We present the design process and implementation of fully open-source, ultra-low noise programmable current source systems in two configurations. Although originally designed as coil drivers for Optically Pumped Magnetometers (OPMs), the device specifications make them potentially useful in a range of applications. The devices feature a bi-directional current range of ±10 and ±250 mA on three independent channels with 16-bit resolution. Both devices feature a narrow 1/f noise bandwidth of 1 Hz, enabling magnetic field manipulation for high-performance OPMs. They exhibit a low noise of 146 pA/[Formula: see text] and 4.1 nA/[Formula: see text], which translates to 15 and 16 ppb/[Formula: see text] noise relative to full scale.

Self-oscillating atomic magnetometers, in which the precession of atomic spins in a magnetic field is driven by resonant modulation, offer high sensitivity and dynamic range. Phase-coherent feedback from the detected signal to the applied modulation creates a resonant spin maser system, highly responsive to changes in the background magnetic field. Here we show a system in which the phase condition for resonant precession is met by digital signal processing integrated into the maser feedback loop. This system uses a modest chip-scale laser and mass-produced dual-pass caesium vapour cell and operates in a 50 $$\mu $$ μ T field, making it a suitable technology for portable measurements of the geophysical magnetic field. We demonstrate a Cramér-Rao lower bound-limited resolution of 50 fT at 1 s sampling cadence, and a sensor bandwidth of 10 kHz. This device also represents an important class of atomic system in which low-latency digital processing forms an integral part of a coherently-driven quantum system.

Machine learning (ML) is an effective tool to interrogate complex systems to find optimal parameters more efficiently than through manual methods. This efficiency is particularly important for systems with complex dynamics between multiple parameters and a subsequent high number of parameter configurations, where an exhaustive optimisation search would be impractical. Here we present a number of automated machine learning strategies utilised for optimisation of a single-beam caesium (Cs) spin exchange relaxation free (SERF) optically pumped magnetometer (OPM). The sensitivity of the OPM (T/Hz), is optimised through direct measurement of the noise floor, and indirectly through measurement of the on-resonance demodulated gradient (mV/nT) of the zero-field resonance. Both methods provide a viable strategy for the optimisation of sensitivity through effective control of the OPM’s operational parameters. Ultimately, this machine learning approach increased the optimal sensitivity from 500 fT/Hz to <109fT/Hz. The flexibility and efficiency of the ML approaches can be utilised to benchmark SERF OPM sensor hardware improvements, such as cell geometry, alkali species and sensor topologies.