How does a fluxgate gradiometer work

#|

The gradiometer instruments directly measure variations in the magnetic gradient which are then interpreted to produce maps showing possible sub-surface archaeology. On very large sites arrays of single caesium vapour sensors are now routinely used in towed cart and sled-mounted systems to measure variations in the Earth's total magnetic field.

In this case the magnetic gradient is derived from the field data and then interpreted to produce maps showing possible sub-surface archaeology.

The depth of investigation using the gradiometry technique can be up to 2m depending on the nature of possible sub-surface targets and the survey instrument used. Twin fluxgate gradiometer Photo: Kevin Barton. Practicalities, pitfalls and new developments in airborne magnetic gradiometry. First Break , 22 , 59— Nabighan, M.

Metalliferous mining geophysics — sate of the art in the last decade of the 20th century and the beginning of the new millennium. Geophysics , 67 , — Schmidt, P. Advantages of measuring the magnetic gradient tensor. Preview , 85 , 26— The magnetic gradient tensor: its properties and uses in source characterization.

The Leading Edge , 25 , 75— Veryaskin, A. Magnetic gradiometry: a new method for magnetic gradient measurements. Sensors and Actuators A: Physical , 91 1—2 , — New archaeological discoveries through magnetic gradiometry: the early Celtic settlement on Mont Lassois, France. The Leading Edge , 25 , 46— A three-layer sandwich form, glued together, to form a long central cavity to accommodate the strip core, is suggested.

The picture here shows just such a coil made from a business card. It has a flattened H shape, the bar dimension being 45x10mm to take the winding. This is to locate the core centrally with overlap at the ends to ensure complete magnetization. This winding is scramble-wound by hand, covering about three layers; it should be evenly distributed along the length. A quick rough and ready test is to power the coil with core fitted from a variable DC power supply, at the same time measuring the magnetic field created at the end.

The core saturation point is clearly shown as a slope change at about 30mA. The voltage used for this test was limited to about 5V to avoid overheating this 55 Ohm coil. The resistance could be beneficially reduced with thicker wire than the 0. This can be adjusted to look like the Excitation Current of figure 1 above, by varying the frequency and voltage drive, V.

To reach core saturation, the frequency must be low enough to allow the coil current to reach core saturation well inside each half cycle. The coil example above had an inductance of about 4mH, measured. Meaning it will take about 0. All these parameters, including coil turns, can be adjusted to optimize the design.

It should also be noted that a twin core device needs to have well-matched cores and coils, for good elimination of odd harmonics. A gradiometer will need two fluxgates, so possibly as many as four cores.

The EAS strips are likely to be reasonably well-matched, at least in one tag, and hopefully from one tag to another of the same manufacturer, but not perhaps between manufacturers.

The coil can be used to measure individual strips with some imagination. The width of the strips is quite large for fluxgate use, with a high mass, potentially affecting noise and resolution. Only testing will reveal performance. Suffice it to say that smaller cores would probably improve things, and could be addressed by cutting the strips into thinner pieces, which is possible with a sharp pair of scissors.

However, it might now be difficult to maintain good matching. Also note that only low temperature mechanical working of these materials is advisable. Of course different sizes will require a revised coil design, which should fit as closely to the core as possible, and appropriate electronics adjustments. The use of mu-metal and similar alloys has been somewhat ignored thus far, but a quick comment is in order. To prepare, for example, wire cores of perhaps 0. After this, mechanical damage such as bending or cutting will degrade them, so they need to be protected until installed into the fluxgate.

It is possible to achieve some limited success by heating these cores in a naked flame, but that will require considerable experimentation. Amorphous material is much more tolerable of mechanical damage, so it can be bent and cut without too much concern.

The condition of the strips discussed above is unknown, but may well already be in this state for its EAS application. For use in a fluxgate the cardboard coil former may work for a quick prototype, but is not a good long-term prospect for repeatable performance, which requires well defined, matched and stable coil forms. One possible approach is to employ a similar design to the cardboard, but using thin Alumina or Macor sheet, possibly even thin plastic. While the latter is the least preferred choice, this general approach could be used to produce a twin core fluxgate by bonding two coil assemblies together.

Application of varnish to the coils can also help improve stability, but is difficult to optimize in terms of flexibility and expansion. Winding directly over the paired coil assemblies as in the Vacquier design, figure 2, is possibly the easiest approach. The coefficient of expansion of this coil with temperature is the primary factor controlling the temperature coefficient of the fluxgate, as it expands, moving the turns apart creating a negative temperature coefficient.

This is probably of less concern for a gradiometer than a fluxgate used alone, because the differencing should reduce the effect. The coil will again need to be many hundreds of turns, subject to optimization.

The use of a transformer drive from the electronics could eliminate the need for this extra coil, for the inconvenience of this extra custom component. Once built and connected to the electronics, failure to operate properly can often be traced to reversed or otherwise wrongly connected coil connections. On a more practical note it is all too easy to incorporate materials into the fluxgate design which have small and not very obvious magnetic moments, which can then introduce mysterious offsets.

A prime candidate for this must be connection pins, where plating often conceals a magnetic nickel layer; test items with a Neodymium Iron Boron magnet.

Good results for excitation of the core can be obtained with a square wave, and this is certainly easier to generate than a sine, so is to be recommended. Digital generation of square wave excitation, together with a 2f reference frequency, is conveniently accomplished by means of a microcontroller or a few discrete logic devices. To ensure good low noise, the core material should be driven well into saturation by 10 to times its saturation field.

Easier capacitive drive can also be used with a feedback coil, provided that low leakage capacitors are chosen to avoid spurious and variable offsets. To reiterate, the voltage drive to the fluxgate together with the excitation coil inductance, determine the current ramp rate in the coil, and therefore the maximum frequency attainable. If the frequency is too high for the cores to reach saturation, the device will simply not work at all. If too low it will result in excessive power consumption and possible overheating.

The drive voltage should therefore be chosen to suit the frequency desired, with attention paid to the saturation point of the core within the cycle. One benefit to operation at higher frequencies is the ability to follow rapid field changes, however this is not usually required for these field levels, especially for geomagnetic work of this nature. In this context one advantage of using a microcontroller to generate the waveforms is the ability to easily adjust the phase shift between the f and 2f outputs for better control of the demodulation.

The sense winding is normally fed into a preamp with mild tuning at the 2 nd harmonic of excitation frequency. In practice this can cause problems because of pulse stretching of the voltage spikes passing through, and a broader band amplifier with a reasonably high slew rate can help avoid this. Demodulation is usually accomplished with a phase-sensitive detector, typically a CMOS analogue switch, following the preamp. It is important to choose a switch with low charge injection to avoid further offset problems.

The switch drives an op-amp integrator, which needs to be a low offset, low temperature coefficient device. The integrator time constant is chosen as a compromise between response time and noise performance. The feedback voltage to current converter can be as simple as a resistor, or an active device such as a transconductance amplifier can be employed. When a resistor is used, its temperature coefficient will be added directly to the magnetic field measurement, so care needs to be taken here.

In addition, should the voltage from the integrator be taken as the field output, the proportion of voltage across the feedback coil will be subject to the large temperature coefficient of copper. This can be taken out in software or by the use of additional analogue circuitry.