METAL DETECTORS

Links to more pages on Metal Detectors:

http://www.next.gr/sens-detectors/metal-detector-circuits/

The Fundamentals of Electr onic Prospecting - a discussion on different types and their capabilities
http://www.goldgold.com/the-fundamentals-of-electronic-prospecting.html

Geotech: Lots of circuits of well-known metal (gold) detectors:

http://www.geotech1.com/cgi-bin/pages/common/index.pl?page=metdet&file=schematics.dat

Rem em ber this: alm ost all circuits (up to about 4 - 6 transistors) have the same capability: detecting a
20mm coin at about 100mm . It doesn't matter if the circuit is simple or com plex, The sensitivity r evolves
around the circuit driving the coil. Some circuits are m ore sensitive to "interference" and we have
dem onstr ated this in Circuits 1 - 8 at the end of the page. The m ost-sensitive circuit is a 100mm dia coil
with just 12 to 20 turns and operates at about 200kHz as shown in circuit 8. The frequency of the circuit
will change by one Hertz and this can be detected on an AM radio. You cannot get better than this.
The simplest circuit is shown below : Another Simple Metal Detector Circuit

To learn more about the basics of circuit-design and recognise com ponents, component-values, go to our :
Basics Electronics 1A

We also have two Metal Detector Projects:

Metal Detector MkI

Metal Detector MkII

BASIC CIRCUITRY of
METAL DETECTION
by Charles D. Rakes

Note by Colin Mitchell: The first part of this dis cussion is a very old article using US imperial measurem ents, by C harles D. Rakes . A
table of wire gauges is provided at the end of the article. The rest of the circuits are from different sources.
All these circuits have about the same sensitivity as the single transis tor circuit shown in fig 7 of Part II (shown below), using an AM radio as
the receiver. They have been included to show the ingenuity of design-engineers, in an attempt to improve the performance.
Here is a reference from another website with exactly the same views as myself:

The Beat-frequency oscillator (BFO) is the simplest (and oldes t) ty pe of metal detec tor technology and is a good starting point for learning
how metal detec tors work. The basic beat-frequency metal detector employ s two radio frequency oscillators which are tuned near the same
frequency. T he firs t is called the sear ch oscillator and the other is c alled the refer en ce oscillato r.

The outputs of the two oscillators are fed into a mixer whic h produc es a signal that contains the sum and difference frequency c omponents.
This s ignal is feed to a low-pass filter removing the harmonics. As long as the two oscillators are tuned to the same frequenc y, the output
will have no signal.

When a metallic object disturbs the magnetic field of the search c oil, the frequenc y of the search osc illator shifts s lightly and the detec tor will
produce a signal in the audio frequency range.

Although once popular, BFO's are no longer being made by professional metal detector manufacturers. They are simple and inex pensive, but
do not offer the accuracy and control of modern PI or VLF detectors. Attempts have been made to add new features such as disc rimination
and more advanced models were produced in the 1970s, but they were soon replac ed by recent, more sophisticated technology.

BFO designs are still used in cheap hand-held devices and in low quality, toy ty pe detectors.

The Sim plest Metal Detecto r Circuit is als o shown below and it only requires 4 components.
Using a Faraday Shield around the search coil will reduce the effect of the ground altering the frequency if the ground has a large amount of
iron in the rock s. Simply wind aluminium foil around the turns of the s earch coil and leave a small gap where the wires exit.
It is pointless going to a lot of work building a complex receiv er (as shown in a number of the circuits below ) as the result will be no better
than the simples t c irc uit.
All these circuits are limited to picking up a coin at 90mm to 150mm. Bas ically, a 90mm c oil w ith pic k-up to 90mm and 150mm coil will pic k-
up to 150mm.
An AM radio will detect the change in frequency of a few cyc les at 150Hz and y ou c annot get better than that.
To get a deeper penetration, you need to deliver very high energy to the c oil to produc e magnetic flux that enters the ground and gets stored
in the gold nugget.
The coil is then turned off and the circuit listens for the collapsing energy from the gold nugget being released and detected by the c oil.
This is called Pulse Injection tec hnique and will be covered in later circuits.
For now, here are some simple circ uits:

Metal Detection Basics

One of mans greatest challenges throughout history is to s ee w hat cannot be seen, to detect what is hidden, and to reap ric hes from these
treasures. This v isit were going to look at some very bas ic metal-detection circuits . Now don't get me wrong; the circuits we'll share here
most likely will nev er locate a v aluable treasure, but they c an be put to us e performing other more practic al applic ations. Howev er, in the
early days of the last century, even the simplest of metal detectors were succ ess ful in dis covering some very v aluable buried treasures.
Simplicity often is the bes t route to tak e in solv ing a seemingly difficult task . Never give up on an electronic adventure because you don't
have the latest and greatest equipment.

Ferr ous Ferrets

Our first example of a ferrous detec tor is a s im ple mechanical device shown in Fig. 1.

The detector is a m odified balanc ed s cale, which indicates ferrous objects and magnetized items. A magnet is attached to one end of the arm
and a simple north/south sc ale is attached at the opposite end. A pivot is located near the magnet end of the arm and a slide balancing
weight is on the opposite end.
The magnetic scale should be balanc ed w ith no ferrous items near by. Any non-magnetized ferrous object positioned below and c lose to the
magnet will caus e the pointer to go up due to the magnetic attrac tion.
The magnetic scale should be balanc ed w ith no ferrous items near by. Any non-magnetized ferrous object positioned below and c lose to the
magnet will caus e the pointer to go up due to the magnetic attrac tion.
A magnetized object with the south pole facing up will caus e the pointer to go dow n, and when the north pole faces up the point er will ris e.
This ultra-simple m agnetic detector is v ery sens itive and w ill easily determine w hat objects are ferrous and the polarity of magnets.

Electr onic F er rous Ferr et

Our first electronic metal detector circuit, see Fig. 2, us es a Hall Effect sensor to detect weak permanent magnetic fields. Almost all ferrous
objects retain some degree of magnetism, and those that do are easily detected with our Hall Effect ferrous-detector circuit.
Fig. 2. The electronic v ersion of the F errous Ferret uses a simple Hall Effect IC.
Weak magnetic fields can be detected with this eas y-to-build devic e.

The HAL 115UA-C IC Hall Effect sens or is the heart of the weak-field detector circ uit and is available for les s than a buc k from D igi-Key. T his
Hall Effect sens or is a bipolar devic e that is s ensitiv e to a magnet's north pole on its branded s ide and to the s outh pole on the oppos ite side.
The branded s ide, see Fig. 3, is the side that display s the part number.

Fig. 3. Let's get up close and pers onal with our friend— the HAL I15UA-C.
The branded s ide-where the part number is displayed— is sensitive to a magnet's north pole,
while the opposite side is s ensitiv e to a magnet's south pole.

The sensor's output (pin 3) is norm ally low when no external magnetic field is pres ent. Plac ing a magnet with its north pole facing the
branded side of the sensor will caus e the output at pin 3 to go high. Placing a magnet with its s outh pole facing the non-branded side will als o
cause the output to go high.

Fig. 4. Utilizing some skill and patience, induc tors can be hand-w ound.
Here is a s imple diagram showing the typical inductor needed for metal-detector c ircuits .

Here's how the circuit operates. Two gates of a 4093 quad, 2-input, NAND Schmitt trigger IC are connected in a low-frequency square-wav e
oscillator circuit operating at about 100 Hz . The output of gate "C" drives the base of Q1, whic h is connected in an emitter-follower c ircuit
supply ing the 100-Hz signal to L1. Induc tor L1's drive level is set by R6. The output (pin 3) of IC 2 is connected to an LED and a metering
circuit.
Inductor L1 supplies a low-frequenc y AC bias to the backs ide of the Hall Effect sensor, IC2. This AC bias in effect increases the Hall Effec t
sensitivity many times ov er and also allows it to detect both north and south pole magnets from the branded side; however, the circuit is
much more sens itive to north pole fields. The arrangement of L1 and the Hall Effect s ensor is shown in Fig. 5.
Fig. 5. The Hall Effect IC works in conjunction with the inductor.
A low-frequenc y AC bias is s upplied to the back side of the IC via the inductor.

The Hall Effects output waveforms are shown in F ig. 6. The waveforms are obs erved at pin 3 of the Hall Effec t IC. Output waveform "A" is set
by adjus ting R6 for a sym metric al output without any ferrous metals in the pick-up area. If a scope is not handy, a DC voltmeter can be us ed
to set the output to about 4.5 volts . This setting will produce an output waveform v ery close to the one shown in F ig. 6A. The "B" output
waveform occurs when the north pole of a magnet is brought in proximity of the Hall Effect sens or. T he s outh pole of a magnet produc es the
output waveform shown in Fig. 6C.

Fig. 6. Here are the waveforms that might come from pin 3 of IC2.
Resistor R 6 can be adjus ted to calibrate the c irc uit.

Win ding L1
Inductor L1 (see Fig. 4) is fabricated by jumble winding 500 turns of #32 enamel-c ov ered copper wire on a ¼-inch diameter ferrite rod. T he
rod's actual diameter and length are not critic al, and any s iz e rod material from ¼inch to ½inch in diameter will do. T he rod's length can be
anything from 1 inc h to 3 inches . The ty pe of rod material suitable for this application can be salv aged from an old AM transistor radio or
some older TVs.
If the rod material c annot he loc ated, don't give up 'c ause there are other paths to take. A relay coil with a resistance of 10 ohms or greater
will generally work for L1. Some miniature audio transformers have s traight sec tions of laminations that can be used in plac e of the rod
material. Most of the rod material I've used and have recommended here is actually designed for much higher frequency use. As a last ditch
effort, try a number of small nails taped together as a core for L1 and s ee w hat happens. Here's a great place to experiment with v arious coil
core materials and windings to improve or vary the c irc uit performanc e. Keep me informed on your efforts.

Try Th is On e
Som ething els e came to mind after disassembling the circuit, and due to time restraints I was never able to check it out. I would lik e to
challenge you to do so. What if a second Hall Effect IC s ensor was added to the circuit but placed beside IC2 with its branded side facing
L1's core?

Fig. 5. The Hall Effect IC w orks in conjunction with the induc tor. A low-frequency AC bias is supplied to the backside of the IC v ia the
inductor.
Duplic ate IC2's circuitry with the new IC, but leav e out the metering circuit. See Fig. 7 for details. Try to get like waveforms from both circuits
by adjus ting R6 and positioning the tw o ICs on the end of L1. C onnec t one lead of a digital DC v oltmeter to pin 3 of IC 2 and the other meter
lead to pin 3 of the added IC. If I'm c orrect, the circuit should be as sens itive to the south pole of a magnet as the original circuit was to the
north pole. If not, try connecting a DC voltmeter to the output of IC2 and another v oltmeter to the output of the added IC. IC2 should remain
more sens itive to the south pole of a magnet, and the added IC s hould be more sensitive to the north pole.

All Metals Detector Circuit

Our next metal detec tor circuit tak es us back into the early years of the last century where tubes were king and sem ic onduc tors were only
diodes. It was discovered early on that any metal object plac ed near the tank circuit of an os cillator w ould shift its frequency either up or
down. A tank c irc uit is the c ombination of an induc tor and c apacitor that make up a tuned circuit. Ferrous metals near the inductor of a tuned
circuit cause the os cillator's frequency to go down and non-ferrous metals cause the frequenc y to increase. This is the bas ic effect that the
Beat Frequenc y Oscillator (BFO) ty pe of metal detec tor us es to detect all metals. Figure 8 s hows a block diagram of the c ircuits making up a
typic al BFO detector. A searc h loop is usually wound in a c ircular fashion to serve as the induc tor in the search osc illator's tank c irc uit. The
reference oscillator is v ery similar to the searc h os cillator with a much smaller induc tor, whic h is usually shielded from the searc h loop. RF
signals are tak en from both osc illators and fed to a c ommon mixer, where the sum and differenc e frequenc ies of the two os cillat ors are
mixed. T he s um frequenc ies are filtered out, leaving only the audible difference frequencies to pas s on to the amplifier and headphones .

