The proliferation in low-cost Internet-of-Things (IoT) devices with narrowband radios, such as Bluetooth Low Energy (BLE), has facilitated a range of applications, including keyless entry, remote patient monitoring, and asset tracking. Accurate distance estimation is essential in these applications. Phase-Based Ranging (PBR) has been highlighted as a key technique in Channel Sounding (CS), which was introduced in the recently released BLE Core Specification version 6.0. The oscillator in these IoT devices serves a vital function as their signal generator, frequency synthesizer, and internal clock. Because of cost-saving requirements, typically these oscillators do not provide high frequency stability. As a result, synchronization in devices is error-prone, leading to phase offsets in the PBR approach, and the conventional Two-Way Ranging (TWR) method applied in the PBR compensates for only a limited portion of the phase offsets. This paper introduces a novel TWR method based on a mirrored tone exchange together with a corresponding new tone-exchange scheme, achieving improved mitigation of the phase offsets. To address the ambiguity issue of the PBR, which still exists in the new procedure, a hybrid framework is developed. Simulation and measurements demonstrate that the new method effectively mitigates a substantial portion of the residual phase offset and improves the precision of the ranging results by nearly an order of magnitude for outdoor scenarios and by approximately 27% for indoor scenarios. For repeated measurements targeting the same precision of the averaged distance estimate, the improved precision corresponds to estimated total energy savings of about 89% in the anechoic chamber and 63% to 90% in the outdoor scenarios, despite the increased energy required by each individual novel TWR estimate.