Iluminatri, sorry for the tardy reply. I thought I had email notifications enabled for new replies. From what I can tell, Repetier sends the steps/e value to the controller's memory. I was sure to change the value in Cura and verify that Cura's g-code output had the correct value. I am using a packing density of 1.
I am printing PLA at 210-220. I was printing reasonably large objects at 100% fill so this will make an over extrusion problem readily apparent because the extra material has nowhere to go. When I said the filament gets sheared, I didn't mean that it snaps. I guess you could say it is the failure mode beyond grinding and slipping. The bolt will leave crescent shaped depressions in the filament where it was "grinding". It is different failure mechanism than grinding with the stock gear because although it has a lot more bite, once it tears the crescent in the filament, there is little to no chance of it catching again. This is much more preferable to me because it is more tolerant of little errors and let's you know when something is VERY wrong.
The pressure plate onto the bolt is pretty tight. Eventually I will move to a roller bearing pressure device but I haven't had success yet. I have not had filament get stuck in the bowden tube from deformation due to the hobbed bolt. When the tube pops out, the head has already been scraping the print for a while, indicating to me that I have compressed filament in the tube.
The bed is properly leveled and homed. I have dialed back the lower layer speed a lot (back down to the stock ~20mm/s). It is possible that with a "too fast" lower layer, I was having to bury the first layer into the bed to get it to stick. This still doesn't explain why the problem gets worse as the print goes on.
I have not verified X Y or Z calibration. I assumed because my prints are coming out square and straight then no steps are being missed but assuming is a dangerous thing. I will perform a repeatability test on the Z-axis.
After much fussing, I seem to be getting good prints at 885 steps/E. I am not fudging the filament diameter. I am torn between "if it ain't broke, don't fix it", and wanting to know what the root cause is. I will do some testing and report back.