Fig. 7. Increas e your odds at detection with this simple modific ation. An additional Hall Effect IC is added to balance the circuit's sensitiv ity to
the north and s outh magnetic poles.

Fig. 7. Increas e your odds at detection with this simple modific ation.
An additional Hall Effect IC is added to balanc e the circuit's sensitivity to the north and s outh magnetic poles.

As a practical example, we'll s et the search oscillator up to operate at a frequency of 100,100 Hz, and the referenc e os cillator to a frequenc y
of 100,000 H z. T he differenc e frequenc y between the two oscillators is an audible 100 Hz that is fed to the headphones. The search c oil is
then moved ov er a small ferrous metal object causing the os cillator to drop in frequency to about 100,050 Hz . T he audible 100 Hz tone drops
to 50 Hz indicating a metal object is located somewhere near the s earch loop. A non-ferrous objec t near the loop will cause the oscillator to
increase in frequency and produce a higher audio output tone. A carefully adjusted BFO metal detec tor can be us ed to dis criminate between
ferrous and non-ferrous metals.
Fig. 8. The popular loop-detector circuit has been a mainstay for many treasure hunters .
A set of headphones allows the user to hear an indic ation of ferrous material and magnetic fields .

Two-Transistor B FO Detecto r
One of the simples t BFO metal locators to build is the two-transistor circuit shown in F ig. 9.

Fig. 9. This is the sch em atic for a B eat Fr equ ency Oscillator metal-d etector.
Two transistor s are used as the oscillators in this circuit.

The circuit m ay be set up to operate on any frequency between 50,000 Hz to over 1 MHz by selecting the tank circuit components . J ust
about any good general-purpos e NPN trans istor suitable for low RF applications will w ork just fine. T he s earch loop c an be as s mall as a
dime or three feet or larger in diameter. A small loop works best for small objects buried shallow and a large loop works bes t for large objects
buried at greater depths. The two oscillator circuits s hould be s eparated and shielded from eac h other to reduce frequenc y pulling between
the two. A really well constructed BFO detector will be able to operate with a difference of less than 100 Hz between the two oscillators. T he
lower the audio output tone the easier it is for the ear to tell a small frequency shift. The detector's maximum sensitivity is obtained when the
two os cillators are operating jus t a few cycles apart. Believ e me, this is not an eas y task to acc omplish, but one well worth the effort.
Here's how the simple BFO detector operates. Transistor Q1 along with its associated components make up a Colpitts os cillator c ircuit with
the search loop, C1 and C3 forming its tuned circuit. Transistor Q2 with its ass oc iated components mak e up another C olpitts oscillator circuit
with L2, C2, and C4. forming the tuned circuit. The emitters of Ql and Q2 are coupled together through R1, R2, and the low-im pedance
headphones . T his circuit arrangement functions as a simple RF mixer c ircuit. The audio frequencies are fed to the headphones, and the RF
frequencies are by pass ed to ground through C 8.

Win ding And Scroun ging

The loop may be w ound on almost any round insulated non-metallic form, s uc h as wood or plastic. Induc tor L2 can be an old ferrite rod
antenna coil remov ed from an AM transistor radio, or one c an be made by winding a coil on a round insulated form. Let me offer the follow ing
winding suggestion to get you going on building the BFO circuit. Locate a 10- to 12-inc h wood or plastic hoop that's about 3/4-inc hes wide
and clos e wind ten turns of #18 to #22 enam el-covered c opper wire evenly around the forms. Tape over the wire with plastic electric al tape
and connect to the circuit w ith a length of two-wire zip cord. If an antenna coil cannot be found for L2, then clos e wind about 80 turns of #22
enamel-covered copper wire on a 1-inch plastic pill bottle or plas tic pipe.
One important thing to do in selec ting the two inductors is to be s ure that the reference oscillator can tune to the same frequency as the
search oscillator. If a frequency counter is available then the chore will be super easy. If not, some experimenting with different pairs of
capac itors (C1 and C 3 or C 2 and C4) will be nec ess ary to bring both oscillators to the sam e frequency.

Part II
Single Transistor Circuit
Before getting into the circuitry, w e had better take a quick look at how the single-transistor detector system operates . I'm sure that at s ome
time y ou've heard a whistle or tone while tuning your AM broadc ast receiv er or, even more lik ely, w hen listening to an AM short-wave
broadcast station. In radio circles, this is referred to as a heterodyne signal. An AM receiver detecting two RF s ignals , w hich are very close in
frequency, usually causes this condition. If the two RF frequencies are less than a few kHz apart, an audio tone (difference frequency ) will be
heard. T his is basically how our s ingle-transistor detector c ircuit operates .
In our single-transistor circuit, see Fig. 1, only one RF os cillator c ircuit is used. The other RF s ignal is supplied by one of many AM broadcast
radio s tations. A portable transistor AM radio receives the two RF s ignals and outputs an audible tone. The mixing and audio amplification is
handled by the transistor radio. If either RF signal shifts in frequency, the audio tone will increas e or decreas e by the same amount. Since the
frequency stability of all licensed AM broadc as t stations is rock s olid, only our searc h os cillator will produce a s hift in frequency. T he end
result is a detector that operates like our two-trans istor circuit, but requires les s parts and time to c onstruct.
The oscillator circuit in Fig. 1 is very s imilar to the os cillators used in our previous circuit. Transistor Q1 is c onnected in a Colpitts osc illator
circuit with components C2, C3, C5, C6, and LI making up the os cillator's tuned circuit. C hanging any one or any c ombination of thes e
components will vary the oscillator's operating frequency.
Increasing the value of any capacitor will lower the oscillator's frequency and decreasing the value will inc rease the frequency. Inc reasing
L1's inductance will also cause a decrease in frequency and vice versa.

Fig. 1. Here is the schematic for the single-transistor cir cu it.

Transistor Q1 is a g eneral-pu rpose, NPN tr an sistor ; an d it serves as the heart of a Colpitts oscillator circuit.

PARTS LIST FOR TH E SIN GLE-TR ANSISTOR CIRC UIT (F IG. 1)

SEMICON DUCTOR S
Q1— 2N3904, or similar general-purpos e NPN trans istor
RESISTORS
(All resis tors are ¼-w att, 5% units.)
R1—1000-ohm
R2—270,000-ohm
CAPACITORS
C1—.01-uF, ceramic disc
C2—.0001-uF, ceramic disc
C3—-.005-nF, ceramic disc
C4—.1-uF, ceramic dis c
C5 4 to 34 pF, 7-mm, ultra-miniature trimmer, Mouser part #24AA113
C6— 12-100 pF, Mouser part #242-3410-70
ADDITION AL PARTS AND MATERIALS
S1— SPST s witch
L1—Loop, see text
The 6 turns of copper wire can be w ound on a rigid material, such as wood or plastic.
Fig. 3. This detailed diagram of the loop shows the leads
extending from the copper wire, as well as the m akeshift Faraday s hield.

Fig. 4. Here is an ar tist's d iagram of the comp leted metal-detecto r un it.

Any inexp ensive AM tr an sistor radio can be used in conju nction with th e detecto r.

Building Th e Loop
The search loop may be constructed in several different way s; however, the method offered here should get y ou headed in the right direction.
Refer to Fig. 2 as a guide for construc ting the loop. T he loop form should be cons tructed from non-metallic and non-m oisture-absorbent
material. A sealed wood form will do, and it can be either s olid or hoop-lik e. The form should be % to 1 inch wide to allow room for the coil
windings. Clos e wind s ix turns of #20 enameled or insulated wire on the form. Wrap the w indings with at least two layers of good quality
plas tic electrical tape. Put the loop as ide and cons truct the osc illator circuit on a piece of multipurpose PC board with pre-drilled holes.
Stability is one of the most important cons iderations in building any stable oscillator circuit, s o keep all component leads short and s olidly
mounted.
The tw o variable capacitors s hould be mounted in a manner that allows tuning from outs ide the enclos ure. In order to achieve the best
results, the circuit should be hous ed in a metal c abinet to which the c irc uit ground is connected. Temporarily connect the loop to the circuitry
with about 30 inc hes of shielded microphone c able or 2-conduc tor intercom wire. Any wire gauge from #18 to #24 will do. Actually two
insulated wires may be twisted together by hand and used.
Plac e the loop away from any metal objec t and apply power to the circuit. Loc ate a trans is tor radio near by and tune in a station somewhere
near the middle of the dial. Adjust both C5 and C6 to a frequency that will heterodyne with the broadcast station. If nothing happens, it is
most likely that the osc illator is not operating near the desired frequenc y. Now, how do we determine if the osc illator's frequency is too low or
too high? Naturally, a frequenc y counter would be the eas ies t way to determine the oscillator's frequency. If one is not available, what then?
A shortwave receiv er that runes both below and above the standard AM broadc as t band c an be used to ferret out the os cillator's frequency.
Onc e the oscillator's frequency is determined, adjustments can be made to move the frequency into the broadcast band. Reducing the total
capac itanc e of the oscillator's tuned circuit or lowering the induc tanc e of the loop w ill raise the frequency. Lowering the frequency is
accomplis hed by increas ing the c apacitance of the tuned circuit or by increas ing the induc tanc e of the loop. Remov ing or adding a turn to the
loop is a good method to use if the os cillator is way off frequency.

Add ing A Faraday Sh ield

The search loop normally s cans the ground in a parallel manner in search of metal objects. The loop's parallel position to the ground forms a
capac itanc e to ground, which shifts the oscillator's frequenc y. As the loop mov es up and dow n above the ground, the oscillator's frequenc y
shifts in a like manner. Adding a Faraday shield to the loop w ill help in reducing the ground-effect frequenc y-shift problem.
The Faraday s hield is a metal s hroud that is formed around the loop with an insulating gap in the middle. A shield can be formed out of
aluminum foil by cutting a length that's 3 inches wide and long enough to go almost completely around the edge of the loop while leaving a
gap of 1 to 2 inches in the middle, see Fig. 3. Once the aluminum foil is formed, add a 4-inch length bare wire under the foil at one end and
glue the shield in place. Place the loop on a flat surface and place a s olid object on top to secure the foil to the loop form. After the glue dries ,
connect the other end of the bare wire to the loop's ground-end connection.
An old broom handle or dowel rod is attached to the middle of the loop and serv es as the handle and s upport for the loop and detector circuit.
See Fig. 4. T he AM radio m ay be attached to the handle as well or carried separately.
Pos ition the loop over the area to be s earched and tune the oscillator to produce an audible beat frequency tone. Max im um sensitivity is
achiev ed when the oscillator is within a few cycles of the broadcast s tation. This detector will detect all types of m etal, so be ready to dig, and
then dig some more.

Fig. 5. The crystal-filter m etal-detector circuit is sh own above.

The nar row band-pass of the crystal allows for a h igh sensitivity to minute frequency changes.

PARTS LIST FOR TH E C RYSTAL-F ILTER DETECTOR (FIG. 5)

SEMICON DUCTOR S
D1, D2— 1N9L4 silicon signal diode
Ql, Q2—2N3904, or similar general-purpose NPN transistor
RESISTORS
(All resis tors are ¼ watt, 5% units .)
R1,R3— 1000-ohm
R2—270,000-ohm
R4—-See text.
CAPACITORS
C1—.01uF c eramic disc
C2—,0001uF ceramic dis c
C3—.005uF ceramic disc
C4— 0.1uF ceramic disc
C5—See Parts Lis t for Fig. 1
C6—See Parts Lis t for Fig. 1
ADDITION AL PARTS AND MATERIALS
XTL1— 1-MHz crys tal
Ml— 50-uA to 1-mA meter
Metal cabinet, PC board material, etc.

