1 Introduction

Martian magnetic field plays an important role in investigating the Martian interior structure and evolution, and monitoring the space environments (Johnson et al. 2020). Martian magnetometry measurements can make us to acquire important observation data to study the space distribution of Martian crust magnetic field, the origin and evolution of magnetic field, the relation to geological structure, and the interaction between solar wind and Martian ionosphere. The expected research results would help us to understand Martian internal structure, space electromagnetic environment, and the evolutionary history of atmosphere and climate.

The magnetic explorations of the Mars have been accomplished by magnetometers on the flyby spacecraft and the orbiters since 1960s. The payload information of Martian magnetic field measurement based on the effective magnetic field data is shown in Table 1 (Smith et al. 1965; Dolginov et al. 1973, 1976; Acuña et al. 1998, 1999, 2001; Auster et al. 2008; Banfield et al. 2019). Mariner 4 flew by Mars on July 15, 1965. The first magnetic measurement is implemented at a distance of 3.9 Mars radii (Smith et al. 1965; Riedler et al. 1989). MGS produced an unprecedented global map of crust magnetic fields at around 400 km altitude (Connerney et al. 2005). MAVEN performed more complete study of the magnetic field magnitudes at a lowest target altitude of 125 km (Connerney et al. 2015). The magnetometer installed on the Insight Lander, which is launched in 2018, brought opportunities to directly measure the magnetic field on the surface of Mars. It was found that the field on the surface is ten times stronger than expected and the data implied it was coming from nearby sources (Johnson et al. 2020).

Table 1 The missions of Martian magnetic field measurement
Full size table

The magnetometer mounted on Insight lander gives new knowledges of the Martian surface magnetic field. However, due to the limitation of the single point measurement, the sources of some new discovering phenomena are open questions. The RoMAG payload of TianWen-1 mission will provide the first mobile magnetic field measurements on the Martian surface. High space resolution multi-point mobile measurements of RoMAG could make us to obtain the fine-scale profile of Martian crust field along the rover track, and also could investigate magnetic field distribution around a magnetic target in detail. Combining the two-site observation by RoMAG and Insight and the Obiter magnetometer observation of TianWen-1, we can add constraints in modeling the global crust magnetic field, explain the interior structure of Mars, and study the features of the Martian magnetic waves on surface. It would provide an unprecedent opportunity for Martian magnetic field investigation.

2 Science Requirements

As one of the important payloads of the Chinese Mars Rover, RoMAG will provide high space resolution magnetic measurements of surface. The major unresolved questions are: When and how did the Martian dynamo die? What cause the asymmetry in Martian crust magnetic field strengths? How does the magnetic field interact with the solar wind? The science objectives are listed as follows:

  1. (1)

    To obtain fine-scale structures of crustal magnetic field based on the mobile measurements on the Martian surface.

  2. (2)

    Combining with Chinese Mars Orbiter observation, to study the ionospheric conductivity by inverting the ionospheric dynamo current.

  3. (3)

    To study the Martian internal structure.

RoMAG consists of two identical tri-axial fluxgates (Fig. 1). The two sensors are mounted on the top and bottom of the mast of Rover, respectively. The distance between them is 67.5 cm. Above the magnetometer sensor 1, a Multispectral Camera (MSCam) and a Navigation and Terrain Camera (NaTeCam) are mounted. Other payloads, Rover-mounted Subsurface Penetrating Radar (RoSPR), Mars Climate Station (MCS), and Mars Surface Composition Detector (MarSCoDe), are also installed on the Mars Rover. The six instruments are designed for a detailed characterization of the Martian surface. The sensor installation environments and the above-mentioned science objectives enforce some basic requirements to the magnetometers. RoMAG requirements are listed as follows:

  1. (1)

    The range of magnetometer needed to exceed the sum of the Mars surface magnetic field and the Mars Rover magnetic field. Therefore, we chose a range for the magnetometer of \(\pm65{,}000~\text{nT}\).

  2. (2)

    Sample rate sufficient to provide temporal resolution of 32 seconds or better.

  3. (3)

    The diurnal variation of the magnetic field on the surface is about 30 nT (Johnson et al. 2020). In order to trace the actual field variation of small changes, a resolution of at least 0.1 nT is required.

  4. (4)

    The magnetometer noise requirement is \(0.01~\text{nT}/\surd \text{Hz}\) at 1 Hz.

