Reduction of noise is a major issue for orthogonal fluxgates, since the main drawback of this kind of sensor is higher noise than parallel fluxgates. Fundamental mode orthogonal fluxgates were successfully proposed to reduce the noise level, thanks to dc bias added to the excitation current. It has been shown that the main reason for noise suppression in fundamental mode orthogonal fluxgates is the reduction of Barkhausen noise obtained with a large dc bias, which makes the excitation current unipolar and avoids reversal of magnetization. Thus, one might believe that a low ac current and a large dc bias are always advantageous. In this paper we will show that this is not always true when this sensor is used in a magnetometer operated in feedback mode. For a given dc bias the 1/f noise will drop when we increase the ac current due to the increment of sensitivity inside the loop. The 1/f noise will reach a minimum level, and then if we further increase the ac current it will rise back up because the increment of Barkhausen noise becomes predominant over the increment of sensitivity. The minimum point is also verified by measurement of circumferential flux-current loop. By properly choosing excitation current parameters, we achieved a 2.5 pT/√Hz noise floor and 7 pT/√Hz noise at 1 Hz. A general rule is finally proposed to easily minimize the noise.

In this paper we analyze the contribution of different sources of noise to the total output noise of a magnetometer based on fundamental mode orthogonal fluxgate. We show that the main contribution to the noise is not given by noise generated in the electronics nor due to the noise of the excitation current in the wire. In fact, a large part of the noise has magnetic origin and it is generated within the core. Therefore, any attempt to reduce the noise of the magnetometer should be focused on the core at a first stage. We give a model, which describes the behavior of both 1/f noise and noise floor as a function of several variables affecting the working mode of the sensor. This model can be used to optimize the parameters of the magnetometer in order to achieve the lowest possible noise. This model can be also useful as a reference to move toward a further reduction of the noise by developing a sensor with better parameters.

Orthogonal fluxgate in fundamental mode has become a competitive sensor for low noise measurement of low frequency magnetic fields. The noise of this sensor has been recently reduced to 2.8 pT/√Hz at 1 Hz; however, it appears that noise lower than 2.5 pT/√Hz at 1 Hz cannot be achieved because of the physical limitations due to the intrinsic noise generated in the core of the fluxgate. In order to decrease the noise below 2.5 pT/√Hz at 1 Hz, Barkhausen noise generated in the core must be reduced. For this purpose, we annealed the amorphous wire used as the core of the sensor in an infrared furnace while a dc current was flowing through the wire. The dc current generated a circumferential field large enough to saturate the wire in a circumferential direction, and by annealing the saturated wire, we increased the circumferential anisotropy. We illustrate the dependence of circumferential anisotropy on dc current in the wire as well as on temperature and time of annealing. We then demonstrate that larger circumferential anisotropy helps reduce the noise of the sensor suppressing the Barkhausen noise when the fluxgate is operated in fundamental mode. As a result, a 1.8 pT/√Hz noise at 1 Hz was achieved with annealed wire, which is far below the 2.5 pT/√Hz minimum noise achievable for sensors with as-cast wires.

Orthogonal fluxgate in fundamental mode has low noise output but its offset is often very large. A method has been proposed to compensate the offset consisting in periodical switching of the dc bias of the excitation and subtraction of the signal obtained with positive and negative dc bias. This method has been already proved to effectively suppress the offset, however no investigation has been made for the temperature drift of the offset. In this paper we show that this method can be used also to compensate the drift of the offset for a wide temperature range: compensated output shows about 120 nT maximum variation of the offset in -90°C to +90°C temperature range, making orthogonal fluxgate in fundamental mode competitive to cheap parallel fluxgate. Moreover the dependence of the open loop sensitivity on the temperature is also investigated: no substantial drop of the sensitivity is observed in the same temperature range.