Cr ystal-Filter Detecto r
Our next entry is a vers ion of one of my favorite metal-detector circ uits. A loop and an oscillator circuit similar to the one in our previous
detector are the basic ingredients used in the crys tal-filter detector. The addition of an emitter follower gives isolation to the oscillator and
supplies a low-impedanc e source for the crystal. The output is rectified by D1 and D2 and fed to the meter. Take a look at Fig. 5, as you
continue to read the circuit desc ription.
Here's a brief description of how the crys tal-filter metal-detector circuit operates. The oscillator is tuned to the series resonance frequency of
the cry stal, which c an be any frequency from 100kHz to over 1MHz. Howev er, in our circuit, a 1-MHz crystal is used. When the os cillator is
operating at the cry stal's frequency, the output at the meter is at maximum.
Any shift in the oscillator's frequency will cause a reduction in the meter reading. The circuit is very sensitive to small frequenc y s hifts due to
the cry stal's narrow band-pass characteristic s in the s eries mode. The bas ic loop c onstruction used in the previous detector circuit may be
used here as well.
This detector's circuitry should be cons truc ted in the same manner as our previous circuit. If any component moves or vibrates during use,
the meter will falsely indicate a detected object. Build it solid. The choic e of the meter used for M1 can vary from a sens itive 50-uA to a 1mA
movement. The value of R4 is selec ted for a full-scale meter reading when the os cillator is operating at the series -resonance f requency of the
crys tal.

Fig. 6. The detec tor shown in the above diagram is excellent for deep level searches.
The "90-degree out-of-phase" relationship of the two s quare loops helps limit cross-interference between the transmitter and receiver, thus
eliminating feedback during operation.

Transmitter/Receiver Detector
Our last detector circuit is suitable for locating large metal objects at greater depths— feet instead of inches . This two-box detector has been
around for about 75 y ears and is still one of the mos t popular deep-searching detectors. The bas ic sy stem is shown in Fig. 6.
Two non-metallic box es serve as the 2 hous ing for the electronics and the forms for the loops. The transmitter and receiver boxes are
mounted on a 3-foot-long wood handle, with the receiv er placed in a horizontal position and the trans mitter in a vertical position. This 90-
degree relationship between the trans mitter and receiver allows for minimum transfer of signal between the two loops . Placing a large metal
object between the two loops causes the trans mitter's field to distort, allowing some of the signal to reac h the receiver's loop. The signal is
amplified by the receiver and indicated on the meter as metal detec ted.

Building Th e Tr ansm itter

We'll s tart with the transm itter circuit firs t, (see F ig. 7) because it is the simpler of the two units . T he transmitter circ uit is v ery similar to our
prev ious two oscillator circuits, with a slight variation in the base bypas s c ircuit. The values of frequency-determined capacitors, C1 and C2,
are the same. Depending on the s iz e of the loop, they can vary from .01 to .1-LlF.
The receiv er loop normally requires a capac itor equal to Vi the value of C1 or C2 in the transmitter c irc uit. The transmitter loop is tuned with
C1 and C2, which are always the same value. The actual value of capacitance across the transmitter loop is Vi the v alue of either C1 or C2.
It is mos t important that both loops are tuned to the s ame frequency.
About any loop size from 8 to 12 square inches will do, but we'll stick to the 12-inch box and offer values for that s iz e. The loops are formed
by c lose winding 20 turns of #24 to #26 wire around each hous ing. Run about 8 inches of wire from each end of the loop to the ins ide of the
housing for c ircuit connections. Tape the winding in place with plas tic electrical tape.
The operating frequenc y will be somewhere between 35 kHz and 50 kH z. T he c apacitor values for C1 and C2 are 0.1uF for the transmitter
and .05-uF for C1 in the rec eiver circuit. Les s turns or smaller loops may be used for higher frequency operation. Try and k eep the operating
frequency below 200 kHz, as this type of metal locator works best at low frequencies.

Fig. 7. The transmitter circuit in the above schem atic operates in a r an ge of 35 to 50 kHz.
The oscillator circu it is similar to the previous two m entioned.

PARTS LIST FOR TH E T RANSMITTER (FIG. 7)

SEMICON DUCTOR S
Q1— 2N3904. or similar general-purpos e NPN trans istor
CAPACITORS
C1.C2— .01 to 0.1uF, ceramic dis c (see text)
C3, C4— 0.1-u.F, ceramic dis c
RESISTORS
(All resis tors are ¼watt, 5%)
R1—1000-ohm
R2—270,000-ohm
R3—220.000-ohm
ADDITION AL PARTS AND MATERIALS
S1— SPST s witch
L1—Loop, see text

Building Th e Receiver
The receiv er (s ee Fig. 8) is a simple two-trans is tor RF amplifier circuit with an isolated emitter follower input. The R F signal is picked up by
the loop and coupled through Q1 to the input of the firs t RF amplifier stage, Q2. Transistor Q2's RF gain is set by R10. The signal from Q2's
collector is fed to the bas e of Q3, and Q3's output is c oupled to a two-diode detector circuit. T he DC output is indic ated by Ml.
The receiv er c irc uitry will fit on a 2 x 3-inc h piece of multipurpose PC board material. Mount the components clos e to the board with short
leads and keep the input components away from the output c ircuitry. The meter c an be any DC type with s ensitiv ity of 50-uA to 1-mA. If a
50uA meter is us ed, R11 may need to be increased to a 10K potentiom eter. Mount the circuit in the receiver box and connect the loop.
Mount the transmitter box on one end of the wood handle and the receiver on the other. The receiv er w ill need to be mounted so that it c an
be tilted up and down to obtain a balanc e betw een the tw o loops. T his can be acc omplished by using a s mall hinge attached to the end of the
handle and the receiver housing. Onc e the balance point is found, the rec eiver c an be mounted in that position.
Fig. 8. The rec eiver circuit.

PARTS LIST FOR TH E R ECEIVER (FIG. 8)

SEMICON DUCTOR S
Q1-Q3—2N3904, or similar general-purpose NPN transistor
D1, D2— 1N914 silicon signal diode
CAPACITORS
C1—.005 to .05-uF c eramic disc (see text)
C2 0.1— .05-uF ceramic disc
C5-C9 — 0.1 uF ceram ic disc
C10— 470uF 25WVDC electrolytic
RESISTORS
(All resis tors are ¼watt, 5%)
R1,R2— 100,000-ohm
R3-R5—1500-ohm
R6— 100-ohm
R7—470-ohm
R8, R9— 220,000-ohm
R10— 1000-ohm potentiometer
R11—2500-ohm potentiometer
ADDITION AL PARTS AND MATERIALS
S1— SPST s witch
M1—50-uA to 1-mA DC meter
L1—Loop, see text
Simplest Metal Detector Circuit

The simples t metal detector c ircuit cons is ts of 4 components.

The detection-coil consists of 70 turns of 0.3mm wire 120mm diameter.
Plac e an AM radio near the searc h-coil and tune it to pick up a squeal. When a coin is placed near the coil, the tone will alter.
The circuit operates at about 250kHz and the the radio picks up a harmonic . It will detect a bottle-top at about 90mm.

Another simple Metal Detector Circuit c onsists of two c oils that overlap to provide feedbac k:

Plac e an AM radio near the coils and a tone will be heard.

Sim ple BFO Metal Detector

The first two transistors are deigned to produce the exact same frequency. The res ult is fed to the third trans istor to produce a beep-beep-
beep in the piezo speaker.
When a metal object is detec ted by L1, (the search coil) the beeping increases.
The performance of this circuit is no better than the s ingle trans is tor circuit abov e and an AM radio. When it is all boiled-down, the first
transistor drives the coil and the frequency of the circuit changes when the inductance of the coil is altered by a metal object near it.
The second and third trans istors are equivalent to an AM radio by detecting the frequency of the oscillator and producing a result that is the
difference between the frequency produced by the firs t transistor and the frequency produced by the s ec ond transistor.

The nex t c irc uit is even more com plex.

It uses integrated c irc uits (IC's) to detect the difference between the two frequencies.
The circuit is very old. All the IC 's can now be replaced by an 8-pin microcontroller and the circ uit will be much simpler.
Don't forget, it is no better than a single transistor and an AM radio and is not worth cons tructing, howev er the text is interesting . . .

Simple, sensitive metal detector

Phil Wait
The metal detecting hobby is enjoying quite a boom at the moment and treasure hunters are not just after gold. The price of the precious
metal has risen in recent months, at around $1,600 an ounce it's worth going after. Old coins and relic s fetch high prices too, so there's lots to
find out there...

METAL DETECTORS depend on detecting one of several effects that c an be obs erved when a metal object influences the magnetic field
surrounding a coil of wire c arrying an alternating c urrent. T he princ ipal effects are: the pattern of the magnetic field surrounding the coil will
be altered and the inductance of the c oil w ill change.
The various types of metal detector devised exploit these changes, electronic ally detecting the alteration induced in the coil by the metallic
object. N on-metallic objects or material can also affect the coil in similar way s.
There are three basic methods employed to exploit the above effects. "Induction Balance" (IB) metal detectors employ two coils. One is
driven by a modulated os cillator. The other is connec ted to a detector and amplifier. The two coils are carefully positioned with res pect to one
another such that the rec eiver c oil picks up very little of the energy radiated by the transmitter coil when no metal or mineral material is
nearby. When the coils are brought near a metal object, the field pattern is distorted, greatly increasing the transmitted energy picked up by
the receiver coil. The modulated signal is detected and can be indicated by amplifying the recovered modulation to speaker level as well as
indic ating it on a meter. F or obvious reasons, this type of metal detec tor is often referred to as a "transmit-receiv e" or TR detector, sometimes
as an IB/TR detector. Chief adv antages are good pinpointing ability and good depth penetration, and they are not sensitive to small ferrous
objects.

Sensitivity suffers badly in mineralised or ironstone ground.

Most IB detectors operate at a frequenc y between 85 k Hz and 150 kHz . As they are badly affected by mineralised ground a technique was
developed using very low frequency to energise the transmit coil. The 'VLF' ty pes operate at frequencies around 4-6 kH z, a frequency range
which penetrates all types of soil quite well. However, they need to run at a fairly high power to achieve s ufficient sensitivity with s mall
objects, hence battery drain is quite high, and pinpointing ability is poor.
"Puls e Induction" detectors employ coils in the s earch head that are set up in much the same manner as the IB detector. However, the
transmitter is pulsed so that high energy bursts are transmitted by the searc h coil. The rec eiver then compares the phase of portion of the
receiv ed pulse with the transmit signal. When a ferrous or magnetic objec t is brought near the search coils the phas e of the receiv ed s ignal is
advanced with respect to the transmit signal. The opposite occ urs when a non-magnetic conductor is brought near the s earch coils. Thus,
this type of detector c an effectively 'disc riminate' betw een ferrous and non-ferrous metals as well as exc lude ground effects — s imply by
setting the detection circuitry to exclude signals of the unwanted phase characteristics. Thus , a "Ground Exclusion" c ontrol is often featured
with thes e detectors. As the s trength of the rec eived signal also v aries, depending on the 'target' objec t's charac teris tics, this effec t may also
be inc luded in the detection process.
Clearly, an IP detector presents many problems to the home c onstructor.
The simples t tec hnique detects the change in inductance of a single search coil. If this coil is part of the tuned c ircuit of an oscillator, then
comparing the frequency of the 'search' oscillator with a stable reference oscillator will indicate the presenc e of a metal object. This detector
is called the "Beat Frequenc y Oscillator" or BFO type. The two osc illators are set suc h that there is a s light differenc e in their frequencies and
their outputs mixed. The resultant will be a 'beat' frequency which is equal to the differenc e between the two osc illator frequencies. The main
advantages of this ty pe are simple circuitry and setting up along with good pinpointing ability. In the past, most published designs hav e
suffered from a distinct lac k of sens itivity as well as poor tuning stability. A cunning mixing technique and a few other fillips c an overcome
these problems.
Hence, our new metal detector is a BFO type incorporating s ome modern refinements . It has proved to hav e similar s ensitiv ity to our IB
detector, the ETI-549, but is generally eas ier to build and s et up, there being no critical adjustments.

