My printer is using cheap linear rails and I am pretty sure the slight play about the axis of motion (roll) in my x-axis is causing the artifacts shown below. Has anyone had any luck with removing this unwanted play from their linear rails?
The rails are : https://www.robotdigg.com/product/1222/GCr15-MGN9-Linear-Rail-and-SUS-MGN9-Linear-Block
I’ve heard of people replacing the bearing balls with higher quality ones.
Does look like vibration. What’s you Jerk settings?
put in a spring or just a rubberband that keeps stress on it - so your rail is prestressed. As you have only acceleration forces this works to minimize vibrations - also slowing down helps a lot as the gravitation is bigger and taking over the job of prestressing the system.
Turn 2 of your bearings 30-45 deg and try a new test print. Worked absolute fine for me!
Like this
missing/deleted image from Google+
Thanks for all the suggestions! I have managed to find a way to pre-load the carriage by pushing on the sides of the carriage with three M3 set screws. This compresses the carriage just enough to remove the play and hopefully not increase wear significantly.
missing/deleted image from Google+
@Ulrich_Baer I was trying to think of a way to implement your suggestion when I cam up with my set screw solution, so thanks for the inspiration!
@Glen_Beer I’m using Replicape and Redeem firmware with jerk in x & y of 0.05, not really sure if that is high or low. Now that I have the vibration basically zeroed out I’ll experiment with the motion planner settings.
I printed the same part again after my modifications and now I get this. It’s a bit hard to tell from the picture but it is significantly better. Now I just need to tune the print settings I think.
The printer is also practically silent now which is a huge bonus! It used to really rattle whenever it hit a resonance. I’m actually quite surprised how much of a rattle considering it only had a small fraction of a mm of play.
missing/deleted image from Google+
seems those screws are for adjusting the clearance - so this would be the proper action but it is not preloading as there is no constant load (sorry for nitpicking)
@Ulrich_Baer depends on where you define the limits of the assembly I suppose. In terms of just the carriage and rail you are correct, but for the whole assembly as shown then you could argue that there is a constant load on the ball bearings. Don’t worry, I like to nitpick too 
sure it is a question of stiffness. If something is bend it would work as a spring. Preload means that there is a possible movement, while a correct clearance ideally means no free movement.
@Ulrich_Baer I think the distinction here is that preload maintains no free movement when under load. A ‘correct tolerance’ only means no free movement when not under load, or with infinite stiffness. So in any real world application some degree of preload is necessary when precision movement is required, as any structure is compliant to some degree.
I should point out that in this case the set screws are tightened sufficiently to elastically deform the bearing carriage and compress the ball bearings beteeen it and the rail. My apologies if this was not clear previously.
yes, it is mainly a cost factor as for a real clearance without preload you need very precice and sturdy parts else they would jam on certain position. To overcome these expensive small tollerances parts made less stiff so, as you said, can be elastically deformed. Just wonder how such printer would perform in Zero-G probably worse as there is no gravitational preload. On the other Hand i am just thinking if a strong Nd magnet is placed next to the rail it would pull the carriage against it with no friction (but eddy currents) this could offer a cheap solution to minimize vibrations.
welcome to my world
@Ulrich_Baer funnily enough I already have a magnet doing just that, although that is not its intended purpose. It is a 6x6mm Nd 52 magnet used for triggering a Hall sensor, you can see it in the drawing I posted earlier as the big circle between the back set screws. Considering this magnet did effectively nothing I would need a very strong magnetic field. The weight needed for this magnetic approach thus makes the set screw option considerably more desirable.
a 5×5×3 N52 magnet could add 1kg, but only in direct contact. A 40×10×5 N42 is at 13kg (while its mass is only 15g) with a 2mm distance you still have ~2kg which should be enough to minimize vibrations.
That is correct, but… weight.
weight? 15gram? (a NEMA17 is 200gram)
