Induction Balance Metal Detectors




The metal detector is designed for searching for metals in the ground (and beyond) with discrimination by types (ferrous and non-ferrous). Type: VLF-IB. The search for metal objects is carried out by scanning the magnetic field of the ground and registering changes in the primary field due to the influence of metal, followed by phase discrimination and notification by a sound signal. With certain skills, it is also possible to determine the relative depth (size) of the object. What sets the device apart from analogs is an effective system for suppressing the influence of the ground, “hot” rocks, and sensitivity with a small sensor diameter, allowing for the detection of small and very small objects. The device incorporates original author solutions and a modern component base. The swinging motion is powered by powerful CMOS transistors. The channel amplifiers are built on precision, low-noise operational amplifiers.

Thunder-3, or Гроза 3, is an analog metal detector developed in 2004-2005 that operates on the VLF-IB principle. The metal detector features a two-tone sound mode (for ferrous and non-ferrous metals) for target recognition and the ability to disable ferrous metal detection. In addition, the version of the metal detector that came to me for repair has been equipped with a LED scale indication called “Молния” (Lightning).

Here is a translation of the specifications:

Technical specifications:

    • Power: 6V
    • Frequency: 8192 Hz
    • Consumption: ~110 mA (excluding sound)
    • Current in the circuit: ~140 mA
    • Coil diameter: 160 mm
    • Sensitivity in the air: ~35 cm (5 kopecks USSR)
    • Discrimination: 2-tone sound, low tone for ferrous metal, high tone for non-ferrous metal
    • Pinpoint mode: Sliding tone (ГУН or VCO)
    • Ground balance with “stone notch,” sound indication of the balance process in pinpoint mode (no analog)
    • Synchronous detector allowing compensation for any balance mismatch, two half-period (no analog)
    • The system, with the right coil, allows “cutting” grass without reacting to the ground, does not respond to small impacts.
    • Compatible with any balanced coil system. Three active regulators, pinpoint button, and phone jack. All components are SMD. The circuit can be adapted to any pumping current.
    • Protection for power input against polarity reversal and voltage exceeding (<6.8V). Self-recovering fuse.
    • The system is relatively resistant to household interference and similar issues.

Scheme of the Thunder-3 Metal Detector

Scheme of the LED indication scale “Lightning”

Photos of the circuit board of the metal detector ‘Гроза 3’

Photos of the indicator scale board ‘Lightning’ for the metal detector ‘Гроза’

Description of working principal

The phase control scheme with synchrodetectors is borrowed from the renowned A50 project, created by a well-known specialist. In my opinion, such a solution is most optimal for a purely analog device. Moreover, the idea itself applied in that device is simple and effective in combating the influence of soil and stones. The use of the phase method to address these phenomena opens the path to further improving amateur systems built on analog principles. Although this idea has been known for quite some time and applied in some proprietary devices, it seems that no one has developed it (?). Many benefits and new ideas are hidden in this area. Synchrodetectors: The problem that drove me to this SD scheme is quite serious for amateur and non-amateur constructions alike. The challenge lies in the fact that we can achieve a good balance in the search system. However, we cannot maintain it for various reasons beyond our control. As a result of sensor imbalance during operation, the signal mismatch at the sensor output due to the influence of all destabilizing factors can be significant. There will never be a zero balance there, even from the influence of the ground alone. With the appropriate amplification of the first stage, this “error” can reach unacceptable levels for the circuit operation, causing saturation of the preamplifier and the synchrodetectors themselves. To address this issue, for example in the A50, the author increased the power supply, thereby stretching the dynamic range of the SD. To eliminate the error (DC component), isolation capacitors were applied, and this is correct. Many devices, particularly capacitors, are built this way. I took a different approach. Let’s not delve into the operation of the SD, as it is a well-known topic. Let’s focus on a specific solution. The input stage and the entire signal processing scheme are built on operational amplifiers of the OP07 type. In my opinion, these are the most suitable operational amplifiers for our case. They have a standardized noise level, which is very low, specifically for those infrasonic flickers, unlike other operational amplifiers.

The first cascade DA7 has a gain of 11. The second one, DA9, serves as a simple inverter. As a result, we have the input signal scaled to a certain level and the antiphase output of the preamplifier. The signal then goes to the switches of all three channels, and they operate identically.

Let’s consider one channel. The switch KT1 (X channel) is controlled by its phase from its comparator DA14. At each phase control state, the switch alternately connects the phase or antiphase of the input signal at its output. This results in a two-half-period rectifier. The advantages of such a synchrodetector over a single-half-period one are evident, with greater efficiency and more effective rectification of the input signal.

Next, the pulsating signal goes to the input of the next switch in the channel amplifier, KT2. This switch is triggered by a generator shared among all channels at a frequency of approximately 45-60 kHz. The signal from the output of this switch enters the integrating circuit, where all pulsations are smoothed.

Then, DA16 amplifies the signal, and as a mismatch signal different from 0 (idle state) arrives at the error tracking scheme, DA17. This is a typical integrator/inverter. Through the switch KT2, the mismatch signal returns to the input of the cascade and subtracts the error to bring it back to 0. Thus, we have a cascade with automatic zero adjustment. The switching frequency of KT1 is tied to the main frequency (8 kHz), while the frequency of KT2 is several times higher.

This is done to ensure that the circuit “pays attention” to the input signal and its changes as frequently as possible during one half-period, not just compensating for 0 at its output. That is, uniformly, several times during the half-period, and since the frequencies are not multiples, a “sliding phase” effect occurs. This has practically reduced detector errors (losses) to 0. Interference penetration from the use of different, non-multiple frequencies is absent. The difference and sum of their frequencies are too large for us to detect. What magnitude of imbalance can there be, mismatch of the coil?

