CW Ground Balance Theory

[Davor]

Instead of using a more common "VLF IB" I intend to continue using CW as a proper term describing metal detectors that use continuous signal for target excitation.

A topic about PI Ground Balance Theory is a success. The way I see it is, us, the community, regaining the balance over the patented technologies based on dubious theories and silly timing schemes. A true model of a process, described by proper theory, and confirmed by measurements, becomes obvious. Patents can't be granted to non-novel or obvious stuff, except maybe in Australia. So describing the way things tick makes more such things obvious, prevents dubious patents from appearing, and makes room for better detectors. So everyone wins except maybe patent trolls.

I think we need a parallel topic on CW ground balance.

In fact I think we need some more activity about CW technology as it will continue providing superior discrimination to any PI. There are several reasons to this.

-------------------------------------------

I believe the core problem of poor performance of CW detectors on difficult soils and salty beaches is in using a joint Geb and Disc. channel, the way most detectors do. I'll explain it in detail soon.

IMHO dual or multiple frequency rigs are not addressing the problem at it's core. Those are fine rigs, no question about that, but because of their complexity not just everyone's dish. It is possible to have fairly simple solution that works well using a configuration very close to the traditional - using a single frequency, and a few channels to play with.

I guess we'll have a basis for a superior CW rig at the end.

[WM6]

It is possible to have fairly simple solution that works well using a configuration very close to the traditional - using a single frequency, and a few channels to play with.

Best melody to my old tired hobby-detectorist ears.

[Davor]

The reason I believe a CW will remain superior for discrimination purposes to every PI around is skin depth invariability at single frequency.
Skin depth in conductive materials is proportional to 1/sqrt(f). Skin depth for copper at 10kHz is ~0.65mm and for gold is ~0.8mm. Double the frequency and you have 0.46mm for copper and 0.55mm for gold etc.

So everything is fine if you are testing targets at a single frequency, and all targets are thicker than their skin at a given frequency. Target identification is closely represented by a time constant (TC or tau) that is related to the material's conductivity. This effect makes foil appear as if it is made of some poor conductor, with shorter TC.

In PI technology a thick target or a bulk target will appear as a superior conductor, and TC appears to be larger than with CW. That's because the frequency content of PI spans from fairly low frequencies to over 100kHz, so skin depth can become quite large at low frequencies. Past the PI pulse frequency content plummets towards lower frequencies. This makes target ID time variant in PI. Perhaps it will enable readout of target thickness, but will never aid discrimination.

Hence for proper discrimination we are stuck with CW.

[WM6]

How all this is related to size and shape of the surface of target?

Par example: gold coin can be detected at least twice as deeper as gold nugget of the same weight but not of the same surface size and shape. Here I mean that not only skin depth at given frequency but eddy currents orientation (which can cancel at signal each other) can play some role in mono signal CW vs PI (which is structured as broadband signal).

[Altra]

Davor thanks for starting this thread. It will be interesting to see what comes out of it.

The classic single channel CW/TR has the two cases

1. Ground balanced to iron oxides - positive all metal response, including salt

2. Salt balanced - Negative response to iron oxides and ferrous metals, positive response from non ferrous metals

That somes up the single frequency CW problem. How to null salts and feO2 at the same time?

[Davor]

Perfect point!
CW/TR is perhaps the best starting point, and perfectly illustrates the issue.
Let's see how it goes...

There is one channel in all. It is sampled or switched at a phase that is in quadrature with offending ground signal. This results in zero response for ground phase, and some response for all other phases, hence "all-metal" response.

Iron oxides response is near the Tx phase, while salt response is at about 90° against Tx. All practical ground responses will be somewhere between these points, and the best part is that on any "normal" ground there is a unique phase that represents soil at some place. It is also the same regardless of bobbing, waving, or skewing of the coil. AFAIK no CW detector around has a GB channel adjustable in full 90° range.

What I can't be sure of is ... whether there is some exotic kind of homogeneous soil that would respond differently at different power, distance or skew.

[mikebg]

I posted the explanation for phase distortion at CW metal detectors in Doug's forum:

http://australianelectronicgoldprosp...8090/#msg28090

"Conventional VLF metal detectors use AIR signal across TX coil as phase reference for synchronous demodulation. This makes correct target identification only at air test or at shallow target.
However at bad ground or at deep target, the phase information is distorted by ground. We must use GND signal for phase reference and to calculate the true phase of target. For this purpose was created the search head with DOoD loop configuration."

http://www.geotech1.com/forums/showt...725#post169725

[Davor]

I'll give this a thorough thought, but I'm not quite certain it will work as you deduced. In my experience a bad ground does not rotate a phase that severely, but again, I have no experience with extreme grounds. I'll surely give it a second thought.

In any case, let's sort out ground balance first. and deal with discrimination quirks later.

[Carl-NC]

and the best part is that on any "normal" ground there is a unique phase that represents soil at some place. It is also the same regardless of bobbing, waving, or skewing of the coil.

