Slipping stepper gear?

[DeltaCon]

Seems like I have been exchanging one problem for another for about a year now... Getting time to get some descent prints now! What could be the cause of the behavior in this picture? The nozzle does some strange lifting even if still in the first layer. I have already checked the belts and tightened them from scratch when I was getting the towers aligned yesterday.

The bottom left leg of the ring is pointing to the X tower.
I was wondering what's the best way to check the gears slipping and how to tighten them without dismantling the whole thing (again).

[Mac The Knife]

You should be able to check the setscrews for the timing belt pulleys on the motor without having to disassemble much.

[Turpinator]

Is all that infill shifted itself over/away from the X tower?

It looks like theres some adhesion issues too?

[Polygonhell]

Could be number of things, the set screws on the drive gears are certainly worth checking.
But it could just be too much speed/acceleration or jerk in the firmware, as you print at the edge of the build area, because of the delta kinematics the carriages have to move at almost 2x the commanded speed, that is makes axis much more likely to skip steps.
After you've checked the screws, if that isn't the issue I'd just try reducing the travel speeds and halving the acceleration and Jerk to see if that addresses it. Once you get a decent print, you can try edging them back up, I'd start with getting the Jerk back to default settings and go from there.

[DeltaCon]

You should be able to check the setscrews for the timing belt pulleys on the motor without having to disassemble much.

I was hoping so, but I think they cannot be reached without lifting the top baseplate, can they?

Is all that infill shifted itself over/away from the X tower?

Yes you saw that right. It is not because of adhesion problems though, because I can see the the nozzle multiple layer thicknesses above the bed almost spraying the filament to the bed and it still sticks.

But it could just be too much speed/acceleration or jerk in the firmware, as you print at the edge of the build area, because of the delta kinematics the carriages have to move at almost 2x the commanded speed, that is makes axis much more likely to skip steps.

Yes I have thought of that too. But the printspeeds here are really low already like 30mm/s or so, and this is the small version of the calring so it not going near the edges, but halfway at most. But it is worth checking the jerk and acceleration values for sure. Happen to know the defaults for a V2?

This is happening after an attempt to get rid of tower lean, I also redid the belt tension, but I cannot say that it changed much compared to before. I might have touched the skatewheels tension to, because I noticed a few that were not rolling during movement. The skates go up and down as easy as before, except fot the Z tower that seems to drop a bit on its own weight when steppers disconnect. In the first attempt to print this the effect was a lot bigger, where the hotend swung into the air when going to the Z tower, and coming down again when travelling away from it.

[Turpinator]

I was hoping so, but I think they cannot be reached without lifting the top baseplate, can they?

Pull the curved plastic pieces hiding the corner/base of the towers? You should be able to see the stepper right there, and then the output shaft/gear between the two melamine pieces holding the actual tower in place.


As a note on speeds, The slowest I ever print is 30mm/s for perimeters, and ill run all the way up to 75mm/s reliably, with infill speeds eclipsing 120 mm/s. Is the print quality sometimes crappy? Yes, but never any major issues besides surface finish, notably waving due to sharp corners, much like that of an underdamped system response.

[DeltaCon]

Pull the curved plastic pieces hiding the corner/base of the towers? You should be able to see the stepper right there, and then the output shaft/gear between the two melamine pieces holding the actual tower in place.

You mean to stick an allen wrench between the bottom of the extrusion and above the idler that fits the setscrew of the pully? Great... I'll have to get me a longer wrench. Hope I can find a long enough one locally. It's a 2mm one right?

[DeltaCon]

Nope, it's a 1,5mm, just went to the store a second time... Should have tried it on the extruder gear first

And for sure there is no loose gear on any of the towers.

[DeltaCon]

The issue already made printing impossible, but it seems to have gotten a lot worse overnight...
As in my previous post I checked the pulleys, and all setscrews were as tight as I dared turn them. There is certainly none of theme loose, so I think I can rule out slipping gears.

https://youtu.be/8guDPog42Gk

This shows exactly what is happening after I start a print. The next video shows proof of how my printer is behaving when manually moving around.

https://youtu.be/Nk0yIC4NGrQ

It shows that Z=0 gets set. I filmed that horizontal movements do go okay when done by hand sending gcode. Also I show prove that at z=0.3 (first layer height) the paper drags nicely below the hotend. I tried to film the teeths of the belt at 10 to 15cm below the hotend, because that's the position where they are at the gears when the hotend gets near z=0. As far as I can see (the movie is hard to get focussed) they look okay and not flattend by grinding or anything.

So... What did I forget to check? Why could the hotend dive into the bed? This is making me mad!

[Xenocrates]

It seemed like it was attempting to do calibration at the beginning of the print. Would you mind posting a sample of the G code you're generating, say of the 20MM cube?

[Turpinator]

Its printing into the air on the first one because its ram jamming the head into the plate, causing the motors to skip steps, ie not move when the board thought they moved.

Figure out why its jamming itself into the plate (gcode/software issue, most likely, 99.99%).

[DeltaCon]

Its printing into the air on the first one because its ram jamming the head into the plate, causing the motors to skip steps, ie not move when the board thought they moved. Figure out why its jamming itself into the plate (gcode/software issue, most likely, 99.99%).

