This feature is a general descriptive on the subject of advanced metal detector ideas and applications. It covers many areas of metal detector techniques not previously seen described in the popular electronic press and is intended for general discussion - as it contains hypotheses to explain certain features of the operation of the circuit that are not readily authenticated by quantitative analysis.
Persons wishing to enter into correspondence on the contents of this feature are invited to do so, addressing their letters to : - omlcgm@ aol.com or qrachma@ki.ericsson.se
I would ask requests for information be tempered with the consideration that I am not going to be able to deal with a vast amount of correspondence concerning basic theory/electronics, and that I am inviting a technical discussion concerning operational theory, an it's quantification.
Throughout this feature it will be assumed that readers will be reasonably familiar with the operation of basic types of detector , and the theory of their operation. (Detecknowledgy is aimed at this type of preparation). Those of you familiar with metal locators will probably agree that the most prominent problems associated with their use are common to all variations in popular use:
At this point, it should be noted that most commercial manufacturers guard their particular secrets very closely, some going to very great lengths to conceal their actual circuit configurations. This is largely because the actual component cost of the unit is negligible, with a substantial part of the cost being made up from research and development. This would obviously entice a "pirating" of the basic ideas if readily accessible. So little seems to exist in the way of technical theory and background to circuits in use. Ambit has arrived at these conclusions from a process of experimental observation of the practical effects of certain types of circuits, based on the theoretical background of radio engineering. We do not claim total originality for our approaches, since from the absence of any other source of information, it is impossible to be categorical that we have a unique approach. However, to the best of our knowledge, what follows is an original and innovative approach to a low cost discriminating metal locator, that largely eliminates the objections listed above.
Absolute sensitivity is only of any real use when it can be achieved without a sacrifice of these basic criteria. The different types of operator audio/visual feedback can be chosen for the most compatible compromise between the electronic complexity, and physiological effectiveness - and the most easily recognised audio feature is that of the variable pitch tone. A simple "tone - no tone" audio switch has certain limitations due to the logarithmic response of the human ear - especially when attempting to pinpoint a source, and the variable pitch tone also permits a large degree of tuning error due to the ability of the ear to track the tone, and still recognise variations as objects are approached. So we will note this feature for incorporation in the design, since it may readily be controlled by the voltage derived from the detection circuits.
Since sensitivity is basically limited by the circuit drift, (assuming a totally rigid mechanical construction), the answer is to try to reduce the operation of the circuit into a system where "analogue" signal processing and detection is not critical. The example of the calculator points the way, since the basic calculator (digital) has eliminated the analogue interpretation errors that occur when using the slide rule. In the same way, the digital processing of the metal locator signal provides a means of providing a detector that should be entirely free from thermal and voltage fluctuation drift. However, the usual induction balance system operates on the measurement of the amplitude changes in the signal received and detected by the coil arrangement in the search head - and however you look at it, that is a thoroughly "analogue" function. The question of analogue to digital conversion at the signal levels present in the IB system is entirely academic, since before that could take place, much analogue processing would be necessary to arrange the levels. There are many long winded and tedious approaches to providing stability in such analogue systems as the IB metal locator, but they are complex to design, critical of components, and tiresome to set up.
However, the phase angle principle of detection is one that is entirely amplitude independent in terms of signal processing. The phase change can be measured in digital terms, with the "on" period corresponding to the number of degrees of change - and since the reference source for the comparison is essentially "locked" onto the received signal, frequency drift is generally insignificant, within certain limits. (Otherwise the phase relationships of the search coils may alter as the circuit tunes around it's resonant point).
The next prime consideration is the frequency of operation. In the UK, the frequency allocations are basically DC to 150kHz, with certain spot frequencies, such as the standard frequency transmissions of MSF, to be avoided. In fact, the power levels of the metal detector are such that there is really likelihood whatsoever of any interference - but it is understandable that the radio licensing authorities would want to keep tabs on developments in the market. In the basic IB configuration, where amplitude considerations are foremost, for reasons of coil design as much as anything, frequencies chosen are in the region of 100 to 150kHz. However, in this instrument we want to be able to optimise as many feature as possible - and operation at high frequencies does not permit best ground immunity, nor does it permit the type of discrimination technique used in this design, where both junk and sought after objects can be simultaneously indicated without resetting controls. The operational frequency is 18kHz (approximately) - in the VLF region for radio waves, and thus the unit exhibits good ground exclusion and depth penetration. One of the two factors in this choice is the skin effect of an AC current travelling in a conductor - whilst basically a transmission formula - it is applicable to an appreciation of the "full" discriminator technique - as opposed to the partial effect exhibited in the different reactions of a metal of high permeability (iron) to one of relatively low permeability (silver) when detected at higher frequencies.
The skin effect concerns the phenomenon that occurs in a conductor, whereby the depth at which the current appears to flow is proportional to the frequency, permeability (absolute) and the conductivity according to the formula:
d = 1 / frequency . p . m . s
Where s is the conductivity in MKS units, and m is the absolute permeability in MKS units.