This is true for an ideal coil, but non-ideal coils can show amplitude-dependent phase errors. For example, lift-off effects in a concentric coil.

AFAIK no CW detector around has a GB channel adjustable in full 90° range.

I think there are several, they have a "salt" mode which opens up the GB range from ferrite to salt.

What I can't be sure of is ... whether there is some exotic kind of homogeneous soil that would respond differently at different power, distance or skew.

As power increases, slew rate increases. Would this not alter the response of viscous soils?

[Davor]

Yes! That is the spirit!

I am very much interested in effects of viscous soil on CW detectors as experienced in practice. My viscous soil model is far from perfect, so learning about the effect from first hand experience will be very beneficial. If model matches the reality, I'll go with the model.

So far my experiments with Roman bricks did not yield any result that would suggest viscous soil to be exceedingly difficult, but I have no idea whether these bricks are mimicking the effect well enough. Guess not.

The part I did not test yet (and don't know if a pile of old bricks will do the job) is whether the discrimination criteria are affected by viscous soil, and if so, by how much. I don't expect too much skew because the soil represents the matrix, and from a waving coil point of view it is near static. A cyclo-static vector bumps against the motion filter, so this effect should be more notable with non-motion rigs.

[Davor]

There was a mention of automatic ground balance at Serbian most wanted...

If we find a way innovation auto GB I think it would be the strongest projects.

It is possible, but not trivial, and in any case it would require separating a GEB channel from Discrinination All-metal-phase-reference, which is one and the same at the moment.

There is a simple way of achieving auto GB, but not in the GEB/Disc channels configuration. I completely abandoned the idea of auto GB for IGSL because the very auto GEB function would become too complex. I intend to pursue this idea in my future LF build. One day...

Just a bit of explanation about GB.

There is always some kind of response, air or ground, and it may be considered static, or more correct cyclostationary. This component is removed by motion filters. In case of non-motion detectors it is removed by "auto tune" or some other motion-compensation-in-disguise. In both cases the stationary component is removed and it is NOT ground response.

The ground response comes as a signal at a phase which is more or less constant over patches of ground with more of less similar properties. Waving a coil on a same distance from ground would not produce much amplitude change of this signal, hence the proper way of provoking ground response is pumping action (bobbing). This signal, at one place, would have changing amplitude, but a constant phase. That's a key to removing it.

The way to remove ground response, but keep everything else, is by synchronous detection at 90° against a ground response phase. This functions as a notch, so whatever the amplitude of ground response, resulting voltage will be zero. Such detection will provide all-but-ground indication, or more commonly known as "all metal".

.....
Now about automatic GB...

Ground phase comes in a range of phases from 0° (ferrite) to roughly -90° (salt water), hence it is balanced out by synchronous detection with phases from +90° (ferrite) down to 0° (salt water). To establish any kind of working automatic GB it is important to detect a phase, while the changing amplitude of ground response is just trouble in a way of a simple solution.

Phase can be detected directly from a signal at some frequency by some of the phase detectors, provided the amplitude is constant.

In a quadrature zero IF receiver - which is a typical topology of nowadays CW detectors having "X" and "Y" channels, phase can be established as arctan(X/Y) with troublesome outcome for small amplitudes.
Things are simple enough with Y larger than X, and with a limited range of phases required for proper ground balance, it is a workable solution, as it may work in a loop that keeps X at zero while changing the GB phase.
In any case, automatic GB requires it's own set of quadrature channels if we want things remotely simple.

...
There are also questions about interaction between automatic GB adjust and targets...

Ground response greatly coincides with ferrous response, so excluding responses in wrong quadrants may work beneficially only for non-ferrous targets.

Exclusion by amplitude may work disadvantageous for small targets, as ground balance will be mis-tuned in presence of small targets, so very soon such targets are lost as ground balance wanders off and reduces sensitivity in a process. It can be aided by some sort of semi-automatic function, but it is not a serious improvement over a manual GB.

Exclusion by motion seem promising, as target responses are usually much sharper than ground ... unless it is a sluggish deep target response.

The safest option seem to be a combination of everything above. Not trivial at all.

[Carl-NC]

Automatic ground tracking is always a design of trade-offs. Some approaches are constantly thrown off by small iron targets, some track out weak conductors like small gold nuggets. The safest method is to run with tracking locked (but still internally accumulating) and use a Grab method to manually update it periodically.

I once was in a prospecting seminar where a rep for a major detector manufacturer emphatically stated that all serious gold detectors use auto ground tracking all the time, and there is no need for a manual mode. The truth is, there are pros & cons to both methods, and this is an area where user knowledge is generally very weak.

[Davor]

Guess the best approach would be semi-automatic after all. Most of the errors are avoided if you intentionally engage in GB tuning, and bob the coil to give it the maximum ground it can get, at a place you can't sense any targets.