Development of an Integrated Hardware and Software System

Over the past three years, GNPP Aerogeofizika has been developing a new integrated hardware and software system implementing the principles of an airborne electromagnetic system based on the DIP-A method, the airborne variant of dipole inductive profiling. The authors seek to describe the complex course and results of this development. Unfortunately, presenting this material involves certain difficulties. It was tempting to relate meticulously documented results to well-established scientific premises and solemnly announce that the work was complete. We could even have carefully outlined the principal directions of our planned activities and the associated trends in the development of Aerogeofizika’s instrumentation as a whole. However, as developers, we feel, first, that a dry, one-dimensional result presented to the scientific community loses its expressive power; moreover, a developer always wishes to show not only the result but also the elegance of the solution. We therefore chose to present the material as a report—or rather, as individual pages from a field diary—following the entire path step by step, beginning with the basic premises underlying both individual decisions and the project itself. We recognize that this presentation is, regrettably, not without shortcomings. First, everything described here is, in a sense, Aerogeofizika’s “underwear”; second, in an equally specific sense, it embodies hopes of success in fierce competition. Let us begin.

D. Fountain’s excellent review of the history of electromagnetic systems (1) briefly mentions the DIP-A method (dipole inductive profiling), clearly implying that both the DIP method and its hardware implementation are nothing other than the internationally known F400 instrument. This is not entirely accurate. The DIP method originated in the USSR, where its scientific foundations were developed and the first airborne electromagnetic instruments were built. Like so much in Russia, the method has a character of its own. It operates purely in the frequency domain, yet is implemented in a fixed-wing configuration, and has several interesting features. Its theory and use in geophysical interpretation are described in considerable detail in the works of B.S. Svetov and in GNPP Aerogeofizika publications; generally speaking, these topics are not directly within the remit of the group developing the hardware and onboard software.

Our work began with an analysis of earlier hardware and software developments within the already operational method. After brief deliberation, we decided to continue developing the instrumentation specifically for the DIP method. It should be noted that the immediately preceding implementation already employed considerable “heavy artillery”: signal processing, digital filtering and synchronous detection, as well as digital control of automatic phase correction and compensation for the ellipticity of the measured field caused by the electrical conductivity of the aircraft and towed bird themselves. Overall, the instrument demonstrated exceptionally high operating performance, but largely retained an unwelcome feature of all previous implementations: instability. In actual flight, the system could be adjusted to very high accuracy and sensitivity—approximately 100 ppm in the channel measuring the ratio of the polarization ellipse semiaxes—but this accuracy could not be maintained for extended periods. Significant changes in external conditions were by no means always the cause. Most major errors arose from phase instability within the measuring equipment itself. Moreover, in some cases the method began producing results that were, to put it mildly, strange.

This was the situation when our work began. Our first task was to design a digital band-pass filter and a very fast algorithm to implement it. The work took approximately a year, with alternating successes and disappointments, and culminated in a filter providing approximately 150 dB attenuation with very strong suppression of the first sidelobe. The computational performance of the algorithm far exceeded our most optimistic expectations. Suffice it to say that the system under development calculates approximately one hundred quadrature components four times per second on an ordinary Intel Pentium 200 computer. The filter’s frequency response is shown in Fig. 1.

At this stage, we agreed the principal operating specifications of the future system with the geophysicists responsible for interpretation and, consequently, for all potential prospects of the method. A painstaking struggle over every least significant digit culminated in the signing of a technical specification for a frequency-domain electromagnetic system based on the DIP method, with the following characteristics:

Frequency set: just four frequencies—but what a set!—130 Hz, 520 Hz, 2080 Hz and 8320 Hz. Anyone who has worked on this problem will, we hope, appreciate the elegance of the choice. The lowest frequency lies between two very strong harmonics, the second and third, of the 50 Hz mains frequency. The highest frequency, doubled relative to the previous implementation, in turn requires a high signal sampling rate: in our system it is eight times the highest frequency, or 66560 Hz.

