PTB Berlin · Department 8.2 Biosignals

Magnetometer Characterization

Reliable comparisons between magnetometers need well-defined test conditions and a common set of metrics. As an affiliate researcher at the Physikalisch-Technische Bundesanstalt (PTB) in Berlin, Germany’s national metrology institute, I characterize highly sensitive magnetometers in collaboration with its Department 8.2 Biosignals: from optically pumped magnetometers (OPMs), OPM-MEG systems and other quantum sensors such as NV centers to SQUIDs and emerging non-cryogenic magnetometers, including magnetoelectric (ME) sensors and magnetoresistive (xMR) sensors. The characterization takes place inside the Berlin Magnetically Shielded Room BMSR-2.1, on a dedicated non-magnetic testbench, and follows a framework of 15 parameters in six metric groups.

Open to research collaborations for emerging magnetometers

Testbench with a row of optically pumped magnetometers and their cables inside the white magnetically shielded room. — *Device Alignment Actuator (DALAC) with SERF-OPMs of type FL v3 (FieldLine) inside the BMSR-2.1 at PTB Berlin.*

Characterization parameters: 15; Organized in six metric groups.
Shielding factor at 0.01 Hz: 111dB; ≈ 340,000 in the BMSR-2.1 at PTB Berlin.
Residual field at the room center: < 1nT; Drift below 10 pT per hour.
Angular precision of the testbench: < 0.2°; Two 360° rotation axes, ± 2 mm spatial precision.

Benchmarking environment

All measurements take place in the BMSR-2.1 at PTB Berlin. Since its conversion in 2020, the room consists of eight mu-metal layers and one RF layer. Careful degaussing is essential for reaching the low residual fields listed below.

The BMSR-2.1 is the best-shielded room in Germany and ranks third worldwide:

Paul Scherrer Institute (PSI), Villigen, Switzerland: special shield with six mu-metal layers, designed for measuring the electric dipole moment of the neutron (nEDM)
Harbin Institute of Technology, China: general-purpose shielded room with six mu-metal layers
BMSR-2.1, PTB Berlin

Shielding layers: 8 mu-metal + 1 RF
Shielding factor: ≈ 340,000 (111 dB) at 0.01 Hz
Residual field: < 1 nT at the room center
Field gradients: ≈ 1-2 nT/m
Drift per hour: B_x ≈ 2 pT · B_y ≈ 3 pT · B_z ≈ 7 pT

The door of the BMSR-2.1 opens: the layers of the shielding become visible and give a view of the SQUID system inside (22 s, with sound).

Light-coloured cubic building overgrown with climbing plants under a blue sky; a red cube labelled 2.8 m is drawn on its facade. — *The BMSR-2.1 building at PTB Berlin. The red cube marks the 2.8 m shielded room inside.*

Mu-metal wall panels with the wires of the degaussing coils running in a grid, seen from inside the shielding. — *Inside the shielding of the BMSR-2.1: coils within the mu-metal layers enable proper degaussing.*

Plot of the change of the magnetic field component B y over 800 minutes, rising from about 40 to about 80 picotesla, with a sketch of the room, its test field coil, the reference point and the door. — *Change of `B`_y over 800 minutes at the reference point of the BMSR-2.1. Source: Elzenheimer et al., IEEE Sensors Journal 25 (2), 2432–2455, 2025.*

Testbench

The prototype testbench was built by the sample workshop of PTB Berlin from non-magnetic materials and is mounted inside the BMSR-2.1. Named Device Alignment Actuator (DALAC), it positions single sensors and multichannel arrays at the center of the shielded room and rotates them about two 360° axes (α, β).

Materials: Non-magnetic
Location: Inside the BMSR-2.1, at the room center
Conditions: Residual field < 1 nT, drift < 10 pT/h
Devices under test: Single sensors and multichannel arrays
Rotation: Two 360° axes (α, β)
Precision: Angular < 0.2°, spatial ± 2 mm

Tall dark-red testbench on a round base plate with toothed wheels and a rotating sensor mount at the top. — *The Device Alignment Actuator (DALAC).*

Devices under test

The testbench enables benchmarking of different sensor technologies in single-sensor and multi-sensor arrangements. The photograph at the top of this page shows SERF-OPMs of type FL v3 (FieldLine). Two further examples are given as an impression:

Grey sensor heads with an orange glow hanging side by side from a mounting plate, their cables leading upwards. — *⁴He-OPMs (MAG4Health).*

Single small sensor with a ribbon cable on the perforated mounting plate of the testbench. — *SERF-OPMs of type QZFM Gen-3 (QuSpin).*

Magnetic test fields

Test fields are generated with coils on the walls of the shielded room, driven by a low-noise current source with long-term stability. The resulting homogeneous field is well known and verified with SQUIDs. The back-action of the mu-metal is measured and taken into account.

Amplitude: 10 fT to 85 nT
Frequency: 0.5 Hz to 3.3 kHz
Static offset field: Typically 0-50 nT, up to 3 µT
Current source: Keithley 6221
Field verification: SQUID

DALAC, BMSR-2.1 and SQUIDs together are the key to multi-axis testing of ultra-sensitive multichannel magnetometers.

Metric framework

Every magnetometer is assessed with the same framework of 15 parameters, organized in six metric groups.