Design features
Our new metal detector has three controls : COARSE frequency adjust, FINE frequenc y adjust and VOLUM E on/off The coarse frequency
control is used to initially set the frequency of the search osc illator, compensating for the v arious factors affecting any drift in this os cillator
(mainly temperature and battery v oltage). The fine frequency c ontrol is then used to set the note to a low pitc h when the detector is placed
over the ground, permitting com pensation for the effect of the ground on the frequenc y of the search osc illator. The volum e control adjusts
the loudness of the output from the speaker.
The tw o main design problems this ty pe of detec tor presents are the frequency stability of the two oscillators and the minute f requency
change which has to be detected.
The search os cillator w e finally us ed was settled on after some experimentation. Our firs t try employed an LC oscillator built around a CMOS
gate chip. This proved to be not as stable as we required and we found that trying to obtain dc control of the frequenc y by vary ing the supply
rail voltage had drawbacks. After some experimentation with oscillator configurations we hit on a discrete component oscillator whic h we
found behav ed much as we were seeking.
The search coil in the circuit we used is the inductor in a Colpitts os cillator. However, this particular c irc uit may be a little unfamiliar to many
readers. To inc rease the RF current in the coil, it is placed in the collector circuit of Q1. Feedbac k is betw een collec tor and emitter and the
base is effectively at R F ground. The frequency determining capac itance of the tuned c irc uit is 'tapped' to prov ide feedback , C2 and C3
performing this function. Careful attention has been paid to the bas ic frequency stability of this oscillator. Good quality s tyroseal capacitors
have been used for C2 and C 3. These have a temperature coefficient roughly opposite to that of other temperature influences on the
frequency of the oscillator. In general, the short-term stability of this osc illator is quite good.
The particular circuit configuration of the oscillator gave us a very useful bonus — dc control of the osc illator frequency ov er a small range.
Varying the base bias on a transistor will vary the collector-bas e capacitance. In this circuit, the c-b capacitanc e is part of the ov erall 's tray'
capac itanc e that determines the exact frequency of os cillation. As the bas e bias is inc reased the c-b capac itance decreas es , increasing the
oscillator frequency. In this way, the oscillator frequency c an be varied over a range of about ten perc ent. We hav e provided two controls , the
FINE control providing a variation of about one-tenth that of the C OARSE c ontrol.
The search os cillator is loosely coupled via a 47p capacitor to a following CMOS Schmitt trigger and two inverters whic h square the output.
The loos e coupling is olates the osc illator from the subsequent circuitry, further enhancing die stability of the search osc illator.
For the reference osc illator, we chose to use a c ry stal, because of its inherent stability. It has been argued that if an ordinary LC circuit is
used for the reference oscillator it w ill have similar drift c haracteristics as the search oscillator and the overall drift will be reduced. In fac t, the
reference oscillator can be made using a standard 455 k Hz IF transformer. In prac tice how ever the two tend to drift at markedly different
rates. We think the best approac h is to mak e both osc illators as stable as poss ible. Hence the c ry stal — which is an easily available type and
cheaper than an IF transformer!
The reference oscillator is a s imple 'inv erter' crystal os cillator built around one gate from a C MOS quad NAND gate, IC2. This has a square
wave output and drives a divide-by-four circuit, IC3, via the other three gates in IC 2, acting as buffers.
The crys tal we used is a 3.579545 MHz type (NTSC chrominanc e sub-carrier frequency ) commonly av ailable from a number of suppliers.
The output of IC3 is at a frequenc y of about 890 kHz. The ex ac t frequency is unim portant, just so long as its stable.
The search os cillator operates at a little above 100 k Hz, about one-eighth of this frequency.
The secret of our metal detector's overall s ensitiv ity lies in the mix er c irc uit. This employs one section of a 4013 flip-flop. The reference
oscillator's divider output (at 890 k Hz) is applied to the D input of IC4a and the squared-up search oscillator's output is applied to the clock
input. If the c lock frequency (i.e the search oscillator frequency) c hanges by 1 Hz, the output beat (from the Q output of IC 4a) will change by
8 Hz (see 'How it Work s'), thus c onsiderably multiplying the smallest changes in oscillator frequenc y.
The output of the mix er is fed to a s imple audio amplifier driving a loud-speaker. The searc h and reference oscillators must be w ell decoupled
from each other and buffered from the mixer stage to prev ent 'pulling' of the oscillators, which would res ult in erratic operat ion, es pecially
when set for a low frequenc y output. We have used supply line decoupling as well as buffer s tages after eac h osc illator. We also found it
neces sary to use a s eparate battery for the audio stage to prevent the v ery short, but high current pulses to the audio stage affecting the
oscillators .

The search coil

The most important c harac teris tic of the searc h coil is its size. Surprisingly enough the actual inductance doesn't s eem to have much effec t
on sensitivity. The greater the coil diameter the greater the penetration depth, but the less s ensitiv e it is to small objects. As a general rule the
penetration is about equal to the s earch coil diameter, while the sensitivity is roughly proportional to the cube of the object diameter (as
expres sed as a function of the s earch coil diameter). Sens itivity is also invers ely proportional to the s ix th pow er of the distance between the
coil and the object.
All this means is that if the object size is halved the sens itivity is reduced to one-eighth. Also, if the depth is doubled the s ensitiv ity is reduced
to one sixty-fourth. Its easy to see why all metal detectors which are designed to pic k up small objects use small coils, (150 to 300 mm
diameter) and really only sk im the soil surface. If the s earch coil is doubled in diameter for greater penetration the s ensitiv ity to small objects
falls to one-eighth. You rapidly encounter the law of diminishing returns.
Som e of the more ex pensive metal detectors improv e the penetration, while retaining sensitivity, by using a very complex arrangement of
coils w hich modifies the field pattern. This c an be done to some ex tent by making the coil on the BFO detector ov al in s hape.
We chose a round coil of 150 mm diameter to giv e good sensitivity to s mall objec ts giving about 100-150 mm penetration which is easy to
build, but this is open to c onsiderable ex perimentation. R emember though, that if the c oil diameter is increased the number of turns will have
to be reduced so that the searc h os cillator remains at the s ame frequency (about 110 kHz ).

Faraday shield
If the s earch coil is moved around, the c apacitance between it and the ground or other objec ts changes . This changing capacitanc e 'pulls ' the
oscillator frequency and can completely swamp out the small change in inductance we are looking for. The coil can be screened from this
capac itanc e effect by us ing a F araday Shield around the c oil. This consists of a ring of tubing, or in our c ase — a wrapping of aluminium foil,
around the coil but brok en at one point so it does not make a shorted turn. This shield is then c onnec ted to the common supply rail (0V) on
the oscillator.
Construction
We hav e deliberately c hosen commonly av ailable m echanic al and elec tronic com ponents so that cons truc tion of this projec t is as eas y as
possible — especially for the new comer. The search coil is mounted on a 165 mm diameter plastic pot stand which may be purc hased at
hardware stores and nurseries. The electronics are mounted inside a simple aluminium box attached to a stem made from a length of tube
which extends dow n to the search coil and s erves as the handle. Connection to shield the search coil is v ia a length of s hielded cable. The
controls mount on one side of the box housing the electronics. Which s ide y ou mount them depends on whether y ou are right or left handed.
The speak er m ounts on the end of the box fac ing the operator. As c an be seen from the picture, the handle was made with an upw ards bend
at the end which you grip. This balances the instrument reas onably well, avoiding arm strain.
Construction should commence with the electronic s. M ount the c omponents on the pc board, tak ing care with the orientation of the trans is tor
(Q1) and the ICs . Do not substitute another type of capacitor for the sty roseal ty pes specified for C2 and C3 or performance may s uffer. The
crys tal specified comes with fly ing leads and may be s oldered in plac e. Don't use too much heat though, s older quickly and y ou will avoid
possible damage to the crystal.
The nex t s tep is to mak e the stem. The easiest way is to take a length of 25mm diameter electrical conduit about 850 mm long and mak e a
bend about 100 mm from one end for the grip. To do this, heat the point of the bend over a flame (not in the flame) until it softens and then
carefully bend it about 60° from straight.
A length of aluminium tube may als o be used for the handle. The bend for the grip c an be made by first flattening the point of the bend
somewhat with a hammer then placing the short piec e in a vice and c arefully making the bend. A s ec tion of wood dowel or plastic tube
should be plac ed between the searc h coil and the end of the metal tube to k eep the mass of metal about 200 - 250 mm away from t he search
coil. A piece of w ood dow el of the right size, jammed in the end of the aluminium tube, is generally the easies t way to go about it.
We used a small aluminium box whic h comes in two pieces. We drilled a hole in either end of the bottom of this box so that it c ould be
slipped over the stem (see ac companying photograph). A nut and bolt was us ed to secure it to the stem on the side 'below' the grip. The
small speaker is mounted in this part of the box, before it is secured to the stem, on the end which faces upward toward the operator. A sm all
hole is drilled in the opposite end and a grommet ins erted. This permits entry of the cable to the search c oil.
The pc board and controls are mounted to the 'lid' of the box. Position the c ontrols on the side that suits y our handedness. Our model was
made for right handed operators.
Now for the search c oil. This is wound so that it can be tucked inside the rim of the up-turned plastic pot stand. Firs t make a c ardboard
former of the appropriate diameter. Roll a strip of heav y cardboard around the rim suc h that it fits loosely and tape or staple it securely (to
avoid it popping open at an awkward mom ent).
Lift the former off the pot s tand and then wind the coil onto this former as per the details given in the parts list. Leav e a short length of wire
spare on each end to make the connection. Tie the coil up with a few lengths of string at various places and then s lide it off the former. Now
wind two layers of insulation tape around the c oil, leading the two ends out at the same place.
Next, wind the Faraday s creen. C ut some aluminium kitc hen foil into strips about 15 mm wide and wind this around the coil to make two
layers but leav ing a s mall gap about 5 mm to 10 mm w ide where the coil ends c ome out. It is v ery important that the two ends of the F araday
shield do not c onnec t as this would mak e a 's horted turn' and the coil would not work as intended.

To s ec ure the foil tightly around the coil and to make connection to the shield, wind a length of tinned copper wire around the shield with
about a 10mm pitc h (i.e: about 10mm between successive turns). The end of this wire is tak en out at the same plac e as the coil connections.
Now w ind another two layers of insulation tape around the whole assembly. Drill a 3 mm hole in the side of the pot stand and then press the
coil down into the rim with the c onnec ting wires adjacent to the hole. Pass the w ires through the hole. Pour quick- s etting epoxy over the c oil
to hold it in place.
The search head is mounted to the stem using two right-angle brac kets and a bolt pas sed right through the end of the s tem. Small pieces of
metal here don't seem to advers ely affect the operation of the detec tor.
Solder the coil connections to the twin s hielded cable, the Faraday shield connecting to the cables shield, and glue the cable and wires
underneath the pot s tand to hold them rigid. If you wish, the 'underside' of the pot stand may be completely filled with epoxy.
Wind the cable around the stem to keep it mechanic ally rigid and pass it through a grommeted hole in the box. Terminate the cable to the pc
board.

