According to bionumbers, anywhere between 18h and 72h. There is probably a lot of variation in cell cycle times from one division to another, and more between different cells. The circadian clock can gate the cell cycle and synchronize them, at least partially at periods close to 24 or 48 h.
The problem in answering your question is "How does one define the CSCs?" Assays that use cell surface markers to define a so-called phenotype identify a much larger population than the CSCs, and hence, answers based on such data are invalid. (See Kern & Shibata; J. Hill; Lange et al.). Those using a functional definition (Till & McCuloch), as in the Hybrid Spheroid Assay (Lange et al.), can provide reasonable answers, e.g., 22 -24 hrs as the CSC doubling time, but the cycle time may be a bit shorter as there may be a slowing due to the rate of new niche site production and for the extruded daughter to find a vacant niche.