In other words, what magnitudes of errors do we need to deal with? In reality, at the output of the synchrodetector after rectification, the maximum error, either from the action of a large object on the coil (full imbalance), does not exceed 2.5-3V DC with such power supply… So, to subtract it, you need exactly the same magnitude but with the opposite sign. KT2 performs this operation. The subtraction itself occurs on the integrating circuit R42C18.

Similar solutions are applied in well-known Tesoro devices. However, in those cases, a resistive divider is used for error subtraction, which divides not only the useful signal (albeit to a small extent) but also the compensating signal itself. This would require high power to obtain an equivalent subtraction signal.

In my solution, there are no losses in either the signal or the compensating voltage. Therefore, the circuit never goes into saturation, and its “return to zero” time is acceptable. In the worst case, it takes a few seconds with prolonged impact on the sensor. The device does not “freeze” for half an hour…

Thanks to this channel design, the possibility arises to design channel amplifiers in a static mode. It is convenient to implement analog memory. There are no transient capacitors, so you can operate in an “open channel.” The use of a Nyquist plot is feasible; the channels are linear. You can use the total output DA28(Y1+Y2)=Y and before the comparator DA25=X. For the Nyquist plot resolution strobe, you can use the coincidence circuit signal DA26DA27 – for example, a logic 1 (points and outputs for this are provided on the board).

Continuing with the schematic, there are no particular features until the sound generation.

Here, I believe it’s worth pausing. With this method of combating ground interference (phase nulling), there is an inconvenience during the operation of the device related to choosing the optimal balance point for the ground. The method of nulling implies the presence of two channels (in this case), synchrodetectors of which are tuned in such a way that they are symmetrically positioned relative to the ground balance point. This point is nothing more than the phase reversal point of the signal at the synchrodetector output.

In other words, when one channel has a positive response to the ground signal, the other has a negative response, making the channels react in antiphase to the ground. However, under the influence of a metallic object, the reaction of the channels is always the same, in this case – positive voltage at the output. This point is very delicate and fragile. Hitting the optimum precisely is quite problematic. There are no criteria for evaluating the correctness of the “ground balance,” and this confuses the user. As it seems to me, I have managed to resolve this issue.

For generating discrimination signals for black and color targets, I applied the Discriminator Unit (DD7). The state of its control input depends on the voltage from the divider R73R74 and the state of the “color” comparator. High pitch corresponds to a color target, low pitch corresponds to black. The switch KT7 is off when button S1 is released. KT9 is activated by the coincidence circuit on comparators DA26DA27 and connects the load R78 to the switch KT10, which is switched by the frequency of the Discriminator Unit.

The signals from the channels Y1 and Y2 are summed at resistors R64R65. When a target is present, a positive polarity voltage is formed there, the magnitude of which depends on the signal amplitude, i.e., the size or depth to the target. This signal, through the buffer DA28, controls the volume of the signal in the speaker (headphone). This is the main operating mode.

When button S1 is pressed, the switches KT7 and KT8 are activated. Switch KT9 remains open constantly, and the voltage from the sum of the channels (DA28) is applied to the input of the Discriminator Unit. Now, the frequency of the Discriminator Unit is determined by the magnitude of the signal from the target. The higher the signal, the higher the tone. This results in a dynamic Pinpoint mode.

But thanks to this solution, it becomes possible to “hear” the ground and its balance, qualitatively assess that very criterion. The process of ground balance occurs as follows: with the “window” closed (R2 short-circuited) and the P/P button pressed, you can hear the tone (I prefer using headphones; they allow you to hear subtle details without distracting yourself or attracting attention). As you rotate the GB knob from one extreme position to another and bring the detector closer to the ground (up-down), you can hear a shift in the tone frequency, its deviation.

Let’s say, with some movement, initially the tone goes up-down, and in the other extreme, it goes down-up. But there is a position of the GB knob where the deviation of the tone is indeterminate. In other words, the tone “does not want” to go either up or down. This point is quite precise and specific (with the “window” closed!). This is the point of ground balance, the criterion for adjustment. On the sand, by the way, it is practically impossible to determine it due to the low activity of the sand. And that’s a good thing!

Now, releasing the P/P button and slightly opening the “window,” we can confidently say that the channels Y1Y2 are tuned symmetrically relative to the ground. At the summing point of the channels R64R65, the signal is minimal because it is subtracted. The volume is minimal, and the tone deviation is also minimal. The threshold adjuster of the channel comparators determines the sensitivity threshold only in dynamic mode.

For smooth adjustment of the threshold, to avoid accidentally making it coarse, I applied a certain “trick.” I intentionally set the voltage at the outputs of the channels to, say, +50+100mV. But it’s precisely the same on both channels. This is done to bypass the beginning of the regulator, which has a jump in resistance at the beginning of adjustment. Finding a potentiometer with an inverse logarithmic dependence (exponential) is not always possible. However, this approach allows you to use a linear regulator, shorting its wiper to ground with a resistor. By adjusting R66, it is possible to influence the slope of the adjustment characteristic, stretch the start of the sensitivity scale.

The use of a linear audio amplifier for the output signal allows connecting any load from 8 to 50 ohms, any speakers, and headphones. At the same time, the amplifier itself consumes only 2.5mA of current.

The power control of the microcircuit is handled by VT1, a 5V supervisor. But by selecting a resistor at its input, you can coarsen it to 5.3V. The circuit from DD8 to DD11, with a decrease in power to this value, sends short signals to the sound channel with an interval of 3-4 seconds. The device continues to work, but there is information about the battery (accumulator) status.


Related Articles

Back to top button