The receiver coil system has three components. This deserves particular attention. Magnetic field theory tells us that field strength (flux density) decreases inversely with the cube of the distance from the source, assuming a point source. Yet the receiver coils cannot be placed close together because of strong mutual interaction. Conventional air-core coils were generally spaced approximately 0.5 m apart. In our case, even with ideal signal measurement, the ratio of the signals from the front and rear receiver coils in the bird, with a tow cable approximately one hundred metres long, would be 100³/101³ ≈ 0.9706. An orientation-dependent discrepancy of THREE per cent against an expected method accuracy of 50 ppm is simply catastrophic! Perhaps the devil is not as fearsome as his little offspring, and the problem could be addressed by making certain assumptions, but commissioning such a system would require considerable ingenuity. A relatively brief analysis reinforced our already firm intention to use ferrite-core receiver coils, arranged compactly and symmetrically. We must admit that this proposal was—and arguably remains—quite risky because of possible system nonlinearity and the coercivity inherent in even the best ferrite. What sustained us in confronting both our own doubts and opponents of ferrite was that the technology for producing excellent shielded air-core coils had already been established before our work, during development of the preceding instrument or perhaps even earlier, and that we had plenty of such coils available. We could always return to the earlier solution. Even that argument, however, is not entirely sound: doubling the upper operating frequency relative to the previous implementation requires substantially more effective electrical shielding of the coils. Our compact arrangement of individually shielded coils (Fig. 2) is enclosed in a common double shield. It is difficult even to envisage how this could be done with a sensor assembly approximately one metre across. We shall, however, deal with difficulties as they arise. For now, there is no reason to consider our ferrite-core coils unsuitable for operation. On the contrary, we consider them very good.

System accuracy in the polarization ellipse semiaxis measurement channel is 20 ppm with a 100 m tow cable. At 8320 Hz, this accuracy can be expressed in terms of signal phase measurement accuracy and therefore timing accuracy. Simple calculations showed that the timing measurement error in the system must not exceed 2–4 ns. We have therefore taken all necessary measures: the system clock oscillator frequency has been set to the maximum supported by the selected components, 272 MHz; the measuring system incorporates continuous phase tracking; and numerous ground experiments have been conducted. Nevertheless, we are still not fully confident that this accuracy will be achieved in a system with such widely separated components.

The circuit and operating principles of the primary-field transmitter were defined only approximately in the technical specification. From the outset, however, we decided to use the resonant transmitter circuit customary in the earliest DIP instruments. In our view, its advantages are substantial gains in energy performance and a significant reduction in unwanted harmonics in the transmitted signal. The transmitter drives a multi-turn loop dipole with a vertically directed moment vector (the Z dipole in D. Fountain’s terminology). In addition to the main transmitter, the system includes two lower-power transmitters whose precise functions we have not yet fully determined. It is clear only that they compensate for the effect of the electrical conductivity of the aircraft and deployment system on the measurements. The exact signals they will transmit will become clear once one of the competing views of the compensation process prevails. The currents of all three transmitters are measured in exactly the same way as the receiver coil signals.

We implemented the control circuits, long-line data transmission and interface to the onboard computer, which are not fundamental to the method, using XILINX programmable logic arrays, already standard in our designs. In fact, this is the most complex and temperamental part of the system and remains under continual improvement. Nevertheless, its conceptual flexibility and power lead us to regard the remote-unit interface, the high-throughput data link and the onboard-computer interface as the basis for an entire family of future developments.

We built the system prototype as early as January 1999; before flight testing, however, one small task remained: to develop algorithms and write a program to correct the measured data, drawing on the invariance of the polarization-ellipse semiaxis ratio, the complex and real sums of squares of the signals measured by the sensing loops, the constancy of the phase of the probing vector, and common sense.

Fig. 3 illustrates the result.
The figure shows 4 operating frequencies
(the lowest is at the top, as usual).

The red curve shows changes in the polarization-ellipse semiaxis ratio
(full scale -0.002).

The blue and green curves are projections of the in-phase vector onto the sensitivity-axis directions.
They illustrate rotation of the system.
Anyone familiar with this work will appreciate how difficult it is to perform such an experiment correctly on the ground. First, power-frequency interference is too strong near the ground; second, it is difficult to obtain a uniform field in the measurement zone. Moreover, we had to take the equipment and the entire field crew to remote areas with a poorly developed power infrastructure. We made a sincere effort to do all this in the laboratory, but were unsuccessful.

The work generally confirmed the validity of the chosen instrument concept. Once measured, the correction parameters remained stable for 12 hours, although their measurement is very complicated. Furthermore, it is quite clear that all these values will cease to be applicable if the towed bird or the aircraft becomes wet, especially if both do. A different approach to determining the correction parameters is needed, one better suited to in-flight measurements. We are currently developing such an approach. Preliminary results are encouraging.

We also note that some components of the future system are already operating in the previous instrument version that we have upgraded. The receiving-loop amplifiers and prefilters, the control circuits and power-amplifier output stages of the auxiliary generators, and several other components have undergone this experimental debugging and, in a sense, flight testing.

The system is currently being prepared for the main phase of flight testing, expected in spring 2000. We hope it will enter service almost immediately thereafter, adding to the already extensive range of equipment and airborne geophysical methods available at GNPP Aerogeofizika.

References

1. D. Fontain. Airborne electromagnetic systems — 50 years of development. (Exploration Geophysics, Vol. 29, 1998)
2. Svetov B. S. Theory, Methodology and Interpretation of Low-Frequency Inductive Electromagnetic Survey Data. Moscow: Nedra, 1973.