Fig. 1
figure 1

The cartoon of Chinese Mars Rover. Two tri-axial magnetometers are mounted on the top and bottom of the mast of Rover, respectively

Full size image

The performance parameters of magnetometer sensors are summarized in Table 2.

Table 2 Magnetometer sensor performance
Full size table

3 Instruments Description

Tri-axial ring core fluxgate sensors have been used for various space missions. The RoMAG system is designed based on the sensor system developed for THEMIS (Auster et al. 2008), VenusExpress (Zhang et al. 2006), the ROSETTA lander Philae (Auster et al. 2007), BepiColombo (Glassmeier et al. 2010), and MASCOT (Herčík et al. 2016) by the University of Braunschweig in cooperation with the MAGSON GmbH in Berlin. The sensors and electronic units have been manufactured at the Institute of Geology and Geophysics (IGG), Chinese Academy of Sciences (CAS).

3.1 Principle of Operation

The three-axis fluxgate magnetometer (FGM) is a ring core system with vector-compensated consisting of sensor and digital electronics. The measurement principle of a fluxgate magnetometer is based on the high permeable soft-magnetic ring cores being driven into deep saturation by an excitation current with specific frequency. An existing external magnetic field distorts the symmetry of the magnetic flux within the core. A signal in even harmonics of the excitation frequency can be generated and it is proportional to the external field. The sense coil system is designed for the selective measurement of the excitation frequency’s second harmonic. A feedback current in the enclosing Helmholtz coil system is generated to compensate the external field at the ring core position. The single component values of the external field are calculated utilizing the field-proportional feedback current values and the remaining field at the core position.

3.2 Fluxgate Sensor

The fluxgate sensor consists of excitation coils, pick-up coils, feedback coils, thermistor and skeleton. The excitation signals are driven by two crossed ring cores with a diameter of 13 and 18 mm. The excitation coils are surrounded by three pick-up coil pairs for the X, Y and Z components. A vector compensating feedback coil system is used to keep the ring cores in zero field. In order to guarantees a wide temperature range and a temperature independent axis alignment, the material of the feedback system is choosing to have a similar thermal expansion coefficient. As shown in Fig. 2, the sensor coil system is encapsulated in a cylindrical polysulfide shell and connected to the electronics via a long pigtail harness (286.4 cm for Sensor 1, 167.7 cm for Sensor 2). The two sensors with their covers and pigtails have a mass of 259.9 g and 156.5 g, respectively.

Fig. 2
figure 2

The structure (left) and the packaged appearance (right) of RoMAG sensor

Full size image

3.3 Sensor Electronics

The block diagram of the RoMAG sensor electronics is illustrated in Fig. 3. The hardware of electronics unit is composed of excitation circuitry, preamplifier, feedback current sources, high precision ADCs/DACs and temperature circuitry.

Fig. 3
figure 3

Block diagram of the Digital Fluxgate

Full size image

Instrument control, digital data processing, instruction parsing and data processing are implemented in a Field-Programmable Gate Array (FPGA). The temperature circuitry is designed to collect the temperature data, which is used to calibrate thermal drift coefficients, of sensor and electronics board.

The sensor electronics generates an excitation AC current. The fundamental excitation frequency is about 9.6 kHz and the sampling rate is 80–160 kHz. The output signal from the pick-up coil is amplified and then sampled and converted by the ADC and transmitted to the FPGA. In FPGAs, fast digitization of the sensor AC-signal and the subsequent data processing are implemented. Prior to the data processing, a band-pass filtering around the second harmonics is performed to alleviate the aliasing problem. The feedback field improves the overall linearity and stability of the magnetometer. Feedback signals are produced by a set of two 12-bit digital-to-analogue converters (DACs) and a separate pair of feedback coils (Helmholtz coils) per sensor axis. The sense values from the analogue-to-digital converter (ADC) and the feedback values are continuously used for calculating the magnetic field values. Using the D/A digital compensation, the aliasing signal and bias field is gradually minimized and the dynamic compensation is then accomplished. Meanwhile, the instrument noise is then reduced. The noise level of the magnetometers is shown in Fig. 4. The noise levels at 1 Hz on the two magnetometers are shown in Table 3, which well satisfy the scientific requirement.

