That's an interesting model, because mosquitoes are vectors for serious illnesses, so are pretty well studied. One team of scientists are working on genetically altering mosquitoes in Africa to make them unable to transmit the parasite that causes malaria. As the mosquitoes breed, it spreads through the population. In an interview, the lead researcher stated:
In the lab, we saw the mutation spread to more than half the population of mosquitoes we had living in cages. This happened in a time span of 12 to 16 generations of mosquitoes (each generation being between 18 to 30 days in the wild). So it's quite efficient.
12 to 16 generations for one mutation to spread to half of a contained population. In human generations, 16 = 320 years, and that doesn't take into account fertility rates, miscarriages, etc. (Not a problem for mosquitoes, I presume).
How many mutations would it take to make two humans incapable of reproducing without endangering their own health or fertility? Excluding all of the harmful mutations that might just result in disease or decreased fertility? And add to that that the males need to change as well as the females? No one knows.
Imagine a cystic-fibrosis type mutation (one gene) that increased the viscosity of cervical mucus such that normal sperm could not fertilize the egg. There would need to be another mutation in males that would make sperm stronger, or secrete an enzyme that breaks down mucus, or something to that effect. However, look at Cystic Fibrosis. People used to die before adulthood. Most men with CF are sterile. Women with CF, living longer now because of better care and treatment, are getting pregnant, but many have a decreased lifespan post-pregnancy, and almost all the babies are premature. You can't predict "when" a series of beneficial mutations so dramatic as to change a being so dramatically. The answer might be never.
Try to give me a single or a few mutations that would prevent the ability of people from one population and another to mate. Say one mutation led to a smaller pelvis. Most women with that mutation would die in childbirth before a mutation arose to make smaller babies (which would most likely be a deleterious mutation as well).
I am not a geneticist (which is really the person who should be answering this question) but I think the odds of this scenario ever occurring are very low. I just can't imagine mutations making reproduction impossible between two populations being beneficial and selected in one population rather than harmful. Keep in mind that different humans from populations that were isolated for many thousands of years have no problem reproducing.