A pure ground is a soil condition that reacts like it was pure ferrite.  In other words a perfect magnetic condition where no electrical conduction (eddy currents) takes place.  We can think of this as a soil that produces a signal in the detector with zero phase shift relative to the transmitted signal.  This is considered our reference signal of zero phase to which all other signals can be referenced to.  Of course the only real life object that produces this type of signal is pure ferrite.  So ferrite becomes our reference target and produces what we call a pure "X" reactive signal.

Of course real ground conditions do not behave like pure ferrite.  When subjected to a detectors magnetic field small currents begin to flow in the soil.  This will cause the soil signal to be displaced slightly from that of pure ferrite.  We call this difference a phase shift and define it to have an angle in degrees negative relative to pure ferrite.  In addition, this phase shift produces a new signal in the detector which we call the "R" component signal.  We can carry this analysis one more step.  Using Trigonometry the ratio of the X signal to the R signal can be shown to be the actual measured phase of the ground.

All grounds have varying amounts of magnetic and conductive properties.  Therefore, the ratio of the X or magnetic signal and R, the conductive signal, will vary from one location to another.  However, the phase produced by this characteristic will always be negative relative to zero, the phase of pure ferrite.

From my experience most grounds produce a phase that falls somewhere between zero (ferrite) and a -5 degrees.  Some highly magnetic soils can have a phase that is quite low, but it can never be zero.  Once the phase exceeds several degrees the ground characteristics begin to fall into an area where it becomes more saline. This doesn't  mean that its not magnetic.  Its just that the R or conductive component of the ground becomes stronger in relation to the magnetic portion.  Thus the phase becomes greater.

The manual ground adjustment works in this manner:  When you position the “Ground Adjust” control to the phase of the target,  in this case the ground,  any up or down motion of the coil does not produce a corresponding change in the audio volume.  For example, when you position the control to zero phase, and then move a piece of ferrite around near the coil, the audio volume will not change.  In other words you have balanced out to the ferrite.  However, if you now lower the coil to real ground the audio will increase in volume.  Of course this indicates that you are not balanced to the ground.   As you begin to turn the control counter clock wise the ground adjust control phase changes from zero to a more negative amount.  Once you have reached the point of  “ground balance” the control and ground phases match.   Of course as the coil is moved to various locations the ground phase changes slightly and you must readjust the control for a neutral reaction.  As you can see there is no one control phase position that matches every condition since the ground phase varies from one location to another.

The introduction of the Motion detector solved this problem.....sort of.  In a Motion detector design you can calibrate the “fixed” ground adjust control phase to approximately +0.5 degrees and set the audio threshold for silent operation.  If that is done the detector will appear not to respond to the ground.  In reality it is responding.  Its just that you don’t hear it since all ground reactions cause the audio to decrease in volume.

And since the audio is already silent you don’t hear anything.  Remember I said that all real targets, which includes the ground, have a phase between zero and some negative value.  The preset ground control phase of +0.5 degrees is in a location where no real targets ever exist. Therefore, you never have a condition  where you are balanced to anything, least of all the ground.  As you move the coil over the ground, the internal detector signals are continually being driven negative.  Any weak positive target signal is easily over-ridden by the huge negative ground signal.  Of course, if the target is close enough to the coil its positive signal can override the negative ground signal and you will hear the reaction in the audio.  The greater the phase and strength of the negative ground signal the more it will mask the positive target signals.   A manual ground balance design would avoid this since the operator can adjust the control for a (near) neutral reaction on the ground.

For fixed machines the phase error between the internal “ground preset balance” and the actual ground condition can be much more than “slight”.  The internal preset is calibrated for +0.5 degrees. This is in an area where a real ground phase never occurs.  The actual ground phase may be -2 or -3 degrees “negative“.  That’s a huge difference, maybe 2.5 to 3.5 degrees.  This much phase error will in effect cutoff several inches of detection depth.

When fixed ground balance (motion) machines first came out I was opposed to using this technique.  I knew it was in some ways a trick into fooling the customer that there was no ground balance.  The control was simply a fixed internal adjustment.   However, the pressure to compete in the market place was enormous.  So, I eventually gave up the argument and designed my first detector using a fixed ground balance the “Big Bud”.