Sensitivity & transfer function
- Sensitivity
- Amplitude response
- Phase response
- Bandwidth (−0.5 dB, −3 dB)
Noise & resolution
- Effective amplitude spectral density √S_B (RMS)
- B-field resolution
Ranges
- Linear range
- Operating range
Directionality
- Directivity (angle-dependent sensitivity)
- Orthogonality
- Heading error
Timing behavior & stability
- Time delay
- Settling time
- Medium-term stability (days)
Application-specific metric
- Minimum detectable signal (MDS)

Minimum detectable signal and MDS₂₀

Biomagnetic fields decay steeply with distance. A magnetometer with higher intrinsic noise but smaller standoff can therefore detect signals that remain inaccessible to a lower-noise sensor positioned farther from the source. Yet magnetometer technologies are still compared mainly by their noise floor, while the source-to-sensor distance is rarely made explicit.

The minimum detectable signal (MDS) is a figure of merit derived from a rotating magnetic dipole model. It combines two properties that are usually reported separately: the standoff of a sensor and its intrinsic noise.

Start from the source distance

The distance r between source and detector is the outer distance R from the source to the case of the device under test plus the standoff r_min of the detector inside the case. The dipole field falls off with 1/r³ (far-field approximation, as model), so small changes in standoff can strongly affect detectability.

$r = R + r_{\min}$
Combine distance with intrinsic noise

For a required signal-to-noise ratio SNR, a noise power spectral density S_B and a bandwidth BW, the minimum detectable signal at the outer distance R follows as a magnetic dipole moment.

$MDS (R) = \frac{4 π}{μ_{0}} \cdot {(R + r_{\min})}^{3} \cdot \sqrt{0.5 \cdot 𝑆𝑁𝑅 \cdot S_{B} \cdot 𝐵𝑊}$
Harmonize for comparison

Fixing the conditions yields MDS₂₀: one number under identical conditions, given in pAm². The three reference conditions are the “20/20/20”: power spectral density at 20 Hz, 20 mm from the source to the case and SNR = 20 as a power ratio (≈ 13 dB).
- R = 20 mm
- BW = 1 Hz
- SNR = 20
- S_B at 20 Hz
Point to a condition to see where it enters the formula.
${MDS}_{20} = \frac{4 π}{μ_{0}} \cdot {(0.02 + r_{\min})}^{3} \cdot \sqrt{0.5 \cdot 20 \cdot S_{B, 20} \cdot {𝐵𝑊}_{1 Hz}}$

MDS₂₀ calculator and MDS plotter

Enter the stand-off distance and the noise of a magnetometer to obtain its MDS₂₀ and to compare its MDS(R) curve with the reference devices of the BIOMAG 2026 handout.

MDS₂₀ for different technologies (single axis)

#	Device / system	`r`_min [mm]	Noise ASD (typ. best axis)	MDS₂₀ [pAm²]
1	SQUID (Research, SQUID-MEG, PTB)	12.9	0.28 fT/√Hz ¹	0.32
2	SQUID (Research, Small SQUID, PTB)	8	3.5 fT/√Hz	2.43
3	SERF-OPMs (Neuro-1, QuSpin)	6.5	10 fT/√Hz ²	5.88
4	SERF-OPMs (HEDscan, FieldLine)	5	15 fT/√Hz ²	7.41
5	SQUID (Research, VMS 304, PTB)	28	2.5 fT/√Hz ³	8.74
6	⁴He-OPMs (MAG4Health)	4	30 fT/√Hz ^*	13.11
7	SQUID-MEG (Commercial, CTF)	20	7 fT/√Hz ^*	14.17
8	xMR (Nivio, TDK)	15	0.8 pT/√Hz ^*	1085
9	MI Sensor (DJ, Aichi Steel Corporation)	6.5	2.1 pT/√Hz ^*	1236
10	NV Center (Research, GSI & JGU Mainz)	0.18	13 pT/√Hz ⁴	3.38 · 10³
11	Converse ME Sensor (Research, CAU Kiel)	15	15 pT/√Hz ⁵	20.34 · 10³

BW = 1 Hz · R = 20 mm · SNR = 20 (13 dB). Values from the BIOMAG 2026 handout, doi:10.38071/2026-01501-6. * Reported values. ¹ Storm et al., doi:10.1063/1.4976823. ² Elzenheimer et al., doi:10.1016/j.measurement.2025.120140. ³ Elzenheimer et al., doi:10.1109/JSEN.2024.3491164. ⁴ Omar et al., doi:10.48550/arXiv.2601.18843. ⁵ Elzenheimer et al., doi:10.1109/JSEN.2023.3237910.

Interpreting the trade-off

Near source

Standoff dominates. This favors low-standoff sensors such as OPMs, despite their higher noise floors.

Deeper source

Noise performance dominates. This favors low-noise sensors such as SQUIDs.

A lower MDS₂₀ means better source detectability, which makes MDS₂₀ a guide for sensor selection. In a real application, the sensor ranking depends on the source model and the source depth.

Open-access publications

The testbench, the metric framework and the minimum detectable signal are described in detail in the following open-access publications:

JournalOpen Access

Key Metrics and Experimental Test Bench for Assessing Highly Sensitive Magnetometers in Research

E. Elzenheimer, S. Knappe-Grüneberg, J. Zerfowski, W. Evans, F. Grüneberg, M. Höft, S. R. Soekadar, S. Robinson, and J. Voigt

IEEE Sensors Journal, vol. 25, no. 2, pp. 2432–2455, 15 January 2025.

DOI: 10.1109/JSEN.2024.3491164

JournalOpen Access

Characterizing Timing Parameters in Commercial SERF-OPM Multichannel Systems for Biomagnetic Field Sensing

E. Elzenheimer, H. Matz, J. Zerfowski, P. Anders, M. Höft, R. Rieger, S. R. Soekadar, S. Robinson, and S. Knappe-Grüneberg

Measurement, vol. 263, 120140, 3 March 2026.

DOI: 10.1016/j.measurement.2025.120140

Conference / WorkshopOpen Access