Using it
When the cons truc tion is complete, turn on the detec tor, advance the volume control and rotate the coars e frequenc y knob. You will hear a
number of 'heterodynes ' or beats, one being v ery strong. This heterodyne is the one commonly us ed, the others being odd multiples of the
reference signal beating with multiples of the s earch oscillator. You may find that some of these w eaker signals are more sensitive to buried
objects than the stronger one.
Set the fine frequency control to midrange and set the course frequency control to near the s trong heterodyne with the s earch head held
away from the ground. Lower the detector to the ground and you will notic e a frequenc y shift. This is the effect of the ground and will vary
between different types of soil. Use the fine frequency c ontrol to set the beat to a low pitch and s weep across the s urface. A metal object will
cause a change in the pitch w hich is clearly audible.
The ear is more sens itive to c hanges in pitc h at low frequencies than at high frequenc ies and thus it is best to adjus t the fine frequency
control to a low pitc h that can be heard at a comfortable volume from the loudspeaker.
Theoretically, the frequency of the search oscillator should inc rease when a non-ferrous objec t c omes within range of the searc h coil and
decrease when a ferrous (or diamagnetic) object is within range. This effect is difficult to detect in practice as eddy c urrent s in ferrous
materials swamp the effect and they react much the same as non-ferrous metals. H owever, minerals s uch as hematite may show the effect.
With the search oscillator set on one side of zero beat, metal objects near the search coil will c ause the pitch to increase, while magnetic
minerals w ill cause the pitc h to decrease. With the searc h os cillator s et to the other side of z ero beat, the opposite will occur.
You could try a few ex periments to s how up this effect.
Enough theorising. In general operation, try to keep the searc h head a cons tant distance from the ground and s weep from s ide to side in a
regular pattern. The right technique is eas ily dev eloped with a little prac tice.
There are a number of books on metal detecting available and these show the s ort of techniques the suc cessful treasure hunter employs.
HOW IT WORKS
The beat frequency metal detector employs two oscillators: a very stable reference oscillator and a search oscillator. T he s earch
oscillator uses a tuned c ircuit designed to be influenc ed by metal or mineral objects whic h are brought into its field. The two
oscillators are adjus ted so they are harmonic ally related and fed to a mixer. When the search frequency is adjusted so the
reference frequency fed to the mixer is eight times the search frequency, the output of the mixer is zero. The search frequency
is s lightly adjus ted so that an output appears from the mix er which is the difference between the two input frequencies . This can
be adjusted to an audio tone.
When a piece of metal or mineral is brought near the search coil the frequency of the oscillator varies, which in turn varies the
output frequenc y from the mix er. T he c hange in pitc h can easily be heard from the speaker.
The reference oscillator employs a c ry stal in a CMOS oscillator circ uit using one gate from IC2a. The resistor R6 biases the
gate into its linear region. IC2 b, c and d, are used as buffer s tages to prevent osc illator pulling and to further square its output
waveform. Two flip-flops, IC3a and b, divide the reference signal by four to 890 k Hz.
The search osc illator uses a discrete transistor in grounded base configuration, with the search c oil in the collector. Us ing t he
coil in the collector inc reases the s trength of the field around the c oil and hopefully overc omes some of the losses in the ground.
Feedbac k is set by the ratio of C2 to C3 from collec tor to emitter and their value determines the frequency of the osc illator. The
base is grounded at R F by C4.
By varying the bias on the transistor the inter-element c apacitances can be varied. This varies the oscillator frequency as the
trans istor capac itances form part of the s tray s in the LC c irc uit. RV1 and RV2 provide fine and c oarse frequency control. The
res is tors R8 and R9 limit the maximum and minimum v oltage on the bas e to prevent over-dis sipation in the transistor or drop-
out of the osc illator.
The output of the search oscillator is fed to a Schmitt trigger, consisting of IC1a and b, where it is squared and further buffered
by IC 1c and d. The search frequency is then fed to the mixer.
Both osc illators are dec oupled from each other by supply line decoupling R1, C1 and R5, C6.
The mixer cons is ts of half a dual-D flip-flop. The search and reference frequencies are fed to the cloc k and D inputs
res pectiv ely. The flip-flop look s at the referenc e os cillator (D on every positive transition of the search oscillator clock), and
trans fers this level to the Q output until the next clock transition. If the two oscillators are ex actly evenly harmonically related
(i.e: 2nd. 4th, 6th, or in our c as e 8th, harmonic) the D input will always be the s ame level at each clock pulse. The output from
the mixer at the Q pin will alway s be the same — no pulses.
Howev er, if the search frequenc y is varied and the D and cloc k inputs are no longer harm onically related but are changing in
phase with respect to eac h other, after a few clock pulses the D input will no longer be the same — the output will change state.
The effect of all this is to produce a chain of square waves at the Q output, the frequency of which is eight times the change in
frequency of the search osc illator.
Capacitors C8 and RV2 form a differentiating network which feeds a puls e to the audio amplifier, Q2, for each output transition
from the mixer. Eac h cycle from the mixer produc es two pulses in the speak er. If the frequency of the search oscillator is s hifted
one hertz the output of the mixer changes by eight hertz, produc ing an output of eight pulses per second in the speaker.

PARTS LIST
Resistor s all 1/2W, 5%
R1 100R
R2 1k
R3 100k
R4 1M
R5 100R
 R6 10M
R7 22k
R8 R9 4k7

Po tentio meter s
RV1 10k lin
RV2 100k lin
RV3 100k log switc h pot

Capacito rs
C1 100ngreenc ap
C2 1n sty roseal
C3 5n6 s tyroseal
C4 100n greencap
C5 47p c eramic
06 100n greenc ap
C7 10p c eramic
C8 100n greencap

Sem iconductors
Q1, Q2 BC548, BC108, etc.
IC1, IC2 . . . . 4001B
IC3, IC4 . . . .4013
Miscellaneous
SP1 8 ohm speak er
B1,B2 9 Volt battery (type 216)
Xtal NTSC
colour xtal
Metal Detector PC board

Length of twin shielded cable, plastic pot

stand (approx 150 mm dial, length of
steel or aluminium tube (approx 600 mm
long, 20 mm dia), length of plastic rod or
wood dowel to fit inside pipe (approx 200
mm long), 0.4 mm enamelled wire,
aluminium foil, Araldite, box to suit
(approx 105 x 125 x 75 mm), three
knobs, battery c lips, ins ulation tape, tw o
right-angle brac kets.

Matchless
Metal Locator
by Thomas Scarborough

Want to find a fortune? Buried treasure, perhaps? Lost coins on the beach? Or perhaps you
fancy earning some pocket money finding other people's valuables. Either way, this project
should really interest you. It's an el-cheapo induction balance (IB) metal locator that delivers
surprisingly good performance.

An induction balance (IB) metal locator has a good depth of penetration and dis tinguishes well between ferrous and non-ferrous metals. It is
also capable, to a large extent, of rejecting iron and also tin foil This is a boon for anyone who is searching for c oins or noble metals.
My aim with this design was to create a 'minimalist' device — one that would work well but without all the bells and whistles of the expensive,
commercial designs. I found that it was possible, with just a handful of c omponents, to design a high-quality metal locator.
Simple, but it works
An IB metal locator is usually far more complex than the design show n here.
The reas on for the simplicity is that I have dispens ed with analogue circuitry, and instead used a digital transmitter and receiv er.
As the s earch coils pass ov er metal, only digital signals of a certain amplitude break through to a peak detector (IClb). Since these are in the
audio range, they are immediately trans ferred to the piezo sounder or headphones.
On testing the sensitivity of this des ign in air, with optimal tuning and us ing a 25mm-diameter brass coin, it gave a c lear signal at 150mm, and
a 'scream ing' signal at 110mm. It was also able to detec t a pin at 30mm.
Note that these figures may not apply in the ground, where depth of penetration will depend largely on the mineralisation present.
In contrast, the locator is far more reluctant to pick up tin foil. A tin foil disk of the same size as the brass coin w as only detected at half the
distance in air. This rejection of tin foil is due in part to the metal locator's low frequency, which avoids what is c alled skin effect.
Bes ides this , if the tw o coils are positioned as described, ferrous metals (iron) are, to a v ery large ex tent, rejected — to such an extent, in
fact, that a 25mm diameter bras s coin weighing seven grams looks the same to the metal locator as a lump of iron weighing 20 times as
much. Large nonferrous objects are detec ted at half a metre distance and more.
The locator's power consumption is conv eniently low. It draws around 10mA, which means that it may be powered off a small 9V battery. If an
alkaline battery is used, this will provide about 48 hours' continuous use. In my experience, the number of coins that are found on a beach in
an hour or two should easily make up for the cost of batteries !
Finally, while the s tability of the locator is not the best, it's by no means the wors t either. Re-tuning is nec es sary from time to time, es pecially
in the first few minutes of use. One soon becomes accustomed to giving the Fine Tune knob an occ asional tweak — perhaps with every 40
or 50 sweeps of the s earch head.

Circuit description
The search head of a typical IB metal locator contains two coils: a trans mitter (Tx) c oil and rec eiver (Rx) coil.
In this case, the Tx c oil is driv en by a square wave oscillator, w hich s ets up an alternating magnetic field in the coil. The R x coil is then
positioned in such a way that it partly ov erlaps the Tx coil. By adjus ting the amount of ov erlap, a point can be found where the v oltages in the
Rx coil 'null' or c ancel out, so that little or no electrical output is produced. A metal objec t which enters the field then caus es an imbalanc e,
resulting in a s ignal.
The transmitter (IC1a) is a standard 555 osc illator configuration, using one half of the ICM7556IPD dual low power CMOS version of this IC.
Do NOT use the NE556N IC, by the way.
IC1a osc illates at about 700H z, determined by R /C c omponents around pins 1, 2 and 6. The 680R resistor limits the current passing through
the Tx coil.
The receiv er s ec tion (IC1b) is preceded by a simple yet sensitive preamplifier stage, based on Q1, w hich amplifies the signal received from
the Rx coil. This is fed directly to IC1b, which is us ed here as a high-performance sine-square converter. Its input at pins 8 and 12 is biased
by the divider formed by the 10k res is tor and pots VR1-VR3, so that only pulses of a certain amplitude break through to output pin 9.
There is a point at which, w ith careful adjustment, the signal is just breaking through in the form of a crac kling s ound. When the locator's
output is adjus ted to a fas t crackle, the presence of metal turns this into a 'scream'. This is heard from the piezo s ounder or through standard
headphones . T he 7556 IC allow s up to 100mA of output current, therefore no further amplification is required.

Winding the coils

The one drawback to any IB metal loc ator design is its need for two coils, w hich must be very carefully and rigidly pos itioned in relation to
one another. Sometimes there's no room even for a fraction-of-a-millimetre error in positioning these coils. While this partic ular des ign makes
things eas ier than usual, the placement of the coils will s till require some patience. On the other hand, the winding of the coils is relatively
easy. Eac h coil als o includes a electros tatic (Faraday) s hield, which helps to minimise ground effec t.
The winding of the (identical) coils is not c ritical and a little give and take is permissible.
I used 30SWG (0.315mm ) enamelled copper wire, w inding 70 turns on a circular former, 120mm in diameter.
I created the former with a sheet of stiff cardboard with 12 pins stuck through it at a suitable angle (the heads facing slightly outwards). The
coil was wound clockwise around the pins , then temporarily held together with s tubs of insulating tape passed under the coil and pressed
together over the top. The coil may be jum ble-wound (that is, you don't have to w ind the turns on side-by-side in neat lay ers).
Onc e this has been done, the pins are removed, and a s ec ond coil is wound in the same way. In each case, mark the beginning and end
wires. Each coil is then tightly bound by winding insulating tape around its entire circumference.
Now w e add a Faraday s hield to each coil. This is accomplis hed with s ome long, thin strips of aluminium foil. F irs t sc rape the enamel off
each coil's end wire. Solder a 100mm length of bare wire to the winding wire, and twist this around the c oil, ov er the insulating tape. This
prov ides electrical contact for the Faraday s hield.
Beginning at the base of this lead, the foil is wound around the circumference of the coil, so that no insulating tape is still v is ible under the foil
but the foil should not complete a full 360°. Leave a s mall gap (say 10mm) so that the end of the foil does not meet the start after hav ing
gone most of the w ay around. Do this with both coils . Each coil is now again tightly bound with insulating tape around its entire
circumferenc e.
Attach each of the coils to its own length of quality single-c ore sc reened audio c able, with the Faraday shield in each case being s oldered to
the sc reen. Do not us e stereo or twin-core microphone w ire to run both leads together; this may cause interference between the c oils.
Gently bend the completed coils until each one is reas onably flat and circular, w ith each end wire facing aw ay from you, and to the right of
the beginning wire. N ow bend them further until they form lopsided ovals like capital Ds (s ee F ig. 2). The bac ks of the Ds ov erlap each other
slightly in the c entre of the search head. This is the critical part of the operation, w hich w e shall c omplete after having c onstructed the circuit.
Last of all, wind strips of absorbent cloth around each coil (I us ed s trips of thin dishwashing cloth s uc h as Chux ), using a little all-purpose
glue to keep them in place. Later, when epoxy resin is poured over the coils, this c loth meshes the coils into the res in.

