What's my problem?

[oliof]

two remarks:

• you can print the ringing tower and measure the ripple frequency to tune input shaping. Works both for klipper and RepRapFirmware.

• a friend of mine figured some vibrations on their railcore at certain speeds and has to print slower or faster for those to go away. so they either print at 25mm/sec max or at 120mm/sec min. Might be worth investigating.

[oliof]

as to picking x or y: the vibrations for X and Y should be similar on a corexy setup -- differences point toward uneven tension in the belt system. If the difference isn't too big you will be able to pick either peak frequency and at worst will have to choose an Input Shaping algorithm with a damping factor.

[zapta]

@gnydick, it's possible that the V0 just has a more stiff and accurate movement system, because of its smaller size.

[Phaedrux]

Don't take this as me telling you to get help elsewhere, as that's not the case, but there is a Railcore discord server. Perhaps asking at the source might be useful. Maybe it's a common problem or someone has seen it already.

https://discord.gg/Sy569Hq

[gnydick]

@oliof Thanks for the info. Yeah, I've noticed that too, but never took the time to find the thresholds.

[gnydick]

@oliof Thanks for the help. Are there any simple ways to determine which algorithm and settings?

[gnydick]

@phaedrux Not at all, thank you. I've been there. Totally forgot about it

[gnydick]

I did some investigation and so far have found that I have some temperamental linear rail blocks. I built an entirely new kinematics that is wider to sit on top of the existing frame and hold the e3d tool changer setup.

I had to buy new longer rails and it turns out that if the screws used to mount into the block are even a hair too long, the blocks actually bind on the rail.

The screws don't even impinge directly on the rail, so it must be deforming something inside the block.

[oliof]

@gnydick I printed the ringing tower again cycling through the algorithms every 5mm and decided based on that. For the bed slinger I did this on, mvz came out front.

[rjenkinsgb]

Again: Check the belts.
"Tune" them, so the equal length long runs give the same note when you pluck them.

The long runs on my CoreXY machine (520mm) ring at near the same pitch as the next to lowest string on a bass guitar, around 55Hz.

If they are significantly lower than that for a similar length, or very unequal, it will cause print accuracy problems.

I don't use input shaping, just very low pressure advance.

[gnydick]

Thanks for the input, all. My belt path is quite long and it really needed tensioning. It wasn't tight enough to even resonate. That made a marked improvement.

I've done a bit more fiddling and things are a lot better now. The print head is heavy and the cross beam is just the mgn rail which twists quite easily, so I think I'm at the envelope of what I can expect without input shaping.