My mk1 router, built from MDF, used M10 threaded rod as leadscrews. I used M752 drivers running off 68V. Never had a missed step, as far as I could tell. Had a lot of other structural problems, mind you... That was a bit OTT but I intended to re-use the electronics in a later version.

I'm still slightly unsure of the "missed step" assumption. Can you measure how much error there is after, say, 50 vigorous Z moves? 100 moves? Are the results consistent, or random? Change the speed and acceleration and try again? Before junking the drivers and replacing them with different cheap low-power drivers, it might be useful to get a better handle on the underlying problem. I might even swap a couple of drivers between axes and see if the problem follows the driver. Reroute the wires and see if anything changes. A bit of science is called for, carefully measuring and only changing one thing at a time, despite your obvious frustration! I found my Z problem when I was able to see that the error in microsteps after each height-setting operation was almost always equal to the number of direction changes during height setting, for example, and the error was independent of how far the axis travelled.