# Search coil design

Introduction

In radio engineering and metal detectors technology there is such a thing: connected circuits. They come with inductive or transformer coupling. We are interested in the first. The transmitter circuit (it receives power from the generator) is called the primary circuit. Assign index 1 to it ( L1, I1, Ф1). The receiver circuit (it receives energy from the primary circuit) is called the secondary circuit. Assign index 2 to it ( L 2 , I 2 , Ф2).

The principle of inductive coupling is that the current of the primary circuit I1, passing through the coil L1, creates a magnetic field F1 around it, which excites the emf in the coil L2, and, therefore, the current I2. Thus, with inductive coupling, energy is transferred from one circuit to another by a magnetic field. If the coils are separated, then their relationship is described using the coupling coefficient Ksv. The coupling coefficient shows what proportion of the total magnetic flux F1 of the coil L1 is the magnetic flux Ff, penetrating both coils. If we simplify the Maxwell-Faraday equation to the limit, we get:

Ф1=( L1*I1)/w1     Ф 2 =( L2*I2)/w2 ,

where w1, w2 are the number of transmitter and receiver turns

KCB u003d m / sqrt (L1 * L2) ,

where m u003d m12 u003d m21 – the coefficient of mutual inductance of the coils of the transmitter L1 and receiver L2

In the secondary circuit with inductive coupling, as a rule, voltage resonance is obtained. This is due to the fact that the L2 coil itself works as a generator in the secondary circuit. It is included in the circuit in series, which means that there will be voltage resonance in the circuit.
The secondary circuit during tuning somewhat affects the primary and violates its resonance. In general, any change in the setting of one of the circuits affects the other circuit (changes its setting). It is necessary to adjust the primary circuit each time in order to restore resonance after selecting the capacitance C2 or the number of turns N2. If the secondary circuit (receiver) is tuned to the frequency of the primary circuit (transmitter), then it introduces only active resistance into the primary circuit, which is the greater, the stronger the connection. The value of this resistance will determine the losses in the primary circuit (the transfer of a certain amount of energy to the secondary). If the secondary circuit is not tuned to the frequency of the first, then reactive resistance will also be added to the active resistance (inductive when detuning up in frequency or capacitive when detuning down in frequency).

If you remove the frequency response, then its shape will depend on the value of the SW, as in the figure below.

The weaker the bond, the sharper the resonance. With an increase in Kv, the top becomes first flat (critical connection), and then double-humped.

A critical or strong connection (with a small gap between the humps) significantly expands the bandwidth of the receiving path. Strong coupling is characterized by the transfer of energy from the primary to the secondary circuit with an efficiency above 50%. A weak connection has a lower efficiency, but the secondary circuit has almost no effect on the primary. Therefore, strong coupling is used in powerful transmitters, and weak coupling in measuring instruments. We are interested in a strong connection, because with a weak connection, the sensitivity of the sensor will be minimal.

It seems to me that White’s sensors use strong coupling. In this case, the receiver coil is affected by the main field Ф TR and the field of the opposite direction Ф FB . By winding the turns of the FB (feedback) coil, we balance the coil. With strong coupling, the power transfer efficiency is > 50%, which means that the power in the receiver (secondary circuit) is higher than that lost in communication in the transmitter (primary circuit). This allows you to increase the sensitivity of the coil compared to weak coupling and obtain a linear frequency response over a wide frequency range.

Since, with the correct setting of the coil, the phases of the signals of the receiver and transmitter must match, the coil setting is reduced to the selection of the resonant capacitance of the receiver. Tuning the transmitter to the operating frequency is also not the same as usual. It is necessary, by tuning the frequency of the generator within certain limits, to find the extreme frequencies of the hump for the critical connection and then set the average frequency. For a strong connection, you can try to find a dip in the middle of the hump.

Thus it turns out that the bandwidth depends on both the geometry of the coil and the number of turns in the coils. If we want to get a critical connection, then the difference in the diameters of the transmitter and receiver should be as large as possible, and the number of receiver turns should be as small as possible. In practice, we know that the diameters are related as 2:1…3:1. So when changing the diameters of the transmitter and receiver compared to proven sensors, one must be prepared for a lot of experimentation …

Very often one has to read in the description of the MD phrases like “… the resonance of the receiver coil is lower in frequency by xxx Hz relative to the resonance of the transmitter.” Or something similar, but the bottom line is that if the transmitter is tuned to resonance at a frequency of 6.8 kHz , then the receiver can be tuned to 4.5 kHz.

