I have an Algolaser DIY 20W and everything works normally. Recently, I bought a Delta 22W, and when performing the material test, I have this misalignment problem.
I have already tried tightening the belts and checking for vibrations, but nothing, absolutely nothing works.
Does anyone have any ideas on how to solve this?
Check the set screws on the pulley to stepper motor conection.
It looks like backlash. Probably a belt issue.
You can test this pretty quickly, by turning bi-directional scan off. With it off, there shouldn’t be any visible backlash.
Belt drives don’t have any backlash. I’d say go with Tim, something loose there.
Anything mechanical item that moves against something else has some form of backlash.
What Russ Sadler calls curtains is a type of backlash.
He can easily check it by turning off bi-directional. It’s it or it isn’t
A belt doesn’t move against anything else. It’s directly connected to a pulley. When the pulley moves, the belt moves.
Backlash is from threaded drive systems. If you put a nut on a bolt, there’s play between the nut threads and the bolt threads. If you tighten the bolt, the nut is pressed against one side of the bolt. reverse direction and the bolt will spin a little bit before it makes contact with the other side of the nut. That’s backlash. The amount of play before the nut actual moves. Belt drives have no backlash. Any “slop” would be the result of something loose.
I used to think so, too, until we got a report of NEJE shipping mismatched belts and pulleys. As a result, a perfectly adjusted machine had dramatic skips & backlash, because the belt teeth did not correctly fit the pulley teeth.
While I doubt that’s the case here, it’s prudent to consider all machines as a collection of random parts requiring verification of the obvious assumptions we all make… 
Yeah, if you’re using the wrong belt for a given pulley then all bets are off.
But when using the correct parts, belts have no backlash.
It’s more likely a grub screw as others have stated…
If it’s a toothed belt, you can have the belt stretch as it’s coming into contact with the toothed pulley.
If you don’t want to call it backlash, I’m ok with that. In the end, it’s similar, at least.
You can watch Russ Sadler video and see what I’m talking about.
But see that’s my whole point.
I can’t tell you how many posts I see that are blamed on backlash but have nothing to do with it.
Just recently there was a post where the print was moving inches, literally inches, away and the first reply was “That’s called backlash.”
Losing steps is not backlash. Bad USB comms is not backlash. Even ednisley’s example of a wrong belt is not backlash. While it may (will) result in backlash, the problem is using the wrong parts.
For some reason people here go straight to backlash, whether it’s an assumption or not, I don’t know, but it’s not really helpful.
Most CNC controllers have a “backlash compensation” setting that will “remove” backlash by moving the set amount more on reverse moves.
This setting will not fix what’s considered backlash in some of these posts.
Got it … I’ll refrain from suggesting this with belts.
Although I didn’t get a degree, I was advised in college that anything that moves against another object will have some kind of backlash…
I’ll concede as I see your point.