The standard loop size is the best size to use on a fixed GB detector since it was the coil most likely to which the detector was designed.  If you read my post to Reg you will see that I took extra care to try to insure that the other loop sizes meet the same characteristics as that of the standard coil.  It is also true that larger coils pickup the ground more than smaller coils.  So any phase errors due to a detector-coil mismatch will make this problem worse.  The only sure way to get around this is in using a detector with a GB mode and a manual ground balance.

If the ground is very heavily mineralized due to natural mineralization or  pollution, a fixed GB machine would probably be of no value.  An error of several degrees, as I point out above, will translate into a negative offset totally masking all targets.  To make matters worse, most detectors are designed to work in moderately mineralized ground.  Where the ground strength is not excessive.   High mineralization will overdrive the front-end circuits of most detectors making them useless.   Raising the coil above the ground will eliminate front-end saturation.  However, as an operator you may never know just how high to raise the coil in order to avoid saturation.

Loop fold-over was a term I came up with to describe the non-linear characteristics of a loop.  Generally, as you lower the loop to the ground the output increase as a function of the mineral in the ground getting closer to the coil.  However, I have noticed that some coil configurations will reverse signal polarity if the ground is close enough to the loop.  This characteristic is loop design dependent.  In some ways its good since it tends to balance the mineral out.  Its very difficult to observe in an actual “real” ground condition since its simply a small change in the amplitude of the loop signal.  In most cases the loop output does not change polarity. This characteristic is most easily observed using a point source ground, a piece of ferrite.

I don’t want to make a big deal about this loop characteristic.  Its just a second order effect loops generally have.   The problem is that it can make a Automatic Ground Balance (AGB) circuit perform erratically if not designed properly.  As you know the  GoldTrax and CoinTrax both are Micro based designs.  I wrote special programs (call routines) in code for both modules that  reduce or eliminate any problem cause by loop fold-over.  Its not that complicated.  The programs simply readjust themselves to the threshold and balance out the mineral continually to that point,  in other words the threshold level.

Remember, the ground adjust control is just another form of a discrimination control.  Its used for discriminating out the ground.  In this case.....to balance to the ground.   I know you can have unusual effects by offsetting the ground balance, that is something that is best determine by experimentation and experience.

The choice of +0.5 degrees for the fixed ground control phase is somewhat arbitrary.  If the designer sets the calibrated fixed phase to 0 degrees he runs the risk that a ground phase near zero degrees will be picked up.  If this should happen the audio will come on due to the ground.  This would produce an undesirable situation for a preset machine since the operator will have no way of adjusting the control. Therefore, the preset must be made positive by some amount.  But how much positive?  The more positive the preset phase value the greater the sensitivity reduction.  The highest sensitivity would be obtained if we could set the control phase to match the ground.  However, since this is a preset machine that is not a option.  Years ago I found that a preset phase +0.5 degrees was the best compromise.

There is no question that a fixed detector design would be less sensitive than a design with a manual adjustment.  It is interesting to note that normally an air test will not reveal any difference between the two.  The reduction in sensitivity will only take place when you use the fixed ground balanced detector in mineralize ground.
The fixed ground adjust phase is generally calibrated using a typical coil. The phase difference between coils of the same size is usually small, 0.1 degrees or less.   Coils of different sizes would be more.  A well designed series of coils will keep this variation within acceptable limits.  Coil inductance, wire size and operating frequency have to be monitored closely to keep this from being a problem.  Overall, in a good design the phase difference between coils is not a problem.  You would probably experience more sensitivity change due to the coil’s different diameters than due to the phase difference between the coils.

The choice between having a fixed or manual ground control is difficult. Over the years I have experimented with many variations trying to arrive at a good solution.  In many cases the best solution is just to offer both on a single design.   The fixed control solution offers ease of use and reasonable detector sensitivity.  Then if the operator so chooses, manual adjustment produces the greatest sensitivity with some effort.

In 1984 I designed a Automatic Ground Balance (AGB) accessory  for the Teknetics Mark I.  Don’t be surprised if you never heard about this potential Mark I feature.  It was never release or advertised.  At that time there were not any microcontrollers available for that type of design. Therefore, the circuit was designed using discrete digital components. The complete circuit went on a PCB that was about 4 by 5 inches.  It was set up to mount on the back of the main Mark I board.   The AGB feature was integrated into the Mark I operation and did not require any additional switches.  It worked off of the toggle in the handle.  The operator could achieve a Turbo or quick Ground Balance by holding the toggle in one position.  Or you could let the AGB balance to the ground gradually.  A prototype was built and tested.  It actually worked quite well.  It had a very smooth characteristic.  The first time I tested it outside I felt it wasn’t even working correctly since I was not getting any ground reaction.   However, it was working fine.  I just didn’t expect that type of performance.  There was one problem with the design that  today I am aware of but I might not have known back then.  The ground balance setting would track off on targets.  This would have been a problem especially on large objects.   That problem could have been solved given more development.