Knowing what coupled circuits are, let’s simulate the operation of White’s sensor . The sensor model is an unbalanced system consisting of two coils: a transmitter with a resonant frequency of 6.8 kHz and a receiver with a resonant frequency of 6.8 or 4.5 kHz. Capacitances of resonant capacitors can be selected: the parameter {T} for the transmitter and {R} for the receiver.

As you know, a parallel resonant circuit is powered by a current source. Here it is I1 , the maximum current is 50 mA. If we take into account that the duty cycle of the signal is 50%, and the current shape is part of a sinusoid, then the consumption of the sensor is on average 40% of the generator current. Then p=50*0.4=20 mA, which is true.

Now there is no proprietary sensor at hand, but there is a home-made one, here are its parameters:

LTR =442µg               LRC =16mg

RTR =1.2ohm               RRC =70ohm

WTRu003d 31 vit.               WRC =250vit.

DTR =196mm               DRC =106 mm

It so happened that at the time of the experiments I ran out of wire for winding the receiver coils, with a diameter of 0.31 mm. Only 0.12 mm (or even 0.1 mm) remained. Hence the increased winding resistance.

This is what should be obtained when changing K CB , but without changing the resonant capacitance of the receiver.

SW =0.01    SW =0.12     K SW =0.25     K SW =0.5

All this is good, but you need to determine which SW was used by White’s engineers . Considering that the magnetic flux is calculated by the formula Ф u003d (L * I) / W (where W is the number of turns), we can approximately calculate

SW =( U TR /U RC ) * (W TR *W RC ) = ( 400 mV / 41 0 mV) * (31 / 250) = 0.12

Now it is possible to simulate the pickup, while not taking into account the balance signal, and determine at what values ​​of C TR and RC there will be the greatest return and what, in fact, is the resonant frequency. This is a sensor with coils tuned to one resonant frequency, so to speak with “full” resonance.

RC 30.3 n C TR =1 22u  RC 35.8V

Do not be afraid of the number 35.8 V. It just shows the conditional sensitivity of the sensor, taking into account the quality factor, inductance and other parameters of the coil. If we apply the same voltage from the feedback coil, but in antiphase, we will get a zero output signal on the receiver coil.

If you continue the selection, you can get at a frequency of 7.7 kHz, with a capacitance of 24.2 n (20 + 2.2), a voltage of 45.7 V at the receiver output (blue diagram below). By the way, it is recommended to adjust the associated contours in this way – by successive approximation to the best result. Not bad, the return of the receiver has increased by 30%!

White’s resonant pickups

But still, let’s return to the sensor, the resonant frequency of the receiver of which is lower than the resonant frequency of the transmitter. Let’s change the value of C RC so that the resonant frequency of the coil decreases to 4.5 kHz.

RC 30.3 n C TR =1 22u  RC 7.1V

Too low sensitivity. It can be increased by increasing the number of turns of the receiver.

Let’s determine how many times it is necessary to increase the inductance of the receiver coil and how this will change K SV . Let’s increase, for example, the inductance twice, up to 32mG, then for a frequency of 4.5 kHz, the capacitance C RC 39 n . Now, knowing that SW = m / sqrt(L1*L2) , we can calculate the changed K SW :

SV u003d K SVS sqrt (L / L C ) u003d 0.12 * sqrt ( 32 / 16 ) u003d 0.17 , here the indices H and C denote the new and old value.

In this case, the number of turns will increase by the same factor, i.e. W=250* sqrt ( 32 /16 )=35 0 . The White’s coil has about the same number of turns . Here’s what happened:

It is easy to see that after optimization C TR =1.33u, C TR =39n is obtained and the resonant frequency has decreased to 6.6 kHz. However, the output voltage of the receiver increased to 13 V, almost 2 times. Still, such sensors are less sensitive than those with “full” resonance. I observed this repeatedly on the oscilloscope screen – I mean a decrease in sensitivity. True, the reaction of such a sensor to temperature changes is as much less. Maybe that’s what White’s wanted.

