6
amount from several ones to a complete trace, which significantly speeds up the operation
of the entire system and increases the volume of the received data stream. The overall
system clocking occurs at high frequencies (hundreds of MHz) - it allows the reducing of
the time base step to tens of picoseconds. But on the other hand, while the time range per
sample is decreasing, the number of transmitter launches per trace is increasing, because
the speed of the ADC does not allow a true description of the analog signal. That’s why the
central processor uses the ADC in a mixed mode operation (i.e., in a pseudo-stroboscopic
mode), where each transmitter triggering gives several samples on output, but not in a row,
with an offset corresponding to the selected time range and the ADC clocking frequency.
For example, if it is necessary to obtain a complete trace in 10 launches of the transmitter,
then the first time the system receives 0, 10, 20-th, etc. sample, further the ADC clocking
pulse phase is shifting by the corresponding value, and the system begins to receive the 1st,
11th, 21st, etc. sample on the second run. Thus, on the 10th run, the trace is completely
formed. But, if we compare it with the classical stroboscopic method, this happen an order
of magnitude faster. The traces overflow is better to convert to the high stacking (up to
hundreds, thousands and tens of thousands times), that increases the bit resolution and
dynamic range of the data, i.e. it increases the signal-to-noise ratio (one of the main
advantages of this method).
This method significantly changes the general structure of the pulsed GPR. First of
all, this is the refusing from the analog stroboscopic converter, because its functions are
now performed by an ultra-fast field programmable gates array or FPGA. It also eliminated
the need to use a receiver’s electronics with a sample and hold device, since an original
signal goes directly to the differential input of the ADC from the dipoles of the receiving
antenna. But one of the main differences is the mandatory use of FPGA, instead of
microcontrollers or processors. The main feature of FPGA is the parallel execution of
operations, and not sequential, unlike in ARM or RISC processors. This gives the
necessary speed for 64-bit operations with the "fast" ADCs, since the main task of the
FPGA is to quickly receive data from the ADC, store them in memory, stack and transfer
the final GPR traces to the logging device (computer). Such a range of tasks is beyond the
power of even modern embedded multi-core processors. One of the main advantages of the
whole system is the exact clocking of all processes under the control of the central logic,
which eliminates the analog adjustment of parameters and their “floating” during
operation.
Regardless of the GPR family architecture, it is controlled by a PC with the
“
Prism2
”
software, the description of which you are reading right now. While you are
analyzing the GPR data, it should be borne in mind that the propagation velocity of
electromagnetic waves in the sounding medium (if it is not an air) is not equal to the speed
of light, but less than it by a factor of slowing down. The deceleration coefficient is equal
to the square root of the medium dielectric constant; this factor is automatically taken into
account in the
“
Prism2
”
software.
