Georadar “Radalit”. General description, block diagrams, terminology
Block diagram of the georadar “Radalit”
Figure 1 shows a block diagram of the georadar (GR) “Radalit”.
|
Fig.1. Block diagram of the georadar “Radalit” |
Functionally, the GR consists of five modules. The power supply and battery module VT1 and V3 provides power to the GR and battery charge. The transmitter module with the transmitting antenna B6 generates a probing pulse. The receiver module with the B7 receiving antenna provides amplification of the reflected signal. The operation of these modules is controlled by the system unit or signal processing module. It also outputs the processed signal to a tablet or smartphone running the Android operating system, preferably version 4.0 or higher. Screen resolution of the tablet is not less than 800×480 pixels. Tracks are recorded on the SD card available in the tablet. The primary interface to the tablet is the 10/100 Ethernet port . However, recently the USB OTG port has become more and more popular.. In the GR, it operates in CDC mode , emulating a high-speed RS-232 port . This is good, as it allows you to significantly reduce the current consumption from the battery by turning off the Ethernet link . In GR “Radalit”, to save battery power, a physical shutdown of the Ethernet port is applied .
In addition, the signal processing module interacts with auxiliary devices. This is a B2 measuring (tracing) wheel, a GPS module for tying to the terrain and an RS-232 service information port for communicating with a personal computer.
How does GR “Radalit” work? When you turn on the power and the tablet, the battery charge is checked. If it is sufficient, then service information recorded on the memory card is transmitted from the tablet to the system unit of the GR. These are the operating modes of the transmitter and the signal processing module. For a transmitter, this is the frequency of the emitted signal. Processing parameters are transferred to the signal processing module: the size of the tracing window, the number of samples per scan, and other values. For example, the tracing method: from the measuring wheel, from the GPS module or manual mode. All of these values can be changed before and during tracing.
Then the GR is balanced. To do this, without starting the movement, you should raise the GR above the ground to its original position, press the “Balance” button and wait for the “Track” button to light up on the tablet screen. In a few seconds, based on the analysis of the reflected signal, automatic adjustment and synchronization of all radar modules will occur. If necessary, you can adjust the upper level of the trace in the output window, change the window size, number of scans, soil parameters, and adjust the brightness of the image. After pressing the “Track” button, you can start moving.
|
Fig.2. |
Fig.3. |
Since the device was conceived as an inexpensive alternative to complex ground penetrating radars, the question of how and from what batteries to power it is quite acute. Naturally, a 12-volt acid battery is the cheapest, for example, from a computer UPS unit or from a Chinese lantern (do they still exist?). But they are heavy… For example, LiPo batteries have much more capacity with less weight. True, their cost is 20 times more than acid …
Therefore, the GR has a universal power supply in the range from + 8V to + 15V. Three LiPo cells will provide a supply voltage of +8.4 ÷ 12.6V, and an acid battery +9 ÷ 12.6V.
A typical acid battery ( SLA battery) CB1270 provides a voltage of +9 … + 12.6V, its capacity is 7 Ah, and its weight is 2 . 7 kg. The same acid battery with a capacity of 4 Ah weighs 1.2 kg. The battery of two LiPo cells LP65105198 weighs only 580 grams. Each 3.7V element has a capacity of 10 Ah, but provides a lower supply voltage in the range of +8.0 ÷ 11.1V.
GR batteries can be recharged directly “in the field”, from the car battery. The tablet also charges from it. Of course, the charge circuit for a LiPo battery is more complex than for an acid battery. This increases the cost of using a lithium battery.
Since the probability of an explosion of a lithium battery (meaning – a large capacity) is somewhat greater than zero :-), it is placed in a separate case. Cases are known when the explosion of a high-capacity lithium battery irrevocably damaged all, without exception, tools and materials in the workshop. It is clear that the GR will be simply destroyed if the LiPo battery is in the same case as the system unit. And this is a thought… Suggestive when thinking about the problems of protecting development…☺
|
Fig.4. Power and charge circuit |
Let’s look at the block diagram of the power supply of the GR (see Fig. 4). Based on the requirements of decoupling from crosstalk of various devices, and also taking into account that the outputs of some external devices can be shorted to ground (power short circuit), almost every GR unit has its own stabilizer. The abundance of such stabilizers and their power supply from a battery with a voltage of about 12V can cause additional local heating of the board, which is not very good. Therefore, a common low-voltage power rail was implemented, and low-voltage dropout ( LDO ) microcircuits were used as stabilizers . Using as stabilizers LDO LP2985 (LP2981 for low-current units ) with a voltage drop of 100 ÷300mV, it is possible to implement a common power bus with a voltage of + 5.7V, 500mA . Common bus stabilizer B1 is also LDO , type NCP5500 . With a bus current up to 500mA (typical consumption 250mA), the voltage drop is 300 ÷ 500mV and thus the GR circuit is operable up to a voltage of +6.3V. At a higher voltage, the excess power will be dissipated on the IC B1mounted on a small radiator.