Fig. 4
figure 4

Noise in each axis as a function of frequency for the two magnetometers. X-axis noise is in red, y-axis in green, and z-axis in blue. The requirement is in orange color

Full size image
Table 3 The noise level for two magnetometers
Full size table

The instrument electronics unit is placed in the payloads management unit box together with other payload electronic boards. The electronics board uses \(\pm12~\text{V}\), \(\pm5~\text{V}\), \(\pm3.3~\text{V}\) power provided by the payloads management unit and converts these voltages to those required by the electronics. The electronics unit (Fig. 5) is \(23.3\times 28.1\times 1.4~\text{cm}\), the power consumption is 5.48 W, and its mass adds up to 628.9 g.

Fig. 5
figure 5

The RoMAG electronics with the two sensors

Full size image

3.4 Ground Test and Calibration

In order to qualify the sensors for flight, the instrument space-qualification tests, such as the thermal cycle, thermal vacuum, vibration, shock, electromagnetic compatibility (EMC), have been completed in preflight. To verify the performance of the RoMAG in nominal and extreme temperature, the sensors were thermally cycled 12.5 times over the temperature range \(-65~^{\circ}\text{C}\) and \(+60~^{\circ}\text{C}\).

In preflight, RoMAG was calibrated in the calibration facility of the Weak Magnetic Metering Station in the city of Yichang, Hubei Province, China. This laboratory for magnetometer calibrations located on a mountain outside the Yichang city and far away from major magnetic disturbers.

Figure 6 shows the calibrations facility with a large Helmholtz coil system. The test of the dependency of instrument parameter on temperature was also performed. The sensor and electronics boards were put inside a temperature box. The temperature has been varied between \(-60~^{\circ}\text{C}\) and \(+60~^{\circ}\text{C}\). The sensor was placed in a coil system, in which the field can be changed by varying electric current. By generating test fields, the sensor offset can be checked. The RoMAG sensor was placed in a thermal box (Fig. 7), allowing the sensor to be tested at various temperatures in a range from \(-60~^{\circ}\text{C}\) to \(+60~^{\circ}\text{C}\).

Fig. 6
figure 6

The Calibration Facility: a 2.4 m large Helmholtz coil system

Full size image
Fig. 7
figure 7

The thermal box with a coil system

Full size image

The calibration procedure includes the determination of the instrument performance in terms of the scale, offset, and internal misalignment. The calibration parameters are given in Table 4. The magnetometer output \(\boldsymbol{B}_{fgm}\) can be expressed as a linear response of the true field vector \(\boldsymbol{B}_{tr}\) (Olsen et al. 2003)

$$ \boldsymbol{B}_{fgm} = \boldsymbol{S} \cdot \boldsymbol{P}\cdot \boldsymbol{B}_{tr} + \boldsymbol{O} $$
(1)

where \(\boldsymbol{S}\) is a diagonal matrix of scale factors, \(\boldsymbol{P}\) is orthogonality matrix, \(\boldsymbol{O}\) is offset vector. Based on the calibration parameters, the calibrated field can be computed by applying the calibration matrix to the data measured by the magnetometer sensor after removal of the sensor offset.

Table 4 The calibration parameters
Full size table

3.5 Rover-Generated Field Estimation and Compensation

Due to size limitations and mass constraints, RoMAG is accommodated directly to mount on the mast of rover in the vicinity of other payloads and system sub-units. The magnetometers measure the superposition of the ambient field plus the field that produced by different sources on the Rover. It is a challenge to accurately estimate and compensate the rover-generated magnetic field from a variety of sources.

To identify correctly what mechanism or subsystem is responsible for the interference, we conduct specialized prelaunch magnetic tests to identify the most important sources. It is found that there is a strong magnet of micro-motor in vicinity of each sensor. Additional sources are associated with the operation of other payloads. It should be noted that NaTeCam and RoSPR were advised to remain power off during the operation period of RoMAG.

