I've put together a heliostat using 12V 28BYJ-48 steppers. It's not entirely accurate as things go, but acceptable - a full degree of looseness in the azimuth, for example. But this is a small project.
Anyway, it seems to track the sun fairly well; in heliostat mode, I have it configured to reflect at a target directly due south, on the horizon (zero degrees in both axes). It seems to track fine in the mornings, but some time after noon and progressing into the evening the reflection drifts slowly upwards as the sun goes down. The gear ratios should be identical for both axes, and there doesn't seem to be a similar issue in the azimuth - it's loose, but the reflection stays more or less in the same location.
Any ideas? It could be some friction in the design of the altitude motor, preventing it from moving accurately, I guess, but if there are any other causes it would be good to rule them out as well.