The graphs above show the effect of a resonant capacitor on the K SW  of the receiver and transmitter circuits. The green graph corresponds to RC =22n , the pink graph corresponds to RC = 27 n , and the phase shift of the receiver and transmitter is approximately 90 ° . This makes it possible to use a sensor with such a setting in a typical circuit, with the formation of LED control signals on comparators. In practice, this is a sensor with “full” resonance.

If we continue to increase the resonant capacitance, then K CB will increase, the resonant voltage will decrease. This is clearly seen in the diagram – the humps of the resonance seem to move apart. The phase shift between the transmitter and receiver signals also increases, and at a value of RC = 39 n will become close to 180 ° . If we now swap the control signals SD – drX, drY, then again we get a typical control scheme. At this point, for a receiver coil of about 380 turns, the resonant frequency will be about 4.5 kHz. Unfortunately, the output voltage on the coil at this characteristic point is the smallest. How it looks without logarithm can be seen in the right figure. Here the sharp decrease in sensitivity is more clearly seen.

On the oscilloscope, my sensitivity differed by 2 … 4 times, and not in favor of the White’s sensor . The simulation confirms this. But let’s not forget that this simulation took into account only the change in the primary coupling field between the transmitter and receiver, and did not take into account the influence of the balance field from the feedback coil. Although most of the information about the target is carried by the primary field. Perhaps balancing the receiver coil at least up to 50 mV, we get a slightly different picture. But a qualitative change, most likely, will not happen.

So why is this shift necessary? VDI is about the same, the soil effect is less, but proportional to the sensitivity. It is said that the effect of the dynamic influence of the soil is reduced. But what is it? I remember that George Payne first mentioned it, but in the context of phasing the sensor windings. And he did not bother to explain what exactly he meant by this term. The second mention refers to the AGEB circuitry of White’s 6000 Di Pro metal detector . But there, the balance voltage of the X amplifier is simply cunningly adjusted, slightly shifting the VDI value depending on the amount of unbalance in the channel and keeping its value for the soil close to the calculated one.

If we look at the graphs of the sensor with “full” resonance and White’s , we can only see that the resonant humps in the second case are further apart. And the sensor is tuned closer to the second resonant frequency. In the figure above, I deliberately increased the output voltage of White’s sensor by more than three times so that their tops matched. Yes, there are discrepancies.

Maybe this somehow affects the dynamic effect, but I have not yet found confirmation of this. And, by the way, Payne wrote in his notes that the dynamic effect of the influence of the soil is two to three orders of magnitude weaker than the static one.

In general, while the usefulness of the frequency separation is not clear to me …

How to choose the resonant frequency of the receiving circuit of the sensor?

Let’s go back to this issue again. But a little on the other side. As you know, if you adjust the receiving circuit of the sensor in resonance with the transmitter, then the sensor will have maximum sensitivity. For what purpose? Yes, to everyone, including the soil. This is also a goal, however, harmful. We need to get rid of it, but it doesn’t work out in any way …

Let’s do a simple experiment. To do this, take the standard sensor “Ring” and tune its transmitter to resonance. It turned out 6. 1 kHz (such a sensor was caught). Let’s achieve balance with a non-resonant transmitter coil – we got a zero output voltage and a zero (with respect to the transmitter) phase shift of the receiving coil. Everything is clear here. Now we tune the receiver coil into resonance with the transmitter and set the average voltage on the receiver coil so that when approaching the soil this voltage decreases. Taking into account the temperature range from -10 ° C to +50 ° C, we obtained a voltage on the receiver coil of 60 mV with a resonant capacitance of about 47 n .

Now let’s change the resonant frequency of the receiver circuit and calculate both the sensitivity of the sensor to the target (penny, of course) and the soil (a plastic basin of salty sand).

We describe the behavior of the sensor by the coefficients:

Kts u003d U c: 60 – let’s call it the goal coefficient; voltage U c when the penny approaches the sensor increases

Kp u003d U p: 60 – let’s call it the soil coefficient; the voltage U p decreases as the penny approaches the sensor

And finally, the main coefficient:

K = Kts : Kp – it will show the relative sensitivity of the sensor, taking into account the suppression of the soil

The higher this coefficient, the greater the dynamic range of the sensor. It is no secret that with a strong influence of the soil, maximum sensitivity cannot be achieved – you have to reduce the gain of the device in order to avoid false positives. On the other hand, it is always possible to increase the MD gain, if the search conditions allow it.

