Doug. By the way, your link is HTML and no diagrams.

Eric.

Eric,
the best part of the document is the list of the references (where it has been copied, adapted and arranged).

Aziz

[Prospector_Al]

Hi Aziz et al,

It's a very good paper indeed! And the references are very useful. Owing to the frequent use of mathematical formulae, some background knowledge is needed to fully understand what's being described.

Here are some further references for those who really want to delve into the intricacies of the theory:
One of my textbooks: Electromagnetic Fields and Waves, Paul Lorrain and Dale Carson; Applied Geophysics, W.M Telford et al.

Now the really good news: This paper looks at the problems from a scientific point of view. A practical metal detector will be designed by an engineer who knows about Occam's Razor, and applying that criterion, it turns out that understanding the minutiae of the ground signal characteristics is not necessary to design the best possible metal detector.

The proof will appear in patent applications and in the market place...

Just my guess (but an educated one)...

Allan

[mikebg]

Here is the DOD loop system for realisation of algorithm for ground balance described in post #77

http://www.geotech1.com/forums/showt...561#post168561

Search head has 3 receive coils generating signals AIR, RX and GND.

The project was realised in 2008:
http://www.geotech1.com/forums/showt...0200#post70200

[Aziz]

Hi Aziz et al,

It's a very good paper indeed! And the references are very useful. Owing to the frequent use of mathematical formulae, some background knowledge is needed to fully understand what's being described.

Here are some further references for those who really want to delve into the intricacies of the theory:
One of my textbooks: Electromagnetic Fields and Waves, Paul Lorrain and Dale Carson; Applied Geophysics, W.M Telford et al.

Now the really good news: This paper looks at the problems from a scientific point of view. A practical metal detector will be designed by an engineer who knows about Occam's Razor, and applying that criterion, it turns out that understanding the minutiae of the ground signal characteristics is not necessary to design the best possible metal detector.

The proof will appear in patent applications and in the market place...

Just my guess (but an educated one)...

Allan

Hi Allan,

the book recommendation is very fine. The book is really good.

Ok, this section is for poor students only:
(I want the proof of, that you are poor )
1. Start JDownloader
2. Start Internet-Browser and goto site: http://www.filestube.com
3. Enter what you want to search for
4. Select a working link and copy it to the clip-board (JDownloader gets active)
5. Start downloading
6. Inspect what you have downloaded
7. Go and buy the hardcover (if you can afford it)
Example: (don't visit the site!)
http://www.filestube.com/query.html?...ves&select=All


WTF, Occam's Razor?
http://www.filestube.com/query.html?...ect=All&page=1


Data mining?
http://www.filestube.com/query.html?...ing&select=All
Yes, data mining concepts (based on measured facts) work much much better to ignore the (pseudo) scientific papers and mind-dribbling patents. Data mining will make patent-trolls life very hard. And you don't need the scientists with their full of B$ and faulty papers. A good scientist will use data mining concepts as well.
BTW, the WBGB is a result of data mining concept!


Cheers,
Aziz

[Tepco]

Good-bye civilization

Good-bye civilization, back to analog scopes. Just improvised one simple LOG amp setup to try few tests. This is temporary solution, signal displayed directly from log amp video output. Serious noise problems prevented DSO use, also this is short of at least 30-40dB additional dynamic range available but still can be interesting and informative to someone. With different design (sampled, integrated and time resolved video output) i want to build test rig specifically to test different soil responses, over 80-90dB dynamic range and various pulse widths, down to 1us or less. Lot of fun with log amps, also lot of observations and conclusions to be done. Last time i tried this, big fat coil and big fat pulse prevented me to make useful measurements. On picture below, horizontal scale is 200uS (20uS\div), vertical not calibrated, around 50dB, ignore top 1 division. Analog display is better for noisy signals, tried averaging on DSO but in fact DSO monitor itself produce significant interference, so dusty old 1740 did the job.

[Ferric Toes]

Good-bye civilization, back to analog scopes. Just improvised one simple LOG amp setup to try few tests. This is temporary solution, signal displayed directly from log amp video output. Serious noise problems prevented DSO use, also this is short of at least 30-40dB additional dynamic range available but still can be interesting and informative to someone. With different design (sampled, integrated and time resolved video output) i want to build test rig specifically to test different soil responses, over 80-90dB dynamic range and various pulse widths, down to 1us or less. Lot of fun with log amps, also lot of observations and conclusions to be done. Last time i tried this, big fat coil and big fat pulse prevented me to make useful measurements. On picture below, horizontal scale is 200uS (20uS\div), vertical not calibrated, around 50dB, ignore top 1 division. Analog display is better for noisy signals, tried averaging on DSO but in fact DSO monitor itself produce significant interference, so dusty old 1740 did the job.

Looks like pictures of that meteorite that came down over Russia . Can't beat a good ol' analog scope at times. It's good that you are seeing some differences now.

Eric.

[Ferric Toes]

I've just done a couple of plots and superimposed them. Linear plot this time and shows the decay of a piece of Oz ironstone compared to that of a 10gm nugget. Both were initially set at an arbitrary 1000 units amplitude at 20uS. It looks as though the nugget starts at 950, but it does curve up to 1000 due to early time skin effect. Curve fitting software cannot fit mixed exponentials. Optimum nugget/ground SN seems to be 30uS

Eric.