Spot on Alex! I keep meaning to do the maths on the deflections. Does anybody know the mass of the print head assembly? Let's not forget that the y-rods will also deflect, as can the glass due to any load. The advantage of circular rods is that they can be ground quite precisely, whether they are or not is another argument. As such, linear bearings can be matched very well. I'm not convinced that extrusions will offer a more accurate slide. You can dial out gaps in tolerances by preloading bearings but that can lead to problems with wear and maintenance keeping things tight. Using software to compensate for any variations seems to be a good solution to me. Consider all of the places in the build where minor inaccuracies could be. With the best will in the world, we can't build a printer ourselves and achieve perfect alignment, we can get close and then use levelling the get better. Good printing! Jason
My measurements suggested the Y deflection was about twice the X - not surprising given they carry the X load as well. Y 0.10 X 0.04 .. of course that also includes the additional load for the gauge.
OMG NOOOO those result are USELESS then!!! Oh well that's about what I expected... But not too worse... Z doesn't have to move much and not fast to compensate that if you have a decent level algorithm like mesh leveling in Marlin or upcoming UBL with even more points for a finer mesh... Lets see what's @dc42 is coming up with in RRF 1.17...
UBL sounds interesting indeed Alex. I should really go through and do some fine adjusting of my glass plate again to get the MBL values lower but it's producing decent and consistent first layers at the minute so I think I'm going to leave well alone.
Still that is 0.14 at the nozzle... I think Dave's comment that distributed load across multiple short bearings is probably the sanest route to solve this mechanically
I'm not convinced tbh. If the bearings are 60mm or so long, you'll already have the load spread across. To get a great benefit you'd need a much bigger span for two bearings and that would reduce your print area.
This would not have applied to the original X carriage that was wider but there is still room to extend the width. As I said, the wider distributed load spacing would still be effective.