The power module consists of the main unit, it is on the left in the figure and is located next to the power battery. The remaining blocks are shown conditionally, to estimate the current consumed from the battery.
The charge scheme turned out to be complicated. Note that for an acid battery, elements B2, B3, B5, B6 and B7 are not needed. However, consider the complete circuit for charging a LiPo battery.
Connector J1 receives a charging voltage of 11 … 15V from a car battery or from a power supply unit. The maximum charge current is 2A. DC / DC converter B4 produces a stable voltage + 16V, current 2 A. To prevent the explosion of a lithium battery, the charge current is limited to 1000mA using a B3 current source, and the voltage is +12.6V (+11.1V) by a B2 stabilizer. However, this is not enough. Since the batteries are purchased in retail, they cannot be identical in terms of parameters. Therefore, it is very likely that one battery will charge faster than the other, it will start heating up and an explosion is very likely. To prevent overcharging, balancers B5, B6 and B7 are included in parallel with each battery. They limit the voltage on each element to +4.2V (+3.7V). At the same time, when the charge is reached, the LED installed in each balancer lights up. Thus, the charge is considered complete when all the LEDs light up.
Now we can estimate the charge conditions. For example, for a LP65105198 battery , the charge time at a current of 1A will be
10 A*h / 1A*h = 10 hours
Due to the fact that the volume of the device is small, a charge with a current of 0.1C was chosen. Let’s allow a charge up to 1C, but such a mode would require a more powerful converter, current source and balancers. And, therefore, huge radiators. This is a compromise for an overnight GH charge.
For an acid battery, the charge time at a current of 1A is
4 A * hour / 1A u003d 4 hours
More difficult with the battery charge of the tablet. The figure shows the option of charging from a car battery or from the mains. But there is another charging option: this is when the tablet is recharged from the GR battery – this is how, for example, children’s helicopters are charged from the control panel. But this is not very convenient, since it “sits down” the main battery of the GR batteries. This mode can only be used for a capacious battery, at least 5 Ah. Such a charge mode has not yet been implemented. It is more correct to charge the tablet from a car battery or a 220V network. When charging the tablet, the B8 stabilizer is used.
The receiving antenna is a conventional dipole with interchangeable directors. There are two sets of directors, one with a center frequency of 433 MHz, the other with a center frequency of 832 MHz. The directors are fastened with a screw, which allows them to be changed directly in the field. Since such an antenna unit perfectly catches reflections from buildings or trees, then (as an option) a screen made of radio opaque material is provided that covers the antenna from above and from the sides. However, such a screen is quite expensive and bulky, and therefore it is used optionally.
The second (replaceable) “butterfly” antenna. It comes with its own amplifier, which is mounted on top of the antenna. It does not require special fastening, the input wires of the balun are insulated with fluoroplastic cambric and are simply passed through the drilled holes. The screen for the “Butterfly” is much simpler, it was made of foam and graphite filler. Works, although not as well as with a proprietary absorber. But the price of the company bites, however …
The antenna is connected to the LNA using the so-called balun (balun, bal ansed- unbalanced). This is a typical impedance matching device that allows you to go from a 300 ohm dipole resistance to a 50 ohm LNA input. The receiver uses a balun from a decimeter Polish television antenna. The LNA output is also matched to the 50 ohm cable. The cable is connected to the LNA via the CP-50 connector. Cable length approx. 2 m.
