This paper introduces an enhanced speed self-sensing method that tracks the stator flux linkage in induction machines at very-low speeds. Volt–second sensing technology is used to mitigate the inverter nonlinearity, which is the major source of flux estimation error in the very-low-speed range. By integrating volt–second sensing in the Gopinath style closed-loop flux observer, the rotor speed information with more precision and lower noise is extracted from the estimated stator flux. Experimental results show that the proposed self-sensing method extends the low-speed operating range, reduces the speed and torque ripple, and increases the disturbance rejection capability compared with the traditional self-sensing method based on the voltage integration method and the back electromotive force (EMF) tracking method.