Let’s check at what resonant frequency of the receiver circuit the maximum K will be obtained. To do this, we will change the resonant capacitor in the receiving circuit circuit and do not forget to set the receiver output voltage to 60 mV each time. Here’s what happened.

 F p , KHz C, n U p, mv U c, mv Kp kts TO Co. n.r. – 60-45=15 150 0.25 2.5 10 1.07 6.9 33 60-30=30 320 0.5 5.3 10.6 _ 1.14 6.1 47 60-15=45 420 0.75 7 9.33 1 5.7 56 60- 28 =32 350 0.5 3 _ 5.8 1 7 . 6 1.9 5 68 60-40 = 20 240 _ _ 0.33 _ 4 12 . 12 1.3 4.3 100 60-48=12 1 35 0.2 _ 2.25 _ 11 . 25 1.2

The last column shows the relative sensitivity of the Ko sensor, with respect to the resonant frequency of 6.1 kHz.

Now in order. As can be seen from the table, despite the fact that the sensitivity of the sensor is maximum at the resonant frequency of the receiver (Kc = 7), which coincides with the resonant frequency of the transmitter, this does not give the best results for the search. The fact is that the sensitivity to the soil is also high. As a result, the dynamic range turned out to be the smallest of all, even for a non-resonant sensor. But, as you can see, the frequency of 4.5 kHz, to which it is recommended to tune the sensor according to White’s methodalso does not give good results. The best was a certain average frequency of about 5.7 kHz, where the relative sensitivity is the highest. To maintain sensitivity, you just need to increase the gain of the device by 20%. It’s a little, and quite capable. The phase shift at this frequency is about 50 degrees (or 130, depending on which angle you measure from).

On the oscilloscope, characteristic graphs are visible:

1. With a non-resonant receiver coil, the phases of the receiver and transmitter are the same.

2. For a frequency of 4.3 kHz (according to White’s method ) – the phases of the receiver and transmitter are shifted by almost 180 ° , which is equivalent to phase matching.

So, I would venture to suggest setting up the sensors in this way:

1. We wind all the coils of the sensor.

2. We adjust the transmitter coil to resonance at low current (it’s easier to catch the maximum).

3. We increase the sensor current until the beginning of visible distortion of the sinusoid and slightly reduce it.

4. We balance the sensor, the receiver circuit is not resonant. The phase of the receiver signal is the same as the phase of the transmitter.

5. We select the resonant capacitor of the receiver, according to the maximum sensitivity. If the output voltage of the sensor turned out to be large and there is no automatic balance, then an absorbing tab can be used. For example, a piece of tin from a tin can. Changing its size and position, shifting the balancing coil, you need to achieve zero (or so) voltage at the output of the receiver.

6. Increase the capacitance of the resonant capacitor until we get the maximum dynamic range (maximum K value).

7. Set the output voltage of the receiver so that there is no “reversal” of the phase during heating and cooling.

Yes, it’s tedious and long. After all, for an MD without automation, you will also have to select elements in the circuit for generating control signals for synchronous detectors, because a non-standard receiver phase shift has turned out. But it seems to be worth the effort.

Sensors with “full” resonance

I did a couple of experiments in the past. It was when I mastered the “Ring” sensor (around 2001?). I made a sensor, tuned the transmitter coil to resonance and then, by analogy with the Blue Max 950 , picked up the resonant receiver capacitor to match the phase of the transmitter. It turned out something like this diagram.

The phase of the receiver signal is almost the same as the phase of the transmitter signal. Transmitter coil voltage 16V, receiver coil 3mV or less.

I then tested the sensor with test items. The results are in the figure below. Both sensitivity and phase shifts were consistent with the Blue Max 950 test results (I found them in some article).

The blue diagram is the soil, the orange one is the nickel. With this characteristic of the sensor, the discriminator works correctly.