Figure 5 shows the device of replaceable directors of a dipole antenna. Each of the directors is fastened with a screw 1 to an aluminum base 2. The bases, in turn, are screwed with screws 3 to the body. Washers 4 are placed under the screw heads, pieces of wire 5 are soldered to them. These pieces represent an overhead transmission line with an impedance of 300 Ohm.
|
Fig.5. Antenna amplifier assembly drawing |
Fig.6. Receiver assembly |
Figure 6 shows a photo of the receiver, without an antenna connected. It was later tested on a butterfly antenna. During final assembly, the receiver board will be placed in the shield.
transmitter module
One of the main GPR modules – the transmitter – must provide a powerful (800 V, 50 A) and short (3..6 ns) probing pulse on the transmitting antenna. Both the depth of the georadar and its efficiency depend on this module.
The block diagram of the transmitter is shown in Fig.7.
|
Fig.7. Transmitter Block Diagram |
The power supply unit B1 of the transmitter supplies the output stage, assembled on an IGBT transistor, with a supply voltage of +200 V. From the CPU unit, a trigger pulse for the transmitter is sent to the CR1 connector EN_TR . This pulse triggers the single vibrator B3, which in turn generates a trigger pulse. It is amplified by driver B4 and opens the IGBT transistor Q1 . Saturable transformer T1 is magnetized and charge C2 begins. Diode D1 is open at this time. At some point, C1 will be fully charged and the discharge will begin. Diode D1 closes and a short high-voltage pulse is formed on the load. Figure 2 schematically shows the process of pulse formation.
|
Fig.8. Formation of a high-voltage pulse |
Fig.9. Transmitter “Geotron” |
Here ID is the current flowing through the diode. Uc is the voltage across the capacitor. U n – voltage on the load (see Fig. 8). This process is explained in detail in the article “Pulse shaper with a current breaker connected in parallel with the load” posted on this site.
Next, the high-voltage pulse is fed to the transmitting antenna. A bipolar pulse shaping circuit ( PFN ) is connected between the shaper and the antenna . As a result, the antenna emits a monopulse with an amplitude of several hundred volts and a duration of several nanoseconds. There are ambiguities here. The literature says that monopulse is preferable. However, two papers note that a unipolar pulse gives better depth and resolution. Need to check…
Trigger pulse frequency EN_TR 100 kHz, Synchronization pulse length SYN 5 µs, it starts about 100 ns before the formation of a high voltage pulse. Accurate time binding is carried out in the system unit at the start of the GR operation and is subsequently stored using the tracking scheme. Snapping accuracy of several hundred picoseconds, the tracking circuit performs step-by-step calibration every scan.
A photo of the transmitter is shown in Fig.9. The top cover has been removed.
The circuit of the signal processing module is, in fact, a two-channel sampling oscilloscope☺ Figure 10 shows a block diagram of the signal processing module.
|
Fig.10. Block diagram of the signal processing module |
GPR devices are clocked from an accurate master oscillator B2. At the output of the generator, we obtain a reference signal (meander) FRQ with a frequency of 100 kHz and an amplitude of 3.3 V with minimal jitter. From this signal, the processing circuit is clocked. Through the output buffer, this signal is transmitted by cable to the transmitter (connector CR 1) and starts it. The transmitter generates a high-voltage pulse 6 ns long and 600…800 V in amplitude.
The signal processing circuit can be clocked both from the built-in generator and from the trigger pulse coming from the transmitter to the CR2 connector . The trigger pulse is rigidly tied to the transmitter pulse. The clocking mode is selected by switch SW1 (in the first sample, the mode is selected by jumpers).
The trigger pulse is fed to the controlled delay line (LZ) B5. The LZ generates a pulse with a duration of 1700…4000 ns, depending on the control signal TD_WIN (beginning of the measurement window) generated by the microcontroller. Thus, by changing the duration of the pulse with the help of a DL, we will sooner or later match the beginning of this pulse with the high-voltage pulse of the transmitter. Further, such a combination is supported by the tracking scheme.
The circuit has a precision (relatively, of course) sawtooth voltage generator (GPN). This is the calibrating device of the system, it ensures the operation of the tracking circuit and serves as a ruler (and I would like it to be a caliper or even a micrometer ☺) for measuring the depth of the object.
