First few layers linearly shift

[adarcher]

So, it's late and I'm kinda tired of working this out, so I thought I'd ask here for some help.

I've attached some pictures, but they don't show it very well since I couldn't get a good shot of it.

Basically, after a few layers the print evens out and stops shifting slightly to the X.

The easiest to see is in the bridge test: thingiverse link

one cube from retraction test

The towers angle in on all sides equally, yet it looks like it is straight up on one side.

I lowered the bed temp to 35 and did a retraction cube test (couldn't find a single cube on my laptop for some reason).

bridge test

This cube shows a wave in going back and forth actually.

(preview is swapping the images in preview, hopefully they are correct for real)

I've used many slicers and firmware versions, but it's been like this for a month now.

Rostock Max V2 w/ E3d V6
Hot end: 200-210
Bed: 35-65
Speed: 50-100mm/s

All give about the same results. So, I think it's missing steps or something. I've taken apart the machine a few times to try to check on the geared pullies on the motors and such, but it's not changed anything.

Heh, as I write this, I printed out a bridge test at 150% size and it's not nearly as bad looking. But it's really just the first 3 layers that shift in every print.

150% scaled bridge

Anyway, if you've every had to deal with this, I'd love some tips.

I'm currently printing out these to try to make my belt tension more consistent.

[ExplodedZombie]

I am trying to print a simple calibration cube and am experiencing the same thing. After the first 4 layers the entire print shifts to the right, prints great, then the top 4 layers shift BACK to the original spot. Is it...leaning or something? Like maybe my bed isn't level to the floor and everything is sliding? Unlikely. If I figure it out I'll come back and let you know.

[adarcher]

I am trying to print a simple calibration cube and am experiencing the same thing. After the first 4 layers the entire print shifts to the right, prints great, then the top 4 layers shift BACK to the original spot. Is it...leaning or something? Like maybe my bed isn't level to the floor and everything is sliding? Unlikely. If I figure it out I'll come back and let you know.

Thanks!

Hope you find your problem. These shifty problems suck.

[adarcher]

I think I figured it out. I was curious about belt tensions and tried looking up tension guidelines but couldn't find definitive rules to follow. One person said to make all the belts play the same tone, while another said to pull the belt with 6 pounds of force and get it to deflect between 1/8th and 1/2 inch.
So, I made them give a note, and used my thumb to measure tightness.

I guess I was too tight. After I printed the tensioners I mentioned I looked up those posts again and decided I would try it loose, pretty darn loose compared to before. And now no early shift in my bridge test. I'll test with some other models next to see if only that model is changed.

[adarcher]

Nevermind. Still shifts the first few layers in the -X direction but only on the -X side of the model.

[adarcher]

In case someone looks at this: Can the belts stretch? My machine fell over about a year ago and at one point where I had to take the RAMBo out, I reversed a motor and when I homed, it ran into the ground (instead of the top) and I'm wondering if that caused some permanent damage I can't see. I have since taken the belts off a few times to check on them, and they "look fine" no missing teeth or rips (though I have seen a dash of salt sized sprinkle of black under the motors at one point).

[DeltaCon]

I think I figured it out. I was curious about belt tensions and tried looking up tension guidelines but couldn't find definitive rules to follow. One person said to make all the belts play the same tone, while another said to pull the belt with 6 pounds of force and get it to deflect between 1/8th and 1/2 inch.
So, I made them give a note, and used my thumb to measure tightness.

I guess I was too tight. After I printed the tensioners I mentioned I looked up those posts again and decided I would try it loose, pretty darn loose compared to before. And now no early shift in my bridge test. I'll test with some other models next to see if only that model is changed.

I found those tensioners too a while ago, and want to implement the, seems like a good idea. A week or two ago I changed to the new skates so I had to loosen my belts. Although not in the manual, I loosened the upper pulley's too, until they fall completely flat in the mount. In that position I simply pulled the ends of the belt with reasonable comparable strength, and fixed them with the clips on the new skates. I forgot about tensioning afterwards, but it works wonderfully...

So I think the tension is not really very much of a concern, as long as they are not overly tight (be careful with screwable tensioners!) and don't skip the toothed pulley on the motor.