All this fits perfectly into the White’s 6000 scheme , for example. There, it is channel X that has an auto-balance circuit – when the sensor is heated – cooled, the change in the amplitude of the receiver signal is an order of magnitude higher than the change in phase. There is no getting away from this, and if the DC gain is high or there is a G channel ( where the X and Y signals add up) – the presence of an auto-balance is mandatory.

But I couldn’t help feeling that something was wrong here. The reaction of the sensor to the glands was already very weak. I must say that before that I unsuccessfully made designs where the receiver and transmitter coils had the same resonance frequency. It looked like this.

The voltage on the transmitter circuit is 16 V, on the receiver circuit 10 mV. This is more than Blue Max , but the sensitivity of such a sensor was much higher. The figure below shows the response of a sensor with “true” resonance.

It can be seen that the reaction of the sensor is approximately 4-6 times higher than that of a branded copy. The signals are shifted by 90 ° , so I then swapped drX and drY . The blue diagram is the soil, the orange one is the nickel. Everything seems to be fine.

But MD did not work with such a sensor. Everything was fine on the table, but as soon as I went out into the yard, into the hot sun, everything “spread”. Very quickly heating of the sensor led to an overload of the amplifiers of the synchronous detectors and the sensitivity dropped to zero. At that time, I sincerely believed that the higher the gain of synchronous detectors, the better.

What was my surprise when, by connecting Blue Maxof my own design to the board, for a very long time, in the same sun, I found rusty nails and oil bolts in the yard of the stoker. I then rented a room for the laboratory in an old boiler room. He paid in kind, repairing various rubbish in a neighboring, operating one. But every buzz comes to an end, the device “passed out” by dinner. I thought for a bit, reduced the gain of the LED and raised the gain of the filter. The device worked well, if the sensitivity of 15-18 cm per penny can be considered a good result. But even that was a step forward. True, on my sensor with a “real” resonance, I received up to 22-25 cm for the same nickel. By the way, at that time I had an old C1-76, single-channel, and I used the full-length synchronization channel, turning it into a pseudo-two-channel. As the first channel from which the signal was synchronized, I used an oscillogram drawn on the screen with a felt-tip pen (usually, a transmitter) – I just circled the signal. Then he poked the probe at another point and sketched “two channels” on the leaf🙂

But it was then that I realized what was the matter and conducted the first experiments on heating and cooling sensors. There was an old refrigerator in the boiler room, it cooled down to about -5 ° C, and this is in the freezer. On the windowsill, the sensor warmed up to +60 ° C, according to the thermometer. South, however…

So, here are the diagrams of sensors cooling down to -5 ° C and heating up to +50 ° C.

At the beginning of work with metal detectors, I naively believed that it was necessary to adjust the sensors according to the minimum voltage on the receiver coil. But look at what comes out: when cooling down, the signal (blue diagram) “travels” to the left, while its amplitude increases at the same time. When heated, it shifts to the right (red diagram), again, the amplitude increases. For your coil, it may be the other way around – when cooled, the signal shifts to the right, when heated, to the left. It does not change the essence, the amplitude increases.

I fought with sensors, probably for a year – until it dawned on me that if you draw the dependence of phase and amplitude, you get something like this picture.

It can be seen that by slightly increasing the initial amplitude of the signal by a factor of two, a small and approximately linear phase change can be achieved. It remains to find a point lying on the linear section of the characteristic, which would correspond to +25 ° C – approximately at this temperature I make sensors. The blue dots show the approximate location of the sensor’s operating temperature on the linear section.

Looking at these diagrams, I decided to make a metal detector with a “real” resonant sensor. It was a pity to lose such gain because of some kind of thermometer 🙂

I succeeded, the sensors have an increased return, and the phase change is almost linear – this is perfectly algorithmized and the processor code is simple. But it didn’t take more than a year…

Sensor balance “to zero”

For conventional analog metal detectors, it is customary to balance the sensor to zero output voltage. It is often difficult to do this, and is it necessary? You know you have to. Those. doesn’t have to be zero, but close to it. How close depends on the metal detector design. The easiest way to determine whether it is necessary to carry out additional balancing of the sensor is by heating or cooling the sensor by 10-20 degrees from room temperature. This is if you are setting up the sensor in the room 🙂 If after cooling the sensitivity dropped, the discriminator began to work incorrectly, or the device went completely silent, then the whole point is in the poor thermal stability of the sensor.