Considering a metal considered to be "desirable" - such as copper - we obtain the following skin depth values at 18kHz:
Where m = 4 . p . 10-7 and s = 5.8 x 107 mho/m
= 6.62 / 18000
= 0.04934
= 0.05 cm
By increasing f to 100,000Hz, the skin depth is reduced to .02 c m.
(In fact, current density decreases exponentially away from the surface, though in transmission line work, the above formula is considered quite sufficient to analyse results).
The detection of non-ferrous metals largely revolves around the principal that the search coil of the detector induces a current in the metal object, such that the eddy current so created tends to oppose the field producing it. In a material such as ferrite, the field is concentrated and enhanced, thus producing an opposite effect. The magnetic effect of the eddy current in the non-ferrous (low permeability) metal will depend on the total number of electrons buzzing around - and can thus be seen to be proportional to the resistivity (conductivity) of the material together with the skin depth.
At VLF, the current is induced further into the metal, and is thus more likely to be modified by the permeability of the medium - such that distinction between, ferrous and non-ferrous objects is more easily established than with the surface eddy currents of higher frequencies. Ferrous metals treated for the minimising of eddy current effects - such as transformer cores - will always tend to show a ferrous reaction, even at relatively high frequencies. Beyond about 50kHz, ferrites of one shape and form are almost exclusively employed as the magnetic material since the particle nature of their composition rules out eddy current interference.
The actual theoretical basis of the operation of the adjustable discriminator - where discrimination takes place to exclude responses from undesired junk - is something that is difficult to quantify. The principal is frequently employed but the author has yet to find anyone prepared to describe the theoretical basis for the phenomenon. It is not very clear if this is simply professional secretism, or simple mystification. In any case, the system works- but how?
The waveforms and spectral analysis of the instrument show quite plainly that with the coils correctly coupled for a signal transfer null, a solid non-ferrous object in the area of coupling between the coils will bring about a small amplitude change - coupled with a distinct phase shift. High resistivity materials - such as foil etc., cause an amplitude change, but no appreciable phase shift. Indicating that the junk merely reflects or absorbs the fields without having sufficient eddy currents induced therein to modify the magnetic characteristics of the search head.
Modification of the discriminating effects of the system may then be deduced to be brought about by a system that introduces a small portion of the transmitted waveform, directly into the receiver circuits, bypassing the head, so directly modifying the phase, relationship ' by pulling' or forcing the phase. So that when the signal from the head is received at increased amplitude by virtue of the presence of a junk object, the forced signal is itself overridden, causing an apparent phase relationship difference from tine steady state signal.
Considering once again the apparent change in the reaction that occurs when sampling at the higher frequency, it seems feasible that this approach also offers a practical clue to the production of a metal locator system that actually allows the determination between different metals. The comparison of reactions at say 18kHz and 100kHz would provide data that is independent of the mass of the object concerned - since the result is related by the permeability and conductivity term. There is obviously much work required before a practical system can emerge, but the advance of MPU technology indicates that the digital analysis of results is not an unlikely, nor an improbable extrapolation.
Thus we have a sound basis for the development of a practical circuit to exploit these techniques. Referring back to the three basic criteria - from the operator's point of view - we have achieved a system that ignores junk in its phase reaction, but may still be used to acknowledge the presence of junk in its amplitude reaction. In other words, all types of objects may be noted simultaneously, without the need to reset controls. Cross interpretation of the two meter readings leads to a far better understanding of the nature of the object concerned.
The major tuning function - for the desirable object processing - has been moved into a digital analytical system, which will be far more versatile and controllable, permitting an effective auto tune function.
However, all is not entirely infallible, since the orientation of the object in the plane of the detector heads will still
Lead to some anomalous results, and the areas not directly under the overlap will also have influence on the readings where larger objects are concerned. Much as with the problems covered in "Detecknowledgy" on simple IB and TR approaches.
Various theoretical designs have been proposed to try to minimise the effects, and work is presently in hand to try to evaluate and develop a coil configuration better suited to ignoring the spurious side reactions. Any form of electrostatic and electromagnetic shielding of the search head leads to such gross distortions that operation under normal conditions is severely compromised. The 2D head design is generally considered to be slightly less sensitive than the 4B coil head. But neither offer a solution to the anomalies brought about in the presence of large objects. Ideas and suggestions are welcome. Nevertheless, this is not a problem of any greater magnitude than exists with any other detector - and is one that can be minimised through operator familiarity. The basic block diagram thus looks like:
And although that may seem daunting, a practical example has been made with just four CD4011B gate ICs, and a
ULN2233B audio stage.
