In the operation of a SQUID, four basic superconducting phenomena are used: (i) complete loss of electric resistance at temperatures below the critical temperature, (ii) perfect diamagnetism having no magnetic flux inside the superconductor, (iii) quantization of magnetic flux in a superconducting ring, (iv) Josephson effect. Most of the present MEG systems use low-temperature Nb-based SQUIDs. Nb has a superconductive transition temperature of about 9.2 K, and is a refractory and reliable material against repeated thermal cycling between 4.2 K and room temperature. And the noise characteristics of SQUIDs made from Nb/AlO_x/Nb Josephson junction show low leakage current in the low frequency range, which is an important requirement for measuring low-noise MEG measurement. The fabrication technology of Nb SQUID sensors is now well established and fabrication of many sensors on Si-wafers can be done (Lee et al. 1999). The typical size of the SQUID chip is about 10 mm², including pads for wire bonding (Al and Nb).

Figure 2 shows the principle of measuring an MEG signal using a SQUID. For the effective pickup of magnetic field signal, a superconductive flux transformer is used, consisting of a pickup coil of a much larger diameter than the SQUID loop and a multi-turn input coil integrated directly on the SQUID loop. When a magnetic field is applied to the pickup coil, a screening current is generated in the superconductive flux transformer circuit, and this current is converted into magnetic flux through the input coil and magnetic coupling with the SQUID loop. In a typical design of flux transformer and SQUID, about 0.5 nT of magnetic field in the pickup coil corresponds to flux transfer of 1 Φ₀ into the SQUID loop.

Typical dimensions of SQUID loops are about 0.1 mm. Thus, to increase the detection efficiency, pickup coils of a diameter larger than 10 mm is usually needed. Among various types of pickup coils, magnetometers or hardware first-order gradiometers are now used in present MEG systems. Figure 3 shows examples of pickup coils.

Magnetometers have the best sensitivity to both deep and shallow sources. At the same time, however, it is more vulnerable to external noises. The optimum choice of a pickup coil depends on the details of the measurement condition; thickness of the MSR, strength of environmental noises, and signals to be measured. Inside a thick MSR or in a quiet location, magnetometers as the sensing element are preferable. In urban clinics or laboratories, there is often a limitation in the thickness or weight of the MSR. Thus, moderate or medium-thickness MSRs, in combination with gradiometers, is the best combination. Generally speaking, axial gradiometers have longer baselines than planar gradiometers, so that it has better sensitivity to deep sources than planar gradiometers. For shallow sources, planar gradiometers have better sensitivity when the axis of current dipole (y-axis in Fig. 3c) is perpendicular to the field derivative direction of the gradiometers (x-axis in Fig. 3c).

A planar gradiometer can be made on a single wafer, that is, the planar pickup coil can be integrated on the same wafer as the SQUID loop. For example, in the Neuromag system, 2 perpendicular planar gradiometers and magnetometer are integrated on the same element (wafer) (Parkkonen 2010).

A possible disadvantage of the axial gradiometer is the relative complex process in assembling the axial gradiometer, and it needs superconductive connection (bonding) between pickup coil wires and input coil pads. For the superconductive connection between the pickup coil and the input coil, usually superconducting Nb blocks or strips with screw terminals are used. To eliminate pickup area of the magnetic field due to the superconducting connection structure, Nb blocks are sometimes shielded using a superconducting tube (Ketchen 1987; ter Brake et al. 1992; Dössel et al. 1993). Since this superconducting block and tube introduces distortion of magnetic fields, and they are installed at a sufficiently large distance from the compensation coil of the gradiometer to maintain the balancing of the gradiometer against the external fields, as shown in Fig. 4a. Increased length of the gradiometer requires higher level of liquid He to keep both SQUID and pickup coil superconducting. In order to increase the refill interval of liquid He, it is desirable to position the SQUIDs as close as possible to the gradiometers. Recently, some improvements were made to remove the shielding tube, so that the SQUID is positioned at about 20 mm from the compensation coil. But, the stray pickup area due to the superconducting connection structure generates imbalance of roughly few percent. A novel method to simplify the superconductive connection method and to reduce the stray pickup area is direct bonding of Nb wire between pickup coil wires and input coil pads. Thus, the fabrication process of the gradiometer became simpler and the total length of the gradiometer can be shortened (Lee et al. 2009). Considering that the residual fields inside the MSR are not highly homogenous, the intrinsic balancing of the gradiometers needs to be as large as possible with simple fabrication process.