In order to qualify the rover-generated interference, we implement a prelaunch magnetic test in laboratory. As shown in Fig. 8, four testing magnetometers (S3–S6) are mounted on a test support. S3 and S4 are at the same height with Sensor 1, while S5 and S6 are at the same height with Sensor 2. After the rover get out of place when the test is accomplished, S3 and S5 are used to measure the ambient field (absolute value) of the positions of S1 and S2, respectively. During the rover operation period, S3 and S5 are placed ahead the rover. The distance between S3/S5 and S1/S2 is 30 cm. The test support is vertical to the Rover body. That is to say, the distance between S4/S6 and S1/S2 is \(30\cdot \sqrt{2}~\text{cm}\). S7 is used to monitor the time-varying of the ambient field during the rover operation period. In the rover mechanical coordinate frame, z points to the direction of the rover’s motion and y points to the right direction of the rover’s motion.

Fig. 8
figure 8

Prelaunch magnetic test. Four testing magnetometers (S3–S6) are mounted on a test support. S3 and S4 are at the same height with Sensor 1, while S5 and S6 are at the same height with Sensor 2. After the Rover get out of place when the test is accomplished, S3 and S5 are used to measure the ambient field of the positions of S1 and S2, respectively. S7 is used to monitor the time varying of the ambient field during the rover operation period

Full size image

Figure 9 gives the RoMAG observations in the rover mechanical coordinate frame. Each payload operation period is listed in Table 5. It is found that there are strong dynamic interferences during the Multispectral Camera operation period for both sensors. Because Sensor 1 is closer to the Camera than Sensor 2, the dynamic interference for sensor 1 is much stronger than that for sensor 2.

Fig. 9
figure 9

The RoMAG observations (32 Hz) in the Rover mechanical coordinate

Full size image
Table 5 The payloads operation periods
Full size table

Due to different locations, the sensors measure the disturbing sources of the rover in different ways. Therefore, changing rover fields produce different magnetic variations at the locations of the sensors. It would lead to reduction of the relevance between the two-sensor data. In order to investigate the correlation between the magnetic fields at two sensors, we calculate the wavelet spectra of magnetic field components with a period range spanning from 16 s to 150 s for two sensors, respectively. As shown in Fig. 10, during the period of MSCam operation, both sensors show strong magnetic fluctuations. The magnetic spectra of sensor 1 shows broadband spectra signature for all the period range while sensor 2 shows strong spectra at about 140 s. The different spectra signature between the two sensors is probably caused by that the short period dynamic interferences at sensor 2 are attenuated because the distance from MSCam to sensor 2 is larger than that of sensor 1. After that period, both sensors show weak magnetic spectra which indicates small interferences from other payloads onboard the rover.

Fig. 10
figure 10

The wavelet spectra of magnetic field components observed by two sensors

Full size image

In order to study the dynamic interferences at the two sensors more clearly, we also calculate the wavelet coherence between the sensors pair (Fig. 11). Wavelet coherence spectra are computed by using the complex Morlet wavelet as mother wavelet. As expected, all the three magnetic components have low coherence between the two sensors during the interval 14:19–14:50, which coincide with the operation period of MSCam. This is consistent with the magnetic spectra differences between the sensors pair as shown in Fig. 10. In contrast, during the time interval 14:50–15:17, although the magnetic field of both sensors shows weak oscillation (see Fig. 10), their coherence appears to be high. This indicates the magnetic field at the two sensors exhibit similar oscillation characteristics, which may be mostly the natural magnetic signal signatures.

Fig. 11
figure 11

The wavelet coherence spectra for the sensors pair

Full size image

To assure the quality of compensated data, we selected the data with high coherence. We use Eq. (2) to determine if the Rover field changes during a time interval dt, as shown in Fig. 12.

$$ \Delta \boldsymbol{B}_{\boldsymbol{d}} = \left \vert \Delta \boldsymbol{B}_{s1} - \Delta \boldsymbol{B}_{s2} \right \vert < e $$
(2)

where

$$\begin{aligned} \Delta \boldsymbol{B}_{s1} =& \boldsymbol{B}_{s1} (t+1)- \boldsymbol{B}_{s1}(t)\\ \Delta \boldsymbol{B}_{s2} =& \boldsymbol{B}_{s2} (t+1)- \boldsymbol{B}_{s2}(t) \end{aligned}$$

We set \(e\) as 2 nT empirically. If data meet the condition of Eq. (2) and exhibit a duration of at least 10 minutes, it will be selected as high coherence data to be conducted magnetic inference compensation. The select data are indicated by red line in Fig. 12.

