I guess what you are saying is that it now gets really complicated:drowning:
Getting rid of the remaining 2/1000 parallel error would be nice but that is approaching the limits of my measuring capability/repeatibility.
I also took out my old engineering text books and estimated the thermal (bi-metal) bending of a 20mm steel rod firmly attached to a 4.5" wide aluminum bar (accounting for the air channels in the 80/20 profile). Over a 10 degree C temperature span (which can easily happen in my shop) there might be a whopping 6/1000" bending in the middle under worst case conditions. I suppose practically it is much less because the steel rod may be sliding on the supporting extrusion but I will check the remaining error at different temperatures.