The voltage output of a SQUID is periodic like a sinusoidal function and nonlinear with respect to the input flux as shown in Fig. 5a. To get a linear response against the input flux signal, a special operation scheme, called flux-locked loop (FLL), is used. In the FLL operation (Fig. 5b), the flux in the SQUID loop is locked at a constant point using a negative feedback circuit, and the feedback voltage (or feedback flux through the feedback coil) is measured as the final output. Figure 6 shows a schematic circuit diagram of standard-type FLL operation. Since the amplitude of SQUID output against input MEG signals is small, care should be taken in detecting the SQUID output. For example, a typical MEG signal, say, 100 fT corresponds to about 0.2 mΦ₀ in the SQUID loop, which generates a voltage signal of 20 nV (for a typical flux-to-voltage transfer of 100 μV/Φ₀). To detect this level of voltage signal, careful design of a room-temperature preamplifier is needed. For economic operation of multi-channel SQUIDs for MEG systems, simple and compact room-temperature readout electronics are required. For the simple structure of the FLL circuits, output voltages and flux-to-voltage transfers of the SQUIDs should be large enough so that the contribution of preamplifier input noise is negligible in direct readout mode (Drung 1996; Drung and Mück 2004). A double relaxation oscillation SQUID (DROS), based on the relaxation oscillation of a hysteretic SQUID and a reference junction, provides large flux-modulated voltage output and a steep flux-voltage transfer coefficient (Adelerhof et al. 1994; Lee et al. 1999; 2005). One example of a DROS design is shown in Fig. 7, which enables direct measurement of SQUID output using room-temperature preamplifiers, and makes the FLL circuits compact using DC bias current. In the FLL or internal feedback scheme, there is crosstalk between adjacent pickup coils. Induced current in the flux transformer generates a magnetic field, which can be picked up by nearby pickup coils. This stray coupling can be eliminated by using the feedback scheme called external feedback or current feedback loop, in which the current in the flux transfer circuit is kept constant (ter Brake et al. 1986).

Superconductivity is maintained under critical condition, that is, below critical temperature, critical current, and critical field. If the SQUID is exposed to high magnetic field or transient electric pulses, magnetic flux can be trapped in the thin film SQUID. Trapped flux can deteriorate the performance of the SQUID or even make normal SQUID operation difficult. Trapped fluxes can be removed by applying a current pulse of about 1 W, typically 0.1 ~ 1 s duration, to the SQUID. The heater is placed close to the SQUID loop. During heating, the temperature of SQUID temporarily rises to just above its superconductive transition temperature, at the expense of temporary slight evaporation of liquid He.

The voltage output of a FLL circuit is too small to be measured directly by analog-to-digital converter using a computer. Thus, intermediate amplifier and filters are used which consist of high pass, low pass, power-line elimination filter and amplifier. Typical cutoff frequencies for high and low pass filters are 0.01 ~ 0.1 and 100 Hz, respectively. Amplification is 100 or 1,000 times. Use of analog filters makes acquisition easier and real-time monitoring of the acquisition process easier. However, analog filters can change the shape of the signal waveforms and introduce phase distortions. In addition to the separate space needed to house the analog signal processing (ASP) circuits, an ASP usually consumes more electrical power than the FLL circuit does. Recent MEG systems digitize the output of FLL directly, and measure signals through optical fibers. Advantages of using optical readout are (i) reduced power consumption, (ii) compact electronics and reduced installation space by eliminating the ASP box, (iii) elimination of electric interference from outside of MSR, and (iv) increased dynamic range of FLL output. Figure 8 shows the comparison of SQUID output measurement systems with conventional ASP circuits (Fig. 8a) and ASP-free readout system (Fig. 8b).