plik

The "Blind Squirrel" Pulsed Induction Metal Detector

This is my version of a Pulsed Induction type of Metal Detector. I named it "Blind Squirrel" because
even a blind squirrel finds an acorn every now and then.

This project came about when it was suggested by our club for the 2005-2006 winter contest.

After studying the various types of metal detectors and circuits that were available on the Web, I
decided to design my own using ideas from previous designs as well as my own ideas.

The Pulsed Induction (PI) detectors seem to have the advantage of more depth, at least that's what
most folks were saying, so I set out to design another PI detector for the contest.

Most of the PI detectors work the same way:

1. A high current, short duration pulse is repeatedly applied to the search coil at a fixed

frequency which is provided by a timing circuit.

2. Either another coil is used as the receiver or, most commonly, the same coil is used as both

transmitter and receiver.

3. Immediately after the pulse is applied, the receiver coil detects a faint signal which is

produced when the magnetic field in the coil collapses. This collapse or decay is lengthen in
duration when the search coil is near metal objects due to the eddy currents they produce.

4. This weak received signal must be amplified many times in order to detect the very small

changes in the decay length. Since there is also a very large kick-back pulse present, the
weak decay signal must be separated with a sampling circuit which is gated by the same
timing circuit.

5. Once the received signal is isolated, it is further amplified and converted to a relative voltage

which is in proportion to the size and/or proximity of the metal being detected.

6. This voltage is then used to drive an audio and/or visual indicator such as a meter and

speaker.

7. A source of regulated power is necessary because of the high gains and offsets that are

involved.

One of the hardest parts of this design is amplifying the weak decay signal when the kick-back pulse
is present. In just about all designs I have seen, an op amp with a high slew rate and wide bandwidth
is used. Because of the big offset at the input, two power supplies, plus and minus, are needed on
the op amp IC. This is usually accomplished by using a voltage inverter IC to produce the negative
voltage or sometimes a booster is used with a floating ground reference in order to bias the IC
correctly.

One of my goals was to eliminate the extra supply. I wanted my detector to operate on a single
supply with a common ground connection.

Another goal was to simplify the timing circuit. Here, you want to have a stable circuit as far as
frequency and pulse width and it must be simple.  Some of the published circuits simply use a R/C
oscillator with a trim pot to set the frequency.  Some use a microprocessor which makes it simple
and stable, but hard to change frequencies.  Being able to change frequencies makes it easy to
experiment with different search coils and timing schemes.

I believe I have met the goals with the following design, however, I want to emphasize the fact that
this is still experimental and I can't guarantee the results you may have. I am presenting it here as
an alternative to the circuits that are currently available. I also encourage further refinement.
Please play with it and see if you can suggest improvements or refinements. Please share them with
me so I can post them here. I would really like to see pictures of your detector too.

The Schematic

As you can see, there is only one regulated 12 volt supply operating the entire circuit. I was able to
do this simply by AC coupling the received signal to U3 via C10. This allowed me to adjust the offset
within an operating window. By using a tantalum capacitor of reasonably high capacity, the signal
passes with no problems. See Waveforms and Adjustments. This met my first goal.

The regulator IC is a low dropout type capable of handling 3 amp. loads. This is necessary because
the peak current of the TX pulse into the search coil is near 3 amps. and I wanted the circuit to still
operate when the battery dropped to 13.5 volts. With the timing I am currently using, this IC doesn't
get warm. However, the range of the pots and jumpers allows you to adjust the circuit to a point
where Q1 and VR1 will warm up, so I included room for heat sinks on the circuit board so you can
experiment without worrying about damaging these parts.

B1 consists of four 3.6 volt Li-ion batteries in series making up a 16.8 volt source when fully charged.
These were salvaged from old Sony cell phone battery packs. The average current drawn by PI
detectors is usually high because of the power needed to drive the search coil.  This circuit draws
less than 200 mA. which gives over two hours operation between charges with the cells I used.  You
can use Ni-Cds or Ni-MH but just remember to use enough cells to provide at least 15 volts into the
regulator.  Sealed lead acid batteries would work but they would be pretty heavy, perhaps a belt-
worn battery pack.  I haven't tried this circuit without a regulator.  Maybe someone out there will do
some experimenting in this area. Let me know what you come up with for power.