Construction
The PC board of the Matchless Metal Loc ator measures 48m m x 42mm, and is coded 04106021. There are not many c omponents, so it
should be easy to as semble the board using the PC board overlay diagram in Fig. 3.
With the exc eption of the CMOS IC, c omponent values and types are not critical. The one critical c omponent is the IC M7556IPD CMOS IC. I
also tried the TS556CN IC in this position — it w orked, but not as well.
Begin board assembly by s oldering the nine terminal pins, the 14-pin dual-in-line socket for IC1 and the res is tors. C ontinue with the
capac itors , diodes and Q1.
Onc e soldering is complete, carefully check the board for any s older bridges, then use some short lengths of quality screened microphone
wire to attach the piezo s ounder, VR2 and VR3, with the sc reen (or braid) always being wired to 0V. If you wish, add a socket for headphones
in parallel with or in place of the piezo s ounder. Us e insulated hook-up wire to attach the battery and switch S1, keeping the leads short.
Finally, attach the sc reened cables from the c oils, with the s creen again going to 0V, and insert IC1 in the DIL s oc ket. Note that IC1 is static
sensitive, and requires careful handling (discharge your body to earth before handling).
Fig. 5 shows the suggested hardware construction, using PVC piping and joints . Bend the base of the metal loc ator's s haft under very hot
water to obtain the angle shown. Alternatively, a s wivel joint may be made.
The entire electronics (apart from the search coils) is mounted in a metal case, ensuring that no part of the underside of the PC board is
touching the cas e. The adjustment slot for VR1 should be access ible via a sm all hole in the case. Mount VR2 and VR3 where quick and eas y
adjustment is possible.
A metal case is essential, otherwise the c irc uit is affec ted by electros tatic c oupling (or capacitive effects ). The metal case is connected to 0V,
through the tab on the copper s ide of the PC board.
I was unable to obtain a purpos e-made metal case in my c ity (Cape Town), but found that good quality metal sweet tins were readily
available, so I used one of thes e. They are als o cons iderably c heaper than similarly sized elec tronics enclosures and of cours e you get the
sweets as well!

Setting the coils

A completed PC board is needed before we can 'pot' the c oils. These are potted with epoxy resin in a hard plastic dinner plate, the sort you'd
find in a picnic set. Any plastic plate of s uitable size will do, on condition that it is rigid. (A tip: don't pinch them from the family pic nic set....)

First place the coils on top of one another — ensuring that they are correctly orientated, with each end wire facing away from you, and to the
right of the beginning w ire. Adjust both VR2 and VR3 to their midpoint. Adjust VR1 to about 90kQ. Then attac h a 9V battery and switch on.
The circuit w ill mos t likely be sc reaming; that is, beeping loudly and c ontinuously. N ow slow ly move the coils apart. When they are
somewhere past the halfway point, the headphones will fall silent. This is where the voltages in the R x coil 'null'. Continue to move the coils
apart. At a precise point jus t before the coils no longer ov erlap at all — the headphones will begin to sc ream again (there may or may not be
a low-level beep just before this ).
It is at this precise point, and not a fraction of a millimetre either way, that the coils need to be set.
Take an indelible m arker pen and mark out holes in the low er plate around both coils. These holes are used to pass c able ties t hrough, to
hold the coils tightly to the plate. Als o us e a cable tie to hold the audio c ables to the plate. Use some Blu-tak to tightly seal the holes
underneath the plate before pouring in the res in — epoxy resin can be very 'runny' and sticks faster than many glues.
Also at this point carefully bend the coils at the c entre of the plate until you reac h the exac t balance at whic h there is neither silenc e nor
sc reaming in the piezo sounder/headphones , but just a crackle. A little drift should not matter at this point.
Now y ou are ready to mix and pour the resin. Us e a modest amount of catalyst, so that there will be not too much heat and shrinkage in the
resin. Pour the resin over the cloth which surrounds the coils, so as to soak it, and k eep on pouring at least until the entire bottom of the plate
is covered with resin.
The circuit m ay no longer function c orrectly at this point until the resin has hardened, so mak e no more adjustm ents at this stage, but sw itch
the circuit off and leave it for 24 hours or so.
I potted two sets of c oils (that is, two c omplete searc h heads ). The first worked perfectly, precisely as I had set it in the plate. The second
contracted slightly as the resin set, so that no settings of VR2 or VR3 would produce a tone in the headphones. H owever, this is where the
design of the Matchless Metal Locator shows its flexibility. By turning VR1 clockw is e, the circuit was again functioning normally w hen VR2
and VR3 were set to their midpoint.

How to use it
Keep the s earch head away from all metal — and "noisy" computer equipment — and s witch on. Adjust potentiometers VR2 (Tune) and VR3
(Fine Tune) to their mid-points. Then adjus t VR1 with a screw driver or plastic alignment tool until the metal locator is just at the point where a
crackle is heard, between s ilence and a scream (or between a low-level hum and a sc ream). Us e the tune and fine-tune k nobs for any further
tuning.
A fast crackling sound produces the best results . Move a c oin over the search head and the piezo sounder should s cream.
In actual use, the adjustment of the metal locator will be affected by the mineralisation of the ground you are searching, as well as
temperature and voltage variations. So as mentioned earlier, readjustments to VR3 and VR2 are inevitable from time to time.
That's really all there is to it.
Here is a Metal Detector using a CD4011 IC:
The oscillator is built with N AND N1 and a c eramic filter of frequency 470kHz. The second osc illator is with N3 and a LC combination. The
frequency of this osc illator. is adjusted in such way that will produce an audible oscillation of both frequencies . Thru N4, the signal from the
variable oscillator. is amplified. If the sens or coil L1 is closer to a metal object then it w ill m odify the auto-induction of the coil, the oscillator is
unbalanced and the sound will modify.

The metal detector’s coil is made of: 70 turns of enamelled copper with dia. 0.25mm on a 5cm dia former.

Another Simple Metal Detector Circuit:

This c irc uit is identical to Fig 7 abov e and the Metal Detector
in 200 Transistor Circuits

Another Simple Metal Detector Circuit:

SIMPLE METAL DETECTOR USING A TAPPED COIL
This c irc uit has been tested on 23-12-2013. It produces an oscillation at approx 530kH z.
The sensitivity is the same as all the other s ingle trans istor circuits, but it us es a tapped
coil to provide the feedback and this eliminates two c omponents.

CD 4093 IC Metal Detector

This c irc uit uses a single coil and nine c omponents to make a particularly
sensitive low-cost metal loc ator. It w orks on the principle of a beat frequenc y
oscillator (BFO). The c irc uit inc orporates two oscillators, both operating at
about 40kHz . The first, IC1a, is a standard CMOS osc illator with its frequency
adjustable via VR1. The frequency of the second, IC1b, is highly dependent
on the inductance of coil L1, so that its frequency shifts in the presence of
metal. L1 is 70 turns of 0.315mm enamelled copper wire wound on a 120mm
diameter former. The Faraday s hield is made of aluminium foil, w hich is wound
around all but about 10mm of the coil and connected to pin 4 of IC1b.

The tw o osc illator signals are m ix ed through IC1c, to create a beat note.
IC1d and IC1c drive the piezo sounder in push-pull fas hion, thereby boosting
the output. U nlike many other metal locators of its kind, this loc ator is partic ularly
easy to tune. Around the midpoint setting of VR1, there will be a loud beat
frequency with a null point in the middle. The loc ator needs to be tuned to a
low frequency beat note to one or the other side of this null point. Depending
on which side is chosen, it will be sensitiv e to either ferrous or non-ferrous metals .

OP AMP METAL DETECTOR

The circuit above did not work with LM 348 and it was modified as shown below to work with LM348.
Do not use LM 324 - it does not w ork.
The circuit above work s v ery well and only uses 4 components and two c oils. It is the simplest self-contained circuit and the
cheapest to build. Adjust the two coils until a "motor-boat sound" is produced. T his will inc rease when a coil is passed ov er the coils .
The frequency of the two osc illators w as meas ured as 108kHz .
Checked and tes ted: 25-12-2013

Simple Oscillator / Pipe Locator

Som etimes the need aris es to c onstruct a really simple osc illator. This could hardly be simpler than the c irc uit show n here, which uses just
three components, and offers five separate frequencies. D ue to the reactance of the 100-µH inductor and the propagation delay of the
internal oscillator, oscillation is set up around 5M Hz. When this is div ided dow n, Stage 14 approac hes the frequenc y of 260Hz.
The Simple Oscillator / Pipe Locator draws around 7mA from a 9-12 V DC source.

Simple Oscillator/Pipe Locator Circ uit Diagram

From: Francisco The pipe detector using CD4060BE with 150uH inductor didn't seem very useful to me becaus e the beep tone only
changed when the inductor would actually touch m etal.

The inductor has to be as "open" as possible, meaning the coil has to be as large as poss ible with the flux being able to escape to the
surroundings. In other words, the flux-path cannot be a good quality magnetic path as any surrounding magnetic object will not change,
remov e, ups et, the magnetic flux. Air cores are the best as they are the most sens itive to disturbance by introduc ed m agnetic objects.
With this detec tor you will be detecting the c hange in frequency of a 260H z tone. Because the frequenc y is being divided by 14 stages of
division, the actual frequency-change at the coil will need to be many hundreds of cyc les and this makes this type of detector very
insens itive.
All the other detectors compare the frequency of the coil with a fixed frequenc y and as soon as the frequency s hifts by one c ycle per s ec ond,
the result is pass ed to the output. Thus they are over 1,000 times more sensitive. T his circuit shows how NOT TO D ESIGN a metal detector.

TDA 0161 METAL DETECTOR IC

This IC is us eless. It is expensive and performs v ery poorly. It will detec t a s mall coin at 10mm.
Don't waste y our time and money on this circuit. I built it to prov e it is worthless.

Here are s ome c irc uits that perform better than a TDA 0161 IC:
This c irc uit is our Metal Detector MkII circ uit:
Buy a kit: Metal Detect or kit MkII  $15.00 plus postage
All the components fit on the PC board w ith the coils and speaker on short leads

Buy a kit for Metal Det ector kit MkII. $15.00 plus $6.50 postage.

Nail Finder
Kits for Metal Detector ki t - Nail F inder $17.00 plus postage
The Metal Detector k it also comes as a Nail Finder.
A Nail Finder head is connected to the Metal Detector PCB as shown in the follow ing circuit:

Connecting the Nail Finder head to the PCB

 The Nail Finder head

Kits for Metal Detector ki t - Nail F inder $17.00 plus postage

This is a simpler vers ion of Metal Detec tor MkII with LED readout. It us es a ferrite rod with 120 turns and 43 turns .
But it is not as sensitive because you are only altering the magnetic flux ON THE OUTSIDE of the coil, whereas the
magnetic flux on the INSIDE of the coil is more-conc entrated and more-sensitiv e to disturbance.