Yes I noticed that... I used the same gcode that was generated by repetier as in the picture I started the thread off with. The behavior is a whole lot different, but not good in both cases. I will slice a new file tonight. (CET+1)

It seemed like it was attempting to do calibration at the beginning of the print. Would you mind posting a sample of the G code you're generating, say of the 20MM cube?

Hmm, I see what you mean, the HE touches down on x0 y0 indeed, did not notice that before. Maybe there the first steps are lost and that causes the the head ramming and losing more steps. I have no idea why it touches the bed in that position, because the CalRing has an open center. No filament should be laid down there at all. I am definitely NOT doing any probing in my startscript. I will check the gcode and post it tonight (CET+1)

Thanks for trying to help me out guys!

[Turpinator]

Never hurts to do a manual z height calibration through the interface.

settings, all the way to the bottom, advanced, all the way to the bottom again, calibrate or z height or something,
home it
go to the uhhhh... z height adjustment (not the ZERO)
spin the encoder to move the head up/down, get it to zero (actual, not displayed).
Check the z height displayed. Is it close to zero, or like +10mm? If its above 0, that's why its jamming itself into the bed.
go back one screen and press the zero button (or set z min or something).

Congrats on your first manual calibration! And that was the easy part. Doing the end stops and horizontal radius was always tons of fun.

[DeltaCon]

I just did a manual z-height set using the paperdrag method. I can see that the nozzle distance to the bed at first layer is visibly larger. I know that Z=0 grabbed the paper snug before, but z=0.3 did not. So I am sure that Z=0 was not too close to the bed for the motors to start skipping steps. I started a 20mm calibration cube And until layer 25 or so it seemed to go well. Then while blinking my eyes things started to shift again. This time without any bed hammering or anything hearable going wrong, things start to shift in an opposite to Y-tower direction. (Remember that in the first picture in the OP things shifted in an opposite to X fashion?)

https://youtu.be/qbDARNsdy_s

The Gcode of the cube is attached:

20mmCalCube.gco.txt

(219.15 KiB) Downloaded 312 times

Things I already did:
- retensioning the belts (look at video in previous post to get an idea whether they are too loose)
- checked the setscrews on the stepper gears, all tight
- Increased the stepper current 10%
- Checked for warm steppers, they all feel cool to the touch while printing
- loosend the bearings on the skates, might have gotten them a bit snug before, but now I can turn them all by hand with very little resistance
- checked the towers for positions where the skate feel a "click" (dimpels in wheels or extrusions), is smooth now, but had some before loosening the wheels
- checked for grinded belts, there was a slight amount of rubber on all towers, but who know how old (second hand machine). I wiped it, and I see no new grindings developing.
- decreased Jerk and Acceleration to 25 and "on print" to 1000 and "on travel" to 2000 (default 35, 1850 and 3000)

I also resliced the cal ring, and when printing it does not show the short head to bed move at x0 y0 anymore. But the shifting problem is still there. In this case there is an obvious lifting too, so not just horizontal shifting.

https://youtu.be/Ra1XzrKdtOU

If I put these pictues next to eachother, it gets clear where the problem starts (compare print path in repetier to the actual layer being laid down)

The gcode for the ring is this:

CalRingSmall.gco.txt

(914.2 KiB) Downloaded 322 times

All to no avail...
Any ideas left?

[DeltaCon]

Additional to the previous post I can mention that I retensioned the belts a second time, a bit tighter.
I started the Calibration ring a second time, and filmed it. At 1.40 minutes you can see just as the loops are finished it starts the perimeter (very slow because I told him so), it immediately starts doing the air-dance again, without any hammering or hearing the motors skip or anything.

https://youtu.be/7tWISUYGF1g

I am curious as to why the symptoms are so different each time. The problem does not start at the same spots every time, not geometrical and not in time or in printing positions. I don't think it is the gcode (see previous post).

I hope someone can push me in the right direction!

[DeltaCon]

Not many ideas huh?

I contacted SeeMe Support today and JJ picked up my request pretty quickly (thanks!). However while answering his questions and repeating steps in practice something struck me... In my first video above you can see me pulling off some rogue PLA from the bed. Just after that the HE started doing the skydance. I thought that it was a coincidence. Today the exact same thing happened and I got this dejavue of shooting the first video. I tried some 10 times to replicate. Not always setting off the FSR leads to the skydance, but always the skydance is preceeded by the FSR going off...

I am uncertain as to the exact cause. I am guessing that in the first video (where I was NOT triggering the FSR myself!) there was some buildup (maybe over-extrusion) that was touched by the HE and set off the FSR as well. That's why it often appears at different stages in a print I guess.

So I checked the #define ALWAYS_CHECK_ENDSTOPS and changed 1 to 0. The behavior is gone with this setting. I understand that this setting would kill printing or pull-up the HE quickly or something. I don't understand how that would lead to what we saw however. But I am just glad that I found the culprit and can continue my experimentations. The time I lost troubleshooting this...

Thanks for your efforts to help me. I am sure I will need to pick you up on that again sometime