Since the output pulse of LZ V5 has a different and unpredictable duration, a control pulse with a duration of 300 ns is generated using a single vibrator V6 to control the GPN. GPN V7 generates a linearly increasing voltage with an amplitude of 100 mV and a duration of 300 ns with high accuracy and thermal stability. The beginning of the “saw” coincides exactly with the beginning of the transmitter’s high-voltage pulse. This is a reference signal, and if it is applied to the B8 sample-and-hold device (stroboscopic converter), then at the output of amplifier B9 we get a linearly increasing voltage, but transferred down in frequency. The output signal frequency is about 4 Hz, it can already be easily digitized by any analog-to-digital converter (ADC). Suitable even built into the microcontroller (MK).
Knowing the law of change of the reference pulse and measuring the voltage U_RAMP using the ADC of the microcontroller, we can easily calculate the delay time relative to the transmitter signal in ns. We need this time to calculate the depth of the object.
Why is this reference sawtooth signal needed at all? The fact is that the controlled DL has some non-linearity (well, to be honest, decent) in the formulas for calculating the signal delay. Thus, by reading the voltage of the saw, it is possible to correct the actual sampling time relative to the calculated formula and improve the accuracy of the depth calculation. On the other hand, this voltage is a reference voltage for evaluating the influence of the block temperature on the conversion parameters. If there were no such reference signal, then a temperature increase of 10-15 ° C would lead to the complete inoperability of the GR. Fortunately, temperature changes are slow things, and by comparing the values of the reference signal and the beginning of the high-voltage pulse, you can correct the desired values.
The voltage at which the saw voltage begins to increase is stored in the memory of the microcontroller. At the beginning of each trace, this voltage is checked and, if it has changed, the control signal TD_WIN is adjusted accordingly . This so-called control voltage tracking is needed in order to compensate for the influence of various disturbing factors, such as changes in voltage and duration of signals due to changes in temperature, humidity and the phase of the moon☺.
Since when processing the trace signal, its length of 32…256 ns is transferred to the frequency range of 32…256 ms, the tracking of control voltages is quite fast (one step every 32…256 ms) and we have time to correct temperature changes .
Now that we have “attached” the beginning of the trace to the high-voltage pulse of the transmitter, we can start tracing – i.e. measure the amplitude of the receiver pulse (the reflected signal of the transmitter from the search object) at a strictly defined time relative to this same transmitter pulse. To obtain an acceptable depth resolution (5…15 cm), we must carry out measurements with a sampling pulse shift in one step by 0.5…1 ns. Since the trace consists of 32…512 samples, we get a depth at a step of 0.5 ns of approximately 300…5120 cm (or 3…51 m). Of course, for different soils, the depth will change – the wave propagation velocity strongly depends on the soil material (approximately 5…15 cm/ns). And you just can’t get a depth of 51 m – the signal attenuation at a frequency of 433 … 700 MHz is very significant, and to increase the depth of measurement, you need to switch to frequencies of 25 … 150 MHz. In reality, at a central frequency of 700 MHz, you can get a depth of 2 … 4 m, at 400 MHz, 5 … 7 m.
The signal from the receiver goes to the CR3 connector and is amplified by 3…4 times, mainly to compensate for losses in the cable. Then it goes to the B16 stroboscopic converter. Sample pulses (for both stroboscopes) are formed similarly to the “saw” pulse. With the difference that the controlled LZ produces a pulse of 700…1600 ns with a step of 250 ps. For reliable operation of the sample pulse shaper B13, a single vibrator B12 was used. It generates a 1 μs long driver circuit trigger pulse. At the output of the pulse shaper transformer, we obtain pulses with a length of 0.5 … 2 ns and an amplitude of 3 V. These pulses are fed to stroboscopic converters (shown in the figure as a diode bridge).
The resulting sample voltage is fed to several B16 level converter amplifiers. This circuit is similar to the previous one (B9) and is also balanced by the BAL_SIG signal from the microcontroller. The difference is in the overall gain, it is much larger and makes it possible to digitize the input signal with an amplitude of 50 μV … 125 mV. As a result, we get a dynamic range of input signals of about 125/0.05=2500. This sensitivity range is necessary to compensate for signal attenuation. Without gain compensation (according to the logarithmic law), the screen for a depth of more than 1 … 2 m would become too dark. Those. to display a point on the tablet screen, a compressed signal is used in the range from 0 to 256 units.
Next, the U_SIG signal is digitized by the ADC of the microcontroller (total resolution using a special circuit is about 20 bits) and subjected to multiple digital transformations (equalization, median filtering, envelope extraction, bandpass filtering). This makes it possible to significantly improve the visual characteristics of the received signal.