555 METAL DETECTOR

Here is a Metal Detector us ing two 555 IC's:
The large coils m ay make the circuit worth building. The author s ay s: The metal detector has a good sensitivity, it is
able to detect a coin at a depth of up to fifteen centim etres, larger objects up to eighty centimetres.
Here is the website:
https://homemade.metaldetectorsforgold.net/Simple_metal_detector_two_chips_NE_555.htm l

And here is the 4MB .zip file with all the images:
https://www.metaldetectorsforgold.net/w p-content/uploads/Simple-metal-detector-on-two-chips -NE-555.z ip

The author described the c irc uit inc orrectly and I will provide the correc t description.
The circuit is no more than a single coil being pulsed from a 555 running at a frequenc y determined by the 10n and 100k res is tor.
This frequency is "pumped" into the firs t c oil.
The second coil merely taps off a s mall amount of the signal due to the overlapping of the c oils. The same effect could have been prov ided
by a 10n capacitor connected to the top of the first coil.
When the coil detects a metal objec t, the frequenc y changes and this is amplified by the transistor. The amplitude of the signal also reduces.
The second 555 is set with the input at mid-rail, so that when the signal either c hanges frequenc y or reduces in amplitude, the effect is
passed to the speaker.
The first 555 is performing the task of a single transistor os cillator but the sec ond 555 is prov iding cons iderable amplific ation, equal to 3 or 4
transistor stages. But 555 IC's are only 10 cents and the whole circuit is very cheap to build.

Talking Electronics has a matrix board for tw o 555 chips and a pot. You can us e this prototyping board for this project and it will make
construction fairly easy.

You have to look carefully at every circuit y ou find on the web and s ee w hat the stages are doing and work out if a simpler design can be
produced. Prove this by mak ing the 2-coil circuit and then removing the second c oil. The only way to k now if a circuit is good or bad is to
make lots of them.

METAL DETECTOR WITH PRINTED COIL

Here is a metal detector kit with a rectangular coil made by etching a continuous track on a printed c irc uit board. The
track starts at the outside of the top of one s ide of a double-sided PC board and makes loops (square loops) towards
the centre, A v ia (a feed-through from one s ide of the board to the other side) in the centre of the board takes the trac k
to the other side of the board and the same set of track s are on the underside.
This produces about 50 turns . It does not matter if the loops are round or square - the result is about the same, but the
whole concept is much less effec tive than 50 turns of wire - as y ou s ee in the discuss ion below.
The kit is very nicely made and c omes with a plas tic c as e.
It is supplied by IC STATION:
http://www.ic station.com/product_info.php?produc ts_id=3750#.VKU-c clTSE4
and cos ts about $5.00 posted.

BUT the printed circuit coil does not produce a very s ensitive "detection."
If you have ever wanted to know if a printed circ uit coil works as a detector for MAGNETIC DISRUPTION, build the kit.
The magnetic flux is widely distributed ov er a very large area and not very much flux is near the object being detected.
That's why it is so INSENSITIVE.
It is just able to pic k up a nail in the centre of the coil.

The clever part of the c ircuit is the s ingle coil. The single c oil is actually TWO coils with a tapping at one turn. You have
to look v ery carefully at the board to s ee the c oil is tapped at one turn and this turn is actually producing an out-of-
phase output (called feedbac k). The circuit WILL NOT WORK if this feature is not included.
The single turn is taken to the base of the first transistor via a 2n2 c apacitor. The value of the capacitor is not
important. It simply c onnec ts the signal to the base of the trans is tor. The number of turns are not important either. You
can add any number of turns and see how many turns are needed to produce a signal to activate the firs t transistor. In
this case a single turn will produce sufficient amplitude as the base is turned on the by the 220k and the signal just has
to add to this voltage.
What this means is this: The signal emerging from the single turn is out-of phase with the output of the other end of the
coil. In other words, the signal is dropping when the other end is rising. Or, to put it another way, it is rising w hen the
other end is falling.
This means the signal can be used to turn on the firs t transistor to create POSITIVE FEEDBACK. T his feedback turns
the first transistor ON more and more and this provides the pulse of energy to feed the tuned c ircuit.
This arrangement replac es two s eparate coils . To ex plain how the circuit works, we can use two separate coils :
On the diagram, a second coil (blue) has been added. Remove the link to L1 and connect the coil to C2. is picked up
by the blue coil and passed to the base of the first transistor.
The first transistor amplifies this s ignal and passes it to the tuned circuit to amplify the effect by prov iding a pulse of
energy at ex ac tly the right tim e to increase the amplitude of the waveform .
The gain of the first trans is tor is adjus ted by varying the resis tance on the emitter so the whole c irc uit is just at the
point of oscillating.
When an object is placed in the centre of the c oil, some of the flux is passed into the metal object and is lost. This
means the blue coil does not see the same amplitude and the first trans is tor cannot provide the second c oil with
enough energy to keep it osc illating.
The feedback winding produc es a 10mV s inewave @ 1M Hz when the circuit is at the critic al point of being "upset" and
when a metal object is added, the waveform c hanges and the whole circuit stops os cillating (actually reduces in
amplitude) and the s ec ond transistor is turned off slightly. T his allows the 2k 2 resis tor to pull the base of trans istor 3
towards the 0v rail and turn ON the buzzer. The problem is this: only a few turns are near the m etal objec t and this
makes the c irc uit very insens itive.
However we can remove L2 and link C2 to L1 and the circ uit will work exactly the SAME - w ith ON E tapped
WIN DING. T his is the SKILL of seeing how a circuit works and being able to simplify it.
MODIFYING THE CIRCUIT
The circuit c an be made muc h more s ensitive by replacing the printed circuit coil with 50 turns of wire (0.25mm),
70mm diameter for c oil L1 and 5 turns of the s ame wire for c oil L2.
The 100R mini trim pot needs to replaced with 1k mini trim pot.
Turn the circuit ON and the buzzer will s queal.
Plac e the two coils together and the buzzer will gradually stop.
If not, turn the 5-tun coil over and adjust the mini trim pot until the buz zer stops.
The circuit is very sensitive to very small changes in the resistance of the mini trim pot but y ou c an get it to s top
squealing and it will detect a coin at about 5cm when near the centre of the coil.
This is much more s ensitiv e than the original design.
Alternatively, you can create a single TAPPED WIND ING.
Start by winding 50 turns and make a knot in the s tart-wire s o you know the start of the "transformer."
Make a loop (c alled a "tapping") and wind 5 more turns. Sticky -tape the turns together.
You now know the start, tap and end of the c oil.
Here is the diagram for adding the coil:
The circuit is not as sensitive as METAL DETECTOR MkII but gives a good platform for experimenting and
understanding oscillators , feedback and tuned circuits. You can compare the effectiveness of a PC B coil with a wire-
wound c oil.
It is pointless learning the theory about these circuits until you have built at least 10 different designs because the
effectiveness works on a principle called "natural effectiveness" or "natural gain" or "natural activ ity" and we call this
"Q" or "Quality factor" and this is the ability of a circuit to produc e a very high amplitude from a very small supply
voltage.
This amplitude must also be produc ed from very small "pac kets of energy " and this means any piece of magnetic
material entering the "field" will reduc e the amplitude. This important fact has NEVER been mentioned in any tex t book
ANYWH ERE. It is the absolute BASIS of metal detection.
Here's the main reason why this c irc uit is not v ery sens itive.
The amplitude of the signal cannot ris e above 600mV (in one direc tion) because this is the maximum between emitter
and base of the PNP transistor. T his allows the signal to produce a peak-to-peak waveform of 1200mV. And that's
what is does. The single-turn feedbac k winding produces a maximum of about 20-24mV and the (printed c irc uit -
printed c ircuit board) coil has a total of 50 turns.
This produces very little electro-magnetic radiation and the c irc uit is not v ery sens itive.
Another reas on for the insensitivity is the way the coil is cons truc ted.
The coil is c alled an AUTO TR ANSFORMER and this means it is a transformer with a single winding that is tapped.
It works exactly the s ame as a transformer with two windings. A 50-turn winding and a 5-turn winding.
This type of transformer has a feature c alled TURNS RAT IO. The ratio is 10:1 and the output of the 5-tuns is one-tenth
the voltage on the main winding.
This type of transformer is normally used as a C URREN T transformer because the current out the 5-turn w inding c an
be 10 times the current supplied to the main winding.
But we are using it as a v oltage winding. But this winding is only 10% as sens itive as the main winding.
If the voltage drops by 10mV on the main winding, the "feedback winding" only sees 1m V. In other words it is not a
very s ensitiv e way to detect the presence of a m etal objec t.
So, you can see . . . this circuit has a number of faults and limitations that make it a poor design.
The Metal Detector MkII k it has a ratio of 50:70 and is much more s ensitiv e.
Where does all this "s ignal" or "noise" or "activity" come from ? What makes the "signal" in the first plac e?
The signal or waveform produced in the coil comes from the natural ability of the coil and 2n2 (acros s it), producing a
waveform without any other components.
That's right. Whenever a coil (called an INDUCTOR) and a c apacitor (acros s it) receive a short pulse of energy . . .
and then there is nothing connected to the lower terminal of the combination, the two components will produce a
sinewav e signal. T he two components are called a TU NED CIRCU IT (parallel tuned c ircuit) or R ESONANT CIRCUIT
(a circuit c apable of resonating) or NAT URAL OSCILLATORY CIR CUIT and that is exactly what happens when you
turn the project ON and "pulse" the two c omponents . The only other component that "boosts " or helps these two
components produce a s inewave is the first transistor. BUT this transistor D OES NOT PRODUC E the wav eform. It just
supplies the pulse of energy at the correc t instant in each c ycle.
The first transistor turns ON at the beginning of each cyc le and supplies the pulse of energy and then turns OFF. The
tw o components then produc e the smooth sinewave.
The transistor is turned ON by the s ingle-turn or 5-turn feedbac k winding. T hat's how the 1MHz oscillator works .

HOW DOES T HE OSCILL ATOR KEEP OSCILLATIN G?

The first transistor is turned on very lightly by the 220k bas e resistor. This puts a s mall c urrent through the coil and
expanding flux is produced. T his is the same as INCREASIN G F LUX. This flux cuts the turn of the feedback winding
and a small voltage is produc ed . . . . along with a c urrent . . .that adds to the current s upplied by the 220k. This turns
the transistor ON more and the process increas es and increas es until the trans istor is fully turned ON.
At this point the trans is tor is fully turned ON and the flux is a max imum but it is not ex panding (or INCREASING) flux
and the flux does not cut the turn of the feedback winding and does not produce a voltage or c urrent. T hus the voltage
on the feedback winding suddenly stops and the trans istor is s uddenly turned off to the weak condition provided by the
220k.
The flux produced by the c oil starts to collapse and produc e a voltage in the feedback winding that is OPPOSITE
POLARITY and now the feedback s ignal counteracts the effect of the 220k and turns the transis tor OF F fully.
The flux continues to c ollapse and it pass es all its energy to the 2n2 capacitor. When the signal can no-longer act
against the 220k, the 220k starts to turn the transistor ON a s mall amount and the c ycle starts again. All this is
happening at one million times per second.

HOW MUCH DID YOU LEARN?

What are the three m ain faults with this circ uit?
Ans wers at end of article

MICRONTA DELUXE METAL DETECTOR

Here is a very good metal detector from Micronta. The secret of its sensitiv ity is the construction of the search coil.
Unfortunately I do not have any details of this and the circuit has been inc luded for viewing only.