This project was never completed.  It was dropped for several reasons. The add on circuit was fairly expensive.  Teknetics would have had to add over 100 dollars to the cost of the Mark even more if offered as a “Mod”. Also, it didn’t seem to fit into the Mark I thinking.  If you recall the Mark was designed for low mineral operation.  In low mineral, maintaining a true ground balance is less of a problem than in high mineral.  In reality this AGB feature would have been better suited on the Tek 9000 or 8500.  However, those detectors, designed in 1981 and could not interface correctly with this AGB circuit.

Several years ago I designed two modules with AGB for the Discovery Treasure Baron.  Both designs use microcontrollers.  The micro AGB design performs better than the discrete circuit for the Mark I.  Both have the iron inhibit feature.  This feature reduces the potential of the ground balance setting from tracking off on targets.   It works like this:  An internal program measures the phase of all targets.  Any target whose phase exceeds about -10 degrees  (I don’t recall the exact value) will produce an inhibit signal to the AGB program causing it to hold its current ground balance setting.  After the target has passed it releases the inhibit and the AGB continues to track the ground.  We called it iron inhibit but in reality it inhibits on all targets above a predetermine phase.  In this case -10 degrees.   All mineralize grounds have a phase less than -10 degrees. Therefore, the program will not inhibit tracking on mineralized ground. Of course if the ground’s phase exceeds -10 degrees this technique would not work.  The iron inhibit feature can be turned-off in case the operator runs into some unusual ground condition  where the inhibit feature does not operate correctly.  Or in case he or she simply prefers it disabled.

There is a difference  between the standard round and DD loops.  On the coils I designed for Discovery I don’t recall a problem with excessive phase shift between the two coil configurations.  When I first started designing coils for Discovery I decided to pick a particular frequency of operation and inductance for the Transmit and Receive coils.  In addition the “Q” of the Transmit must be control within a certain range.  For those who don’t know......the coil’s Q is the ratio of the coil’s inductance to its resistance at a given frequency.  Whenever the coils change in size the turns are modified to return the inductance to the standard value.  This tends to maintain winding resistance and more importantly, the Q of both the Transmit and Receive coils.  So, its important to have standard coil values to target the design to.  Normally this would be the coils, inductance, resistance and effective Q.  It these values are maintained the resultant coil phase will be maintained over all coil designs regardless of the coils size and shape.

The DD coils can get you if you are not careful.  As you know the Receive is generally the same size as the Transmit on these coils.  Coupling that with the tendency to keep the Receive turns constant can result in a serious change in the coils output phase.   Therefore, the Receive turns must be reduced considerably to lower the inductance back to the standard value.  From a practical standpoint the inductance does not have to be exactly equal to the target inductances.   As I said the tendency is to keep the turns the same as you change from one coil design to another.  This tends to keep the sensitivity the same across many designs.  However, that should not be the consideration.  In this case its more important to control the phase across many designs.  It’s better to look at it this way.  For example, suppose that we build two Receive coils where one coil has twice the diameter of the other.  But we keep the turns the same in both coils.  For this example the larger coil would have an inductance that was twice the smaller coil.  These coils would not have the same output phase.  The larger Receive would easily have more sensitivity than the smaller coil because of the greater turns and coil area.  However, this would not be a good design.  The turns on the larger coil must be cut by .707 times.  This would make both coils have the same inductance.   Ideally we would also need to change the Receive wire size to keep the Receive resistance constant.  Remember the coil Q is the ratio of its inductance to resistance at a given frequency. If we keep the inductance and resistance constant then the Q would also be constant.  However, I don’t generally change the wire size on the Receive because if you maintain the inductance constant the resistance tends to not change as well.  As I said, math calculations show that the wire size should be changed and to what size.  But from a practical standpoint the Receive wire size can be left the same.  When we reduce the Receive turns on the larger coil the coils characteristics approach the characteristics of the smaller coil.  However, the larger coil will still has more sensitivity than the smaller one because of its greater area.  The key here is not to get so concerned about the coil’s sensitivity that you forget about the overall design.