The following happened: when the sensor was heated, the balance point of the Y channel shifted and this affected the sensitivity. The left figure shows the sensitivity diagram before cooling, the right after.

pic 1 pic 2

About resonant sensors with capacitively coupled circuits

Recently, sensors have been used that have a parallel resonance circuit for pumping a transmitter that has a series resonant circuit. The principle of operation of such a circuit is briefly explained in the section Related circuits . I was interested in the capabilities of such a transmitter, especially since the MDs where I met him work quite well. I decided to check what can be “squeezed” out of such a transmitter and what is its efficiency compared to a typical one.

Let me remind you that for me a typical sensor has the following parameters. The number of turns of the transmitter – 31, wound in two wires Ø 0.41 mm. The number of turns of the coupling coil is 11, wound with wire Ø 0.41 mm. The number of turns of the receiver – 350, wire Ø 0.15 mm. Transmitter powered by +8 V. Test results are shown in the table below. Here: Id is the pulsed current in the transmitter coil, Ib is the current consumption from the battery, Utr is the amplitude of the sinusoid at the output of the transmitter, Ctr and Crc are the resonant capacitors of the transmitter and receiver.

 Frequency, kHz ID, mA Ib , mA Is , mA Utr, V Ctr , microfarad With rc , nf 13 380 14 16 0.27 6.8 7 640 29 15.5 0.27+0.68 6.8+6.8

For a sensor with a pump circuit, it was necessary to change the turns of the transmitter, since it uses a coil with a series resonant circuit. It has these settings. The number of turns of the transmitter – 48, wire Ø 0.41 mm. The number of turns of the coupling coil is 16, wire Ø 0.41 mm. The number of turns of the receiver – 350, wire Ø 0.15 mm. Choke inductance Lp 2 .2 mH.

The sensors were tested on the stand, the first one with 8 V supply, the second one with 5 V supply. The switching circuit is shown in the figure below.

The experiment consisted of selecting resonant capacitances for the pump circuit and the transmitter circuit. I used the following series of capacitor capacities: 220 nF, 330 nF, 470 nF, 680 nF, 1 uF, 1.5 uF. For each capacitor in the pump circuit from this series, all values ​​of the capacitances of the capacitors from the same series were checked. The results were entered into a table. The table turned out to be large, 30 rows. As a result, I left the two most efficient lines. They can be seen in the table below.

 Frequency, kHz ID, mA Ib , mA Is, mA Utr, V Cp , microfarad Ctr , microfarad With rc , nf 13 260 28 28 25 0.47 0.22 5.6 7 480 137 62 22 1.0+0.47 0.22+0.47 5.6+15

Unfortunately, with such a number of transmitter turns, it was not possible to obtain a current in the circuit, at least approximately the same as in a typical sensor. But you can compare. Although it has already been suggested that this switching method is suitable for the high frequencies of the transmitter. So, Fisher F70 operates at a frequency of 13 kHz, Troy Shadow – at a frequency of 17 kHz.

Let’s try to derive a certain coefficient showing the efficiency of the transmitter. For example, for a frequency of 7 kHz.

Standard scheme.

At the maximum charge of the batteries, there will be the largest losses, so we will make the calculations for a battery voltage of 13.2 V.

Total power:            Po = V1 * Ib = 1 3 . 2 * 29 = 383 mVA

Power loss:            Pp = (V1 – 8) * Ib = (13.2 – 8) 29 = 151 mVA

Efficiency                                   n = (Po – Pp) : Po = (383 – 151) : 383 = 61%

Power in the sensor:        Pd = Utr * Id * n = 15.5 * 640 = 9920 mVA

Efficiency: E u003d (Pd : Po) * n  u003d (9920: 383) * 0.61 u003d 16

Scheme with a pump circuit.

The biggest losses will be at a minimum battery charge, so we will make calculations for a battery voltage of 3 V. In this case, the efficiency of the converter is 80%, the current from the battery is 137 mA.