Click HERE for larger image

Here is anther vers ion of the metal detector us ing a coil that is printed on the fibre-glass PC board.
The circuit is not very s ensitiv e. The only reason I can give is the des ign of the circuit.
The PCB coil was replaced by a wound coil of 30 turns , 70mm diameter, and one turn for the feedback .
As the s ingle turn is moved closer to the 25 turns, the ac tive buzzer stops s quealing and the LED turns off.
It is only good for detec ting large metal objects at a s hort distance. The bes t result was a 25 c ent coin at about 20mm. This is a very poor
result.
The coil on the PCB is 26 turns on one side of the board and 26 turns on the unders ide, mak ing a total of 56 turns . The inductance was
measured at 0.12mH. The "feedback" winding is jus t one turn. It must be connected around the correct way. It must deliver a signal that
increases the noise produced by the transistor to produc e an osc illator. This osc illator has a certain amplitude and y ou control the amplitude
by the 1k pot in the emitter. This amplitude is pass ed to the second transistor where it is increased about 200 times and effectiv ely prevents
the 100n c apacitor being charged via the 2k resistor.
This k eeps the third transistor OFF and the speaker does not produce a noise. When a metal object is detected by the coil, the amplitude of
the magnetic waves from the 26 turns are reduced slightly and the amplitude of the oscillator is reduced. This means the second transistor
does not keep disc harging the 100n to the same extent and it c harges a small amount v ia the 2k resistor. This puts a small voltage across
the base and emitter of the third trans istor and it turns ON slightly. The speak er is actually an ac tive buz zer and it will produce a tone when a
small DC voltage is applied to its terminals.

This c irc uit has about the s ame s ensitiv ity w hen the coil is wound with 0.25mm wire and its effectiveness could not be improv ed. It is the type
of osc illator used in this design that makes the product not very s ensitiv e.
Here is a discussion on how a number of oscillators work and the surprising
conclusion that one of the oscillators produces results better than all the others:

METAL DETECTOR OSCILLATORS

Generally, an oscillator is designed to produce a STA BLE OU TPUT .
We don't w ant an osc illator to drift - change frequency - but for Metal Detection, we want the oscillator to be sensitive to outs ide influence.
This m akes it v ery SENSITIVE.
These oscillators hav e the coil ex posed to the surroundings and the turns have a very large diameter.
This type of oscillator is used to detect the presence of a m etal objec t. The metal will alter the inductance of the c oil by absorbing some of the
magnetic flux and inc rease or dec rease the frequency of the circuit, depending on the type of metal.
There are two types of metal - as far as magnetism is conc erned. Magnetic and non-magnetic.
But when an osc illating magnetic field is present, another feature is present.
All metals that conduct electricity will produce eddy currents when exposed to an os cillating magnetic field and this will absorb some of the
magnetic flux.
This w ill increase the inductance of the coil and reduce the frequency.
Most of the oscillator circuits are Colpitts and can consist of 5 to more than 15 components.
The whole idea of the c ircuit is to mak e it stable when "sitting around" but allow the slightest magnetic influenc e to change t he frequency.
This is called the SENSITIVITY of the circuit and we will be providing a number of circuits for you to test and compare their sensitivity.

These circuits operate betw een 100kHz and 500kHz and to detect when they shift by as little as 1Hz , we use a stable reference osc illator.
The oscillator is an AM radio. It has an inbuilt 455kHz os cillator and when placed near a Metal Detecting Oscillator, it picks up the signal, (just
like a radio s tation) and produces a whistle. This whistle represents the difference in frequenc y of the two oscillators.
It is called a BEAT-FREQUENCY or Difference Frequenc y.
By tuning ac ross the band you will get a lowering in frequency until the whis tle stops. At this point the two oscillators are t he s ame frequency
or one is a multiple of the other and the difference produces a zero- frequency beat. If a coin is brought near the detecting c oil, the change in
frequency will be heard.
Even though this arrangement is very simple, it is equal to all the BFO (Beat Frequency Osc illator) Gold detectors on the market.

The circuit is a Colpitts Oscillator and is placed near

an AM radio to produce a whistle.
The circuit operates at about 500k Hz.
The radio is tuned so the whistle is reduced to zero.
When a metal object is placed near the coil, the
frequenc y of the transm itter is reduc ed and a whistle
is heard in the radio.
Com pare this c ircuit will the others to determine the
mos t s ensitiv e circuit.

Note: 4n7 between c ollector and emitter (in place of

the 1n) produces a better signal - easier to find on AM
radio.
The thicknes s of the wire makes very very very little
difference. But the number of turns and the diameter of
the coil makes a very BIG differenc e to the actual
frequenc y at which the circuit will oscillator.
In this circuit the frequency of operation does not
matter as it will only change the pos ition on the dial.

In this circuit the coil is not

driven directly but receives a
very small amount of energy
from the feedback provided by
the signal on the emitter of the
transis tor.
The coil and the two 1n
capacitors (in s eries) produce
a TUNED CIRCUIT and will
produc e natural os cillations.
The transistor sees this and
produc es the same waveform
on the emitter but with a greater
amount of "strength."
This "strength" is not needed by
the Tank Circuit but it need s a
signal and the signal from the
emitter is s ufficient to maintain
This c ircuit is a good performer. Current 4m A. Easy to find os cillation.
signal on AM radio. The circuit operates at about 1MHz and
the beat-frequency can be found on the dial at 1,000kHz .

By k eeping the s ize of the s earch

coil nearly the same in all c irc uits,
we c an determine if the
arrangement of the components
improv es the sensitivity.
Circuits 2 and 3 are SH UNT-FED
Colpitts Os cillators.
In a Shunt-fed os cillator, DC is
not passed through the Tank or
Oscillator section.

This is a v ery simple circ uit produced by a

reader. See if it works.

Circ uit does not work with 10k. Reduce

10k to 3k3 and circ uit keeps changing
frequenc y. Very hard to find on AM radio.

This circuit is the same as circuit 1 but the

TUNING C APACITOR across the coil is
tapped and the tapping (feedback ) is
pass ed to the emitter.
 This circuit is the same as circuit 5 but the
emitter res is tor is reduc ed and the
transis tor takes more current. See if this
improves the s ensitiv ity.

Ev ery component in this circuit is the

wrong value.
Build the circuit and see if it works.

This c ircuit is the best performer. C urrent 2mA. Eas y to find the signal on AM radio. T he c ircuit
operates at about 1MHz and the beat-frequency can be found on the dial at 1,000kHz .
The circuit is very similar to #2 but this time the feedback signal from the c ollector has a larger
amplitude and the voltage across the c oil is about 15v p-p. T his produc es a very clean and clear
signal on an AM radio and is the bes t performer of all the circuits.

The Tank Circuit operates in a very unusual way.

The tapping (between the two capacitors) is connected to 0v and does not move.
When the magnetic flux in the coil collapses, it charges both 1n capacitors in series and
although it does not charge both equally, the bottom terminal of the coil drops downwards easily
bec ause the base-emitter junction of the trans is tor is seeing a reverse-voltage and does not
offer any opposition.
This turns the transistor OFF and the top 1n s tarts to charge via the 4k7 resistor. The top
terminal of the coil rises due to the charging of the top 1n and as the voltage produc ed by the
coil reduces, the voltage on both ends of the 100n s tarts to rise and turn ON the transistor.
The voltage across the base-emitter junc tion cannot rise abov e 0.7v, so the transistor turning
ON will remov e the voltage across the top 1n

Changing the 100n "pic k-off" to 10n makes no difference s o the value of this capacitor is not
important. It simply connec ts the tank circuit to the trans is tor and does not affect the timing of
the circuit.

METAL DETECTOR OSCILLATORS - CONCLUSION

The bes t circuit is #8 where the Tank Cir cuit (the self-os cillating circuit - the coil and c apacitors)
will produce a good waveform (large amplitude) that is easy to pick-up on an AM radio at about
1,000kHz .
Don't forget, this circuit is a special type of oscillator that is stable in operation but is easily
"ups et" by the c hange induc tanc e of the coil.
This means the Tank Circuit has to be lightly loaded and although it is connected to the base via
a 100n capacitor, the trans is tor is providing a feedback signal that is compensating for the
losses deliv ered to the base.
The emitter waveform was measured on a CRO as 15v p-p for 9v supply. The transistor is a
common-emitter amplifier and the circuit is a SHUNT-FED Colpitts Oscillator bec ause no DC is
pas sed through the Tank (os cillator) section.
These 8 circuits are a perfec t example for an ex periment to show you cannot predict how a
circuit will w ork and which circuit w ill perform the best.
No simulation-software package will show the enormous difference between the circ uits. That's
why y ou have to build EVERYTHING and OBSERVE the results yourself.
The secret to getting a TANK CIRCUIT (more-accurately . . . a RESONANT CIRCUIT) to
produce a good sinewav e is shown in circuit 8.
The reason why a Res onant Circuit will produce a v oltage up to tw ic e rail-voltage is due to the
fact that the induc tor produces a voltage in the reverse direc tion when it collapses.

FINAL DESIGN FOR A METAL DETECTOR

We now want to produce the s implest and BEST metal detector.
We will be us ing a microcontroller to take the signal from the detecting c ircuit and analys e the
waveform to produce a beep w hen meal is detected.
If you look at all the Metal Detector c irc uits, y ou w ill find the complexity of the c irc uit rev olves
around detecting the s mall c hange in frequency or amplitude of the detecting circuit. This can
take as many as 8 to 15 chips in the older c ircuits as each chip only performs v ery small task.
When using a microcontroller, all this can be done with an 8-pin device and the program does all
the work
We have two choices for the detecting circuit.
The Metal Detector MkII front-end changes amplitude when metal is detected and the remainder
of the c irc uit amplifies this change and produces a tone.
Circuit 8 produces a c hange in frequency.
If we want to detect the firs t sign of c hange-in-frequency, we need to have two osc illator
operating at exactly the same frequenc y and detect the difference.
Alternatively y ou need to count the cyc les for one sec ond and detect the change. This means a
reading will take 1 second to get a result.
Alternatively y ou c an use the Metal D etector MkII circuit and detec t the immediate change in
output voltage. Howev er this circuit has a "s weet spot" and needs to be constantly re-adjusted to
produce the point where it is sensitive to "interferenc e." This adjus tment is diffic ult to do with a
microcontroller.
The first three transistors of the Metal Detector MkII circuit will produce a change is about 1m V
and this can be detected by the Analogue channel in the microcontroller - providing the circ uit
can be automatically adjus ted.
All Metal Detector circuits have been designed along the wrong lines. They hav e a fixed
reference oscillator and detect the change in frequenc y of the scanning coil. But the frequency of
this c oil is changing all the time due to natural drift in the osc illator and the setting of the Metal
Detector has to be c onstantly re-adjusted s o the output is zero and no tone is produced by the
speak er when nothing is being detected.
It is m uch m ore clever to adjus t the reference osc illator on a constant basis and thereby produce
silenc e when not detecting and when something is detected, the two osc illators w ill differ by a
small amount.
But our clev er design goes further.
When y ou s weep over a nugget of gold and then sw eep bac k again, TWO HIGH-LOW PUL SES
are produced within 3 seconds . This is what the mic rocontroller detects and interprets as a
"sample."
In this way we have eliminated 95% of all fals e triggering.
It is s urprising no-one has thought of this concept before.
The final design is in the process of being developed.

6-1-2018

Answers:
1. The PCB coil does not concentrate the flux and only a few of the turns are being affected fully
by the presenc e of the piec e of magnetic material and the other turns are being effect less and
less (as they are further from the object being detected).
2. The base-emitter junction across the coil puts a load on the coil and reduces the amplitude to
about 1200m V.
3. The s ingle turn feedback w inding only deliv ers 1:50 of the change in output of the c oil.

When designing a project, you don't know what WILL work but we have explained what will NOT
WORK.
We hav e covered some of the features of BAD design.
The operation of the "pick-up head" is much more complex than meets the eye.