More Core-T refinements.
TL;DW: swapped the positions of the motors and the idler pulleys to improve belt angle from 90 to 120 degrees, and applied some judicious use of belt twisting so that teeth never push on bearing surfaces.
Result - jerk of 30 and accel of 500 for x and y with no missed steps that I can hear.
Thanks to the members of this community for pointing out these issues!
Progress continues…
If you reprint the main T base (the part where the X arm attaches to the Y rod) you should space the bearings a lot farther apart. You’ll lose some Y travel but it’ll mean way less ringing on Y moves.
I had considered that, but I want to collect some data on this “worst case” scenario first to see how much impact a wider spacing .
Any suggestion on test shapes that would expose Y axis only ringing artifacts?
A box with embossed letters on one side is pretty good.
@Ryan_Carlyle excellent idea!
/me fires up OpenSCAD…
@Ryan_Carlyle farther apart? In y, x or both directions? How would that decrease ringing?
@Thomas_Balu_Walter I mean put the Y bearings farther apart in Y.
There are going to be three basic flex modes contributing to ringing in this gantry:
Making the Y carriage wider in Y will improve performance of #2 and #3.
This is standard stuff for cantilevers… there are various rules of thumb, but for plastic joint parts and relatively linear flimsy rods, I wouldn’t make the cantilever arm length more than 5x the width of the Y carriage bearings. Upgrade to stiffer hardware and low-friction bearings and you can probably get away with 8-10x, but it depends on a lot of factors.
I’m not striving for super precision here - this is just something I’m throwing together out of spare parts for my 6 year old to print with.
I though I’d explore the Core-T gantry design while I’d did so - and I’ve been very surprised how little ringing artifacts I have at the moderate speeds I’m using.
Of course, having a Bowden extruded helps!
@Ryan_Carlyle ah I got your “bearings” wrong. For some reason I was thinking about the bearings for the belt, not the one for the arm itself. I’ve been thinking about doing this with v-slot and wheels on both sides of the arm because the single belt run looks easier to handle than on a core-xy.