Total power:                     Po = V1 * Idc = 3 * 137 = 411mVA

Power loss:                      Pp = Po – 5 * Is = 411 – 5 62 = 101mVA

Efficiency                         n = (Po – Pp) : Po = ( 411 – 101 ) : 411 = 75%

Power in the sensor:        Pd = Utr * Id = 22 * ​​480 = 10560 mVA

Efficiency:                        (Pd : Po) * n = (10560 : 411) * 0.75 = 19

You know, I didn’t expect it myself… It turned out that the pumping scheme is more efficient than the traditional one. It is necessary to check this sensor and this circuit on a “live” metal detector. Something gnaws at me doubts whether the method of calculation is correct. And will there be normal sensitivity …

There is, however, a small fly in the ointment. This is an increased current consumption from the battery. Using batteries with a capacity of 2100 mAh , we get 2100: 137 = 15 hours of continuous operation. For the traditional scheme, the operating time is 2100: 60 = 35 hours. Considering that a 2100 mAh Wart battery costs UAH 42, the savings will be UAH 168. Not bad either, but the choice is yours…

I often heard this opinion: “I don’t have a 0.4 wire, which is recommended to wind the transmitter coil. But there is a lot of 0.3 wire. I’ll wind it up, it’s okay.” I have repeatedly drawn attention to the fact that it is preferable to wind the resonant coils of transmitters with a thicker wire, if there is not one indicated by the author. But my opinion is often not taken into account.

Let’s check in practice how the diameter of the wire with which the coil of the transmitter with parallel resonance is wound affects. To do this, we will make two completely identical sensors, but in one we will wind the transmitter coil in two wires. Like such a litz wire was used.

Number of transmitter turns 31

Transmitter wire diameter 0.41 mm and 2 x 0.41 mm

The number of turns of the receiver 350

The sensor is tuned to 13 kHz with resonant capacitances: 270 nF for the transmitter, 6.8 nF for the receiver. For a frequency of 7 kHz, a 680 nF capacitor is connected in parallel with the transmitter capacitance, and a 8.2 nF capacitor for the receiver.

Let’s set up the sensors. To do this, connect them to the stand and power the transmitter from 8 V, as in PMD. Let us set the maximum possible impulse current Id through the transmitter coil and measure the average consumption current Ib of the sensor from the battery.

Here’s what happened.

 Frequency, kHz ID, mA Ib , mA Utr, V Transmitter coil 7 570 43 14 0.41 mm 7 640 29 16 2 x 0.41 mm

As can be seen from the table, a sensor wound in one wire is less efficient and consumes more power from the battery. And this is a significant loss, about 48% in relation to the second sensor. In this case, the current in the circuit of the first sensor is 11% less than that of the second one. As a result, the sensitivity of the first sensor will be much less.

In principle, I will not give formulas for calculating resonant circuits. I’ve been using the old Kroneger’s handbook since my school days. Who cares – find, look, count and make sure that the active resistance of the circuit is of great importance.

Conclusion: wind the sensors with a thick wire.

The second most important element of the metal detector. If someone forgot – the first is the sensor. It must have a sufficiently rigid structure so as not to work when hitting stones and various grasses.

To depths of 30 cm, I used a Belarusian-made four-wire shielded cable, and everything was fine. Then I increased the return of the sensor, raised the gain of the device and trouble began.

Of course, if the sensor is motionless, then everything is in order – the sensitivity is from 33-35 cm, with the computer and TV running. It would seem that such a hint is understandable even to a fool. But no, I again thought about the scheme, the recognition algorithm, interference from the ground, and so on. And in general, if it is possible to make, for example, ten mistakes, then I will not limit myself to five. I’ll do at least nine. And in a year I will try to make them again, in the same place …

It looked like this. When the sensor was swinging, the device did not respond to a coin lying on the surface of the earth. And if I saw it, then not always, and most often at the end of the swing … I think that at the end of the swing the rod bends – the sensor has some mass, the rod is flexible – this is the design that bends. And the cable is wound on a rod … As a result, the cable is stretched, interference occurs, and involuntary trips begin. Later, when I figured out the essence of what was happening, I also understood the reasons that the device does not work every time on a swing, but in such a way that there are swings with false positives, but there are without them. In fact, there is interference every move. But these interferences have a “wandering” phase, and often it falls into an area that is close to VDIsoil. My device simply ignores such interference, and even tries to reduce their duration. I didn’t understand it right away, but as soon as I understood it, I turned off tillage in the signal analyzer and saw interference in almost every move …

In general, I was tinkering with the circuit for a long time, but finally I realized that the interference was coming from the cable. The fact is that any cable has some rigidity and flexibility 🙂 And if this is not the case, then this is no longer a cable, but a bus (s). The hardness and flexibility of the cable can only be determined practically – for example, by carefully kneading it with your fingers and bending it in different directions. In this case, the sensitivity must be set to the maximum at which there are no interference triggers. If the device is silent, the cable is good.