Alignment is simplicity itself – almost – and comprises switching on with a coil overlap so that a reading is present on the field meter. The oscillator fine tune is set to maximise this field reading, and the heads are adjusted so that the meter reading is nulled as far as possible. Then the oscillator tuning is set to re-peak the field reading once again, and so back and forth. Meantime, the phase meter reading will have been jumping about wildly – don’t worry; the idea is firstly to set the oscillator frequency to the resonant frequency of the receiver L/C tuned circuit head. When no more improvement in level is achieved by tuning the oscillator, set the null carefully, and try to null out the reading on the phase meter. Just as the two coils are pulled slightly apart, the phase meter will show a sharp reaction which is indicative of the operational point of the system. Checking the head responses to different materials will confirm the optimum head alignment. Be prepared to spend some time acquainting yourself with the operational characteristics of the alignment and tuning settings.
Please also note here that I shall not at present be answering correspondence concerning the basic features of the circuit. I.e. if you make the circuit shown, and it fails to operate, then I must ask you to use your own time to learn and become familiar with the system. The circuit as shown here is accurate and has been built and tested, providing results of 2p pieces at 8 – 10 inches, and full anti junk discrimination. Adjustable discrimination is possible by feeding directly from the TX coil to the RX coil via a 2M potentiometer in rheostat configuration. (Carbon types only).
The circuit shown does not include
CMOS standard practices - such as decoupling the supply pins of the ICs with
100nF and taking any floating input to logic 1, these must be taken into
account in your design. The FET audio gate is not an ideal medium - a CD4016
is probably more suitable, but constraints of time and resources have had
to curtail additional development of the basic circuits. The gate reference
voltage should be set on the 25k preset in the source circuit - and this
is the only truly analogue point in the circuit (of the phase detection half
of the circuit) and so is one which the author plans to remove at some stage.
The output of the phase detector is simply a pulse train. The width of the pulses is proportional to the phase error, (data on the CD4046 CMOS PLL is a useful guide to the principles involved), and so the integration of this pulse train provides a DC voltage which may be used to control a variety of "alert" functions. Including the authors' favourite, the swept audio tone. (Voltage controlled audio oscillator).
Why stop at this point you are probably wondering - after all, the system described thus far apparently equals the performance of units costing a great deal more than the £5 odd for this lot! But that is the basic problem, it is not our policy to conceal £5 worth of bits in lumps of epoxy and charge £50 for the idea development. However, this does in no way mean that the ideas so far represent the zenith of achievement, and this section is aimed at supplying the ideas necessary to refine the basic principle. At this point in time, Ambit has no commercial intentions in marketing a complete and "foolproof" constructional kit for some time. Those of you with detectors and nerve may be prepared to use the basic frames to insert your own electronics - but please do not ask ME to do your trouble shooting!
The first development feature to consider is the coil construction. In the basic circuit, the coil exists with relatively low impedance, and hence phase change rates will be lower than a coil arrangement consisting of higher Q and better matching. (See the Ambit 78-1 catalogue theory section.) The coils chosen were those derived from our VCO version metal locator, and were used more for convenience than anything!
A reasonably accurate formula for the calculation of search coil impedance’s, based on diameter, is
L = N2.R2 / 9.R + 10.N.D
N = number of turns
R = radius in inches
D = spacing in inches
L = inductance in m H
This is basically a solenoid formula, but quite sufficiently accurate for most purposes. Further formulae in the catalogue mentioned above.
The head is always the key feature in the detector system, and must be carefully constructed and glued in place. The fine tuning of the coils once the glue has set can be brought about by leaving a few inches of the overlap free, and adjusting this carefully when the main configuration is rigid.
A remote "fine tune" of the head may be brought about by a third tuning coil. Then the reference oscillator can be a crystal controlled oscillator, divided in a CMOS counter to provide .a number of binary related frequencies for the approach of comparing the results obtained at two different search frequencies. The coil tuning Cs would need to be switched (such as by a PIN or similar switching diode).
The AF gate has already been mentioned, since the basic circuit gate provides a basic form of voltage controlled attenuator. A second approach using a Darlington pair biased from rectified AF combined with the control voltage has been more sharply defined, but is more prone to drifting.
The autotune system is basically a circuit that compares the length of the phase error pulse from the phase detector with a reference pulse from a monostable that is triggered from the out put of the master reference oscillator. The "error voltage" is fed back to the reference source in the manner of AFC, since phase is also a function of the resonance of the tuned circuit, adjusting the frequency will compensate phase. The time constant of this function should be long enough so that object signals are not masked by an over reactive autotune.
Well, once again credit is due to Ambit's engineers. Ambit is no longer trading - as far as I know.
I have tried most of the ideas proposed and this has led to a better understanding of the operation of my detectors. Yes, I own three commercially made detectors. I have spent many hours experimenting, but never found the time to finalise my designs. Maybe you will. Let me know how you get on.
Cris G Martin.
Go to Detecknowledgy
Link to UK Detector Net
You are visitor number:
since 19 April 98.