|Z axis rev 2, with worm gear reduced motor drive.|
The gear reduction mechanism I chose was an OnDrives Rino 30:1 worm gear drive that came with its own NEMA-23 size motor (used, $100 via ebay). It completely solved the bed-drop problem, but at some point I ran the plate too far down and the motor kept pulling at the belt which eventually stretched enough to cause it to slip over the teeth on the drive pulley causing the bed supports to shift so the bed would no longer be level.
|Worm gear reducer mounted at bottom of Z axis frame.|
That got me looking at the belt path and thinking about how much the belt will stretch under load. The relatively stretchy, glass core GT2 belt was replaced with with a less stretchy, steel-core HTD-3M belt. I ran some tests, applying loads in the form of filament spools to the bed plate and measured the amount of belt stretch that occurred.
A Zmax limit switch was added to the design to prevent damage to the mechanism in the event that the Z axis is driven beyond the limit.
|Unloaded bed at about the middle of the Z axis.|
|Bed loaded with 4 kg (3 kg filament + 1 kg spools) at about the middle of the Z axis.|
|Belt stretch under 4 kg load measured on the left side of the frame (printer was turned around and gauge mounted on the back side).|
I mounted the gauge on the other side of the frame and checked the stretch and found it to be a little less, as expected because the belt path back to the motor was shorter. This implies that under load, the bed will tilt slightly due to differing belt paths on the right and left side of the Z axis frame. The truth is, the amount of tilt probably wouldn't matter, but the asymmetry bothered me.
I decided I didn't like the serpentine belt path, which led to rev 3 of the Z axis mechanism...