But there is a small nuance here. I ran into it, and did not immediately understand how to check the cables. This is a socket. It also should not cause a reaction of the device – neither when touched (for this, the connector housing is grounded), nor when moved (for this, it is tightened with a union nut). And these bastards (both the cable and the connector) give such interference that you can’t understand what the device is beeping from …

As always, I spent more than one month until I developed a methodology for evaluating cables and connectors. The essence of the technique is simple: first you need to select a “silent” cable, then a connector.

To check the cable, we solder its wires directly to the device board. Turn off the discriminator (“All metals”). Installing the GEB handle(if present) to a position where the instrument responds to ferrite (soil suppression is minimal). Then we set the maximum possible sensitivity – one step less than the one at which there are rare spontaneous responses. Then we wrinkle and twist the cable. Without fanaticism, but not weakly. If there are triggers, then it is only suitable for the trash can. We throw it away, buy another cable and repeat everything … until we find the right one. We remember where and from whom we bought. Although the latter is not necessary – the next time you buy a cable, even if it looks (and costs) the same as last time, it may be from a different manufacturer and cause interference … I run into this regularly. I have a coil of cable, bought the day after the test. And he’s handicapped. Just to check it was cut off from the started bay, I bought a whole one.

Now you can select a connector. They are old Soviet and Chinese 🙂 Both of them are good (more often old Soviet) and bad. Soviet ones are bad because of looseness and oxidation. I do not buy loose contacts, I treat oxidized contacts with “Contactol” and an elastic band. Then they work without problems – the contacts are often silver-plated, one might say – eternal.

Chinese is more difficult. They are shiny, but the contact surfaces can be poor contact or even oxidized and “sparkling” (metal incompatibility). It also happens that the coating is applied to an unprepared connector pin, and after a couple of months it peels off like a stocking. Faced a couple of times with this, it is very difficult to understand what’s wrong.

The most correct way out is to throw out the problematic connector and put in a new one.

In general, that’s all, it remains to show pictures with normal cables and connectors. Or at least describe them.

Cable . It is better if the cores of the receiver and transmitter have their own screens that are not connected to each other. The transmitter requires two wires and a shield. No current should flow through the screen. There is such a cable … but you can buy it with DJKey, from there. Neither in Russia, nor in Ukraine, I have not seen this. Only in branded devices.

If this is not the case, we buy a two-wire audio “noodle” and let the transmitter return current across the screen … This is not correct, but there is no other way out, except to make the cable yourself. Sometimes I do. I take one and a half meters of two-core audio cable in silicone insulation, single-core for the receiver and a plastic rope (hundreds of veins, not twisted) of the same diameter as for the receiver, and heat shrink. In the groove of the “noodles” I put the receiver cable, on the other hand the rope and pull the heat shrink. I set it down with a hair dryer. Then another layer of heat shrink, maybe two. The rope is bent and “banded” at both ends – when pulled, it will take on the breaking force. The view, of course, is worse than the “branded”, but it works and gives up to 40-45 cm of depth 🙂 You can’t get it on a regular cable …

If you are too lazy to do it, pick different ones until you succeed …

Connector . I usually use microphones, they are also used on soldering stations 🙂 Chinese. I take those in which the “dad” pins have a longitudinal section – when contact is lost, and this happens only in the field, 100 km from home – they can be opened with a knife or a screwdriver and continue the search. I cleaned my mother in the field with matches, soaking them in vodka. It should always be with you at any searcher and angler 🙂

Note. All figures and diagrams show the relative dependencies of the parameters and may differ in value from those obtained at a particular moment and on a particular sensor. I also did not attach importance to the polarity of the receiver signal – for me it does not matter.