The frequency of operation on a PI detector varies from design to design and from coil to coil on any
given design. Basically, the higher the frequency, the more power is consumed, so you have to
decrease the TX pulse width to compensate until you reach a point in frequency where the pulse
gets too short and performance starts to drop off. The same thing happens on the low side of the
frequency range, but here, the pulse gets too long to still work and ends up just consuming more
power than necessary. I have found by experimenting, that a good compromise is between 75 and
225 Hz. with a pulse width of about 100 to 175 microseconds in duration. Still, this is with the coil I
am using.

With the jumper setting "110" in the schematic, the frequency out is 109.2390442 Hz. which works
well with 125 microseconds for the TX pulse width which is set by R2.

You can select different frequencies in multiples of 2 by changing the binary code at A,B and C
inputs. The following table shows the frequencies possible with a 3.579545 mHz. crystal:

C   B   A     Frequency
0   0   0   6,991.298828 Hz.
0   0   1   3,495.649414 Hz.
0   1   0   1,747.824707 Hz.
0   1   1      873.91235 Hz.
1   0   0      436.956176 Hz.
1   0   1      218.4780884 Hz.
1   1   0      109.2390442 Hz.
1   1   1       54.61952209 Hz.

For more information on this, see the data sheet on the 4536B. You might want to experiment in
this area too. Let me know what you find.

U2 takes the TX pulse and derives the RX pulse and delay which drives U4 the sample circuit.

The sample then goes to U5A which is a sample and hold circuit which obtains an average DC voltage
from the samples. U5 is an all-CMOS device that has an output which can swing rail-to-rail. Last
time I checked, it was available from Newark, part number 57K3887. If you have trouble getting this
part, let me know, I have several of them that I am selling for $ 1.00 each plus postage. I had a
sensitivity control (R14) on the first version I built, but I decided I didn't need it and later replace it
with a jumper. Also, I changed R13 to 10 K. U5B then further amplifies the voltage to obtain a 0 to
12 volt swing which tracks the distance and/or size of the metal object. This DC voltage drives M1
and OC1, an opto-coupler which provides a varying bias to U6 a 555. The 555 produces a ticking
sound when an object is first detected then turns into a higher pitched squeal as the object gets
closer to the coil. The threshold control is set to a point where the ticking just starts.

The schematic shows a speaker for the audio. You can used a piezo electric transducer in it's place
and eliminate R26. Also, I installed a headphone jack on the output. This helps in noisy
environments.

Parts List

B1    3.6 v Lithium cell x 4 for a fully charge voltage of 16.8 volts.
C1    470 uF 25v
C2    100 uF 16v
C3    27 pF disc ceramic
C4    .1 uF mono ceramic
C5    .01 uF mylar
C6    .1 uF mono ceramic
C7    .0033 uF mylar
C8    .0033 uF mylar
C9    1 uF mono ceramic
C10    4.7 uF tantalum (or other low leakage type, if space permits)
C11    4.7 pF disc ceramic
C12    1 uF mono ceramic
C13    4.7 uF tantalum
C14    .001 uF mylar
C15    .01 uF mylar
C16    .1 uF mono ceramic
C17    .01 uF mylar
C18    1 uF mono ceramic
C19    .01 uF mylar
C20    .01 uF mylar
C21    .1 uF mono ceramic
C22    470 uF 16v
D1,D2,D3,D4  1N4148

L1    Search coil, 27 turns 9.5 inches O.D.# 20 enameled wire flat spiral wound.

M1    50 uA DC Ammeter
OC1      4N35 Opto Coupler
Q1    IRF740
Q2    MPSA70 or any general purpose PNP transistor with a beta of 40 or higher.
R1    3.3k
R2    25k pot.
R3    1.5k
R4    25k pot.
R5    1.5k
R6    25k pot.
R7    470 ½ watt
R8    470
R9    220
R10    1 meg
R11    100k pot.
R12    10k
R13    1k (if using SEN. pot.) otherwise 10k to 15k

Circuit Board Layout

This is the bottom view as a completed board would look, not a mirror image as used by the transfer
method of making pc boards. The red lines indicate three jumpers that must be installed first when
assembling the board because these are located on the top side under IC's.

This is for reference only, should you want to use the same general layout. Everything is to scale
except the holes for IC's, resistors, capacitors and diodes may be too large. I have the holes larger
on this drawing because I use a pen plotter with an etch resist pen to make my circuit boards and
this size allows for the extra width from the flow of the ink.

The layout for this circuit is critical to power and return ground paths. The layout presented here
has been refined and does work, however, there is always room for improvement. Let me know if
you have a better way. I haven't considered surface mount, but it should be easy to implement.