All that being said the DD coils do have the worst phase shift away from the target value.   However, it can be control within acceptable limits as outline above.  I don’t recall the exact phase tolerance on the Discovery DD coils but I think it’s below 0.5 degrees.   We always calibrate the fixed ground phase trimmer to be +0.5 degrees.   The phase of most soils do not go below -0.5 degrees.  Therefore, we have a total difference here of 1 degree.  This 1 degree differential is well above the 0.5 degree tolerance on the DD coil.  As a result we don’t have any serious phase problems with the DD coils.

One question that might seem important here is.......Why be concerned about the output phase on coils if you have a detector with a manual ground balance?  After all, any phase shift between coils can be compensated for with the manual ground balance.  Well that’s true.  But there is more to this consideration that must be understood.  You may have many detector designs some with fixed and some with a manual ground balance.   Also, I have produced many designs that had a fixed ground balance for the motion mode but a manual ground balance for the GB mode.  The bottom line is this.  You must decide on a standard and stick with it across all coils designs.   This keeps everything interchangeable.

Yes, it is important to have good quality caps for the Transmit tank capacitor.  The main reason to use polypropylene is because their capacitance is very stable over time.  Much better than most caps. Polystyrene caps are better but they don’t come in the larger capacitances like those needed for the tank Transmit coil.  If the capacitance is stable over time then the loop frequency is also stable over time.  That’s very important.  If the frequency changes the Receive signal phase changes for a whole bunch of reasons all related to the frequency change.   So it’s important to keep the frequency constant.  The other capacitor characteristic like it Dissipation Factor  (called DA) and leakage are not that important in this application.

You would see no difference in sensitivity between coils with different tank capacitors (polypropylene vs. polyester) if the capacitors had exactly the same capacitance.  However, you might see a slight improvement in drive efficiency.  The polypropylene cap would probably take less current to drive than the polyester cap.   The Transmit wire size has very little bearing upon the coils overall sensitivity.   However, it will greatly effect how much current(or power) is required to drive the Transmit coil.  The designer could make the Transmit wire size very small and reduce the weight of the coil.  That would be very impressive.  But you would not we impressed with the battery life.  The coil would draw huge currents and drain the batteries quickly.

One last point.  The internal fixed ground adjustment is calibrated using a dummy coil.  Not a real live coil at all.  So of like a dummy load on a ham transmitter.  The ones that consume the power but do not radiate into the air.

Litz is not use by most manufacturers because of the following reasons:  Litz is more expensive than standard solid wire and it is harder to work with.  Soldering it is more difficult and as far as I know it’s not available with self-supporting coatings.   Also, it is probably difficult to quantify the improvement in using Litz.  I have been using Litz wire in Transmit coils since 1989 but not for the reasons mentioned by Minelab.  The applications where I have used Litz wire were only in industrial metal detectors where there is no ground considerations.

There are several effects to consider when discussing the ability for a detector to reject the ground mineral signal.  The first is frequency.  Some of the original mine detectors operated at 1000 Hz.  At that frequency the reflective phenomena is almost nonexistent.  At least it is not a design consideration..  At higher operating frequencies the reflective ground effect can be broken up into two main effects, static and dynamic.
Static effects refer to the fact that the ground is not balanced out when the loop is at various distances from the ground.  This is the effect you mentioned.  Using Litz wire for the coils in the loop will reduce the static effect problem.

The second or dynamic effect has not been address by anyone as far as I know.  This phenomenon is due to the motion of the coil across mineralized ground and prohibits you from obtaining a true balance.  It has nothing to do with the reflective ground effect but it appears to be related to it.  But it’s not.  I have been interested in this effect for many years because it directly effects the performance of Motion detectors.  A special circuit design can eliminate the problem.  Some of the detectors that used this circuit were the Teknetics 9000, 8500 and Mark I.  To some degree it was incorporated into the Discovery Treasure Baron.

Although not fully exploited.  You can not tell what this particular component design arrangement is, just by looking at the schematic of the detector.

Ordinary hook-up wire is not the same as Litz.  It very important that the individual Litz wires be insulated.  This distributes the total current evenly in all the wires.  If they are not insulated then you might as well as not have individual strands.  Also, the individual strands are wound in a very particular fashion that minimizes the skin effect in each strand.  The end effect is that the resistance of a properly designed Litz is (almost) completely flat from DC to RF frequencies.

