Measurement of ultra-low power oscillators using adaptive drift cancellation with applications to nano-magnetic spin torque oscillators

This work presents a technique for measuring ultra-low power oscillator signals using an adaptive drift cancellation method. We demonstrate this technique through spectrum measurements of a sub-pW nano-magnet spin torque oscillator (STO). We first present a detailed noise analysis of the standard STO characterization apparatus to estimate the background noise level, then compare these results to the noise level of three measurement configurations. The first and second share the standard configuration but use different spectrum analyzers (SA), an older model and a state-of-the-art model, respectively. The third is the technique proposed in this work using the same old SA as for the first. Our results show that the first and second configurations suffer from a large drift that requires ∼30 min to stabilize each time the SA changes the frequency band, even though the SA has been powered on for longer than 24 h. The third configuration introduced in this work, however, shows absolutely no drift as the SA changes frequency band, and nearly the same noise performance as with a state-of-the-art SA, thus providing a reliable method for measuring very low power signals for a wide variety of applications.