Fig. 12
figure 12

The Rover field changes during a time interval

Full size image

The rover-generated fields can be grouped into three types: permanent, induced, and dynamic magnetic fields. The permanent magnetic fields are caused by the permanent magnetism of the ferromagnetic parts of the rover platform. The induced magnetic fields are created in paramagnetic parts by the ambient magnetic field. The dynamic magnetic fields are current-generated field, such as from the power subsystems, solar arrays, batteries, etc. (Acuña et al. 2001).

Applied to both sensors, the magnetic field can be modeled in the payload coordinates

$$ \boldsymbol{B}_{\boldsymbol{m}} = \boldsymbol{S}\cdot \boldsymbol{P}\cdot \bigl[ ( \boldsymbol{B}_{\boldsymbol{a}} + \boldsymbol{K}_{\boldsymbol{i}} \cdot \boldsymbol{B}_{\boldsymbol{a}} +\boldsymbol{B}_{\boldsymbol{p}} + \boldsymbol{B}_{\boldsymbol{d}} ) + \boldsymbol{O}\bigr] $$
(3)

where \(\boldsymbol{B}_{\boldsymbol{m}}\) is measured field. \(\boldsymbol{B}_{\boldsymbol{a}}\) is ambient field. \(\boldsymbol{K}_{\boldsymbol{i}}\) is magnetic induction coefficient as a diagonal matrix. \(\boldsymbol{B}_{\boldsymbol{p}}\) is the residual magnetic field. \(\boldsymbol{B}_{\boldsymbol{d}}\) is dynamic field, \(\boldsymbol{S}\) is a diagonal matrix of scale factors, \(\boldsymbol{P}\) is orthogonality matrix, \(\boldsymbol{O}\) is offset vector.

Equation (3) can be rewritten as

$$ \boldsymbol{B}_{\boldsymbol{m}} = \boldsymbol{K} \cdot \boldsymbol{B}_{\boldsymbol{a}} +\boldsymbol{b} $$
(4)

where

$$\begin{aligned} \boldsymbol{K} =& \boldsymbol{S}\cdot \boldsymbol{P} + \boldsymbol{S}\cdot \boldsymbol{P}\cdot \boldsymbol{K}_{\boldsymbol{i}}\\ \boldsymbol{b} =& \boldsymbol{S}\cdot \boldsymbol{P}\cdot ( \boldsymbol{B}_{\boldsymbol{p}} + \boldsymbol{B}_{\boldsymbol{d}} ) + \boldsymbol{S}\cdot \boldsymbol{P}\cdot \boldsymbol{O} \end{aligned}$$

Solve the linear equation (4), we can obtain the \(\boldsymbol{K}_{\boldsymbol{i}}\) and the sum of \(\boldsymbol{B}_{\boldsymbol{p}}\) and \(\boldsymbol{B}_{\boldsymbol{d}}\). Table 6 gives the calculated value of \(\boldsymbol{K}_{\boldsymbol{i}}\) and \(\boldsymbol{B}_{\boldsymbol{p}} + \boldsymbol{B}_{\boldsymbol{d}}\) at each install position of two sensors in the rover coordinates. Figure 13 gives the comparison of magnetic field measurements data after compensation with ambient field. The RMS errors of compensation algorithm for three components are all less than 1.5 nT.

Fig. 13
figure 13

A comparison of magnetic field measurements data after compensation with ambient field. Red: data after correction (1 Hz), black: ambient field

Full size image
Table 6 The magnetic interference parameters
Full size table

However, there are many additional disturbances to be further taken into account. For example, due to some engineering factors, the solar arrays were not installed on the rover and MarSCoDe wasn’t in operation in the preflight magnetic test. Therefore, the disturbances generated by these sources should be carefully checked, identified, and eliminated in future.

4 Summary

The RoMAG will provide the first mobile vector magnetic field observations at sample rates of up to 32 Hz and obtain fine-scale structure of crust field. Two magnetometers are installed on the top and bottom of the mast of Rover, respectively. Due to the fact that RoMAG is placed in the vicinity of micro-motors, the magnetic interference compensation technique is developed to remove the magnetic field disturbances. However, the disturbances generated by the solar arrays and surface composition instruments operation should be carefully identified and eliminated during in-flight calibration processing. Together with the magnetic field observations by the magnetometer on the obiter of TianWen-1, the RoMAG observation data will help us to understand Martian internal structure, space electromagnetic environment, and the evolutionary history of atmosphere and climate.