I have built and tested Litz wire loops for consumer detectors.  However, they never went into production.  The improvements gain by reducing the dynamic effect mentioned above and the use of AGB circuits were enough to satisfy our design requirements.  Also, the use of a preset ground balance makes the static effect phenomena almost irrelevant. This is not to say that the elimination of static effects are unimportant.

Here are some general design parameters for a coil:

Transmit Coil -- 25 to 30 turns of 22 gauge wire.  Diameter 7 to 8 inches.
Receive Coil -- 200 to 300 turns of 31 gauge wire.  Diameter 3 to 4 inches
Feedback Coil -- 6 to 10 turns of 22 gauge wire.  Diameter same as receive.
Tank Capacitor -- 0.47uF

Generally the Receive is about half the diameter of the Transmit.  So if you choose an 8 inch Transmit use a 4 inch Receive.  Wind the Feedback Coil on top of the Receive coil.  The wires are insulated so it ok to have the Receive and Feedback touching.  The Transmit and Feedback coils can have the same wire gauge.  Here is a very very important point!  The end of the Receive wire nearest the Feedback Coil must be connected to ground.  In other word it must be connect to the loop shield and to the ground in the circuit.   If you don’t do this the completed coil will not operate correctly.  The R null component of the coil will be excessive and may overdrive the detector.   This has to do with the high capacitive coupling between the Receive and Feedback windings.  Connecting the coils as I have outline above will solve the problem.

The Transmit and Feedback coils must be connected series opposing.  If they are connected incorrectly you will not be able to obtain a null signal from the Receive coil.  This is generally not a problem since it only works one way and not the other.   Basically what is required here is that the magnetic field produced by the Larger Transmit coil must be in the opposite direction to that of the Feedback coil.  This is the key to obtaining a magnetic null for the Receive coil.

One final important point.   The ratio of the Transmit turns to the Feedback turns is about 3.3 times.  This holds only for this size coil, 7 to 8 inch diameter.  In other words to determine the Feedback turns divide the Transmit turns by 3.3.

About the connection of the Receive in relation to the Feedback wind.  It’s seems to be a very minor thing and not important. The only reason this is a problem is due to the closeness (touching) of the windings.  If they were an inch or two apart it would not be a problem.  There is one other little point that I forgot to mention.  Again this is very important.  The Feedback winding must be connected to the un-driven side of the Transmit tank circuit.  For the same reasons I mentioned before this reduces the capacitive coupling between the Transmit and Receive which will allow the resultant Receive null to be small.

The dynamic effect I refer to is a small second or third order effect.  Most designers are not aware of this effect since it is so small.  Other than the detectors I mentioned no one that I know of have ever used this circuit.
This dynamic effect has very little to do with the earth’s magnetic field.  It has to do with the interaction of the metal detector’s magnetic field and the magnetic material in the ground when the loop is moved in relation to the ground.   Unless it’s eliminated it is impossible to obtain a true balance if the coil is in motion.  As you can see this is an important concept for the motion mode.  It helps for the GB mode too.  But, it is not as apparent in this mode since the loop does not have to be in motion all the time.

 Sometimes there are very simple answers why some manufacturers use paint and others use foil or paper for the loop faraday shield.  Usually it is related to ease of construction or cost, not necessarily to performance.  Each technique has its advantages and disadvantages.  And, you may only determine what those factors are by trial and error.  My background has been in using paint shields.  This process can be a little tricky if you don’t know what to avoid.  For example.  It’s best to have a very smooth surface on which to place the paint.  Painting a shield on an irregular surface can cause excessive noise in the detector.  If fact early Discovery loops had only the bottom cover painted in order to avoid too many surfaces that were not smooth.  However, after some problems were worked out all their loops were 100%shielded.  Making an electrical connection to a paint shield is difficult too.  When I was with Teknetics we experimented with many correction methods before settling on one process that produced consistent results.  A poor connection was prone to breaking loose or producing very high detector noise.

Generally its best to have some distance between the shield and the coils.  However, I have seen many designs were the shield is place directly on the wire.  A foil shield for example.  The conductivity of the shield must be low enough to not interfere with the detector operation.  My personal experience has suggested that resistances around 10K ohms per square or optimum. However, the resistance can vary quite a bit without effecting detector operation.   When this value drops below 1k then you can have pickup problems.