[DeltaCon]

In case someone looks at this: Can the belts stretch? My machine fell over about a year ago and at one point where I had to take the RAMBo out, I reversed a motor and when I homed, it ran into the ground (instead of the top) and I'm wondering if that caused some permanent damage I can't see. I have since taken the belts off a few times to check on them, and they "look fine" no missing teeth or rips (though I have seen a dash of salt sized sprinkle of black under the motors at one point).

I am sure belts can stretch indeed, especially when they are over-tightened like you mentioned earlier. But that would take considerable time to become permanent, and they are especially designed to prevent that. If you want to know, you could mark them X, Y, and Z and turn them around one tower. If the shifting follows the belts you know enough.

Here is an other approach: Do you have the bed heat up, and then give it considerable time for the heat to creep to the surface and level out? Maybe you have a bed that creeps on the mounts a bit when the heat dissipates to the upper surface. When it is done creeping, the sifting stops. (e.g. after a few layers).

[adarcher]

In case someone looks at this: Can the belts stretch? My machine fell over about a year ago and at one point where I had to take the RAMBo out, I reversed a motor and when I homed, it ran into the ground (instead of the top) and I'm wondering if that caused some permanent damage I can't see. I have since taken the belts off a few times to check on them, and they "look fine" no missing teeth or rips (though I have seen a dash of salt sized sprinkle of black under the motors at one point).

I am sure belts can stretch indeed, especially when they are over-tightened like you mentioned earlier. But that would take considerable time to become permanent, and they are especially designed to prevent that. If you want to know, you could mark them X, Y, and Z and turn them around one tower. If the shifting follows the belts you know enough.

Here is an other approach: Do you have the bed heat up, and then give it considerable time for the heat to creep to the surface and level out? Maybe you have a bed that creeps on the mounts a bit when the heat dissipates to the upper surface. When it is done creeping, the sifting stops. (e.g. after a few layers).

After loosening the belts, I know they must have been way way too tight. There is no longer a grinding noise from the motors.

Those tensioners are nice actually. I ended up putting them on and making the belts slightly floppy before fully tensioning them. Seem to work really well, I'm not silent, but there is a lot less noise

Re: heat creep
I was thinking about that. While it was freezing weather a month ago and my machine was in the mudroom (so it could run over night and not keep me up) I had the bed 20 degrees hotter than usual. ended up smooshing the bottom of the parts a bit because of overheating.

In most of these tests, I've brought the bed and hot end up to temp (65 bed and 200 he) for a bit to make sure things are a-ok before pressing "go" on the print.

I don't have the picture with me, but I ended up printing 5x20mm squares at [70,70] [70,-70] [-70,-70] [-70, 70] to see the shift across the bed. Looks like they all angle out a bit as if the first few layers are in a bowl.

I was super happy that I was calibrated enough to not have my nozzle scrap the bed at any point (though one corner was a little lower than the rest by about .04mm).

That calibration got me thinking as well. I'm using the octoprint plugin by geneb to get my offsets/angles/rods values. Tonight, I'm going to modify the script to not modify the rod length, since it makes it 139.xx versus 141.xx from the M68 command.

[adarcher]

I thought about making another thread about the calibration plugin, since that what I'm using these days to get a flat first layer, but I figured it may all be related.

It looks like my probing values are a little weird to me, this is my output:
"""
Progress: New calibration is not measureably better than the old - keeping the old calibration
"""
Variables:
X offset: 92
Y offset: 0
Z offset: 37
A angle: -0.79
B angle: -0.44
C angle: 0.00
Rod: 291.06 *always stays the same because of 6 point calib
Radius: 139.59
Z Height: 376.96

Point Deltas:
(outer 6)
0: -0.06
1: 0.21
2: -0.16
3: 0.18
4: -0.18
5: 0.19
(inner 3)
6: -0.12
7: 0.06
8: 0.10
(center)
9: 0.06

I feel like the zigzag of the deltas around the edge should mean something...

[adarcher]

Sadly, I think I may just give up. It's gotta be something hardware related and I just don't know what it is.

[nebbian]

Point Deltas:
(outer 6)
0: -0.06
1: 0.21
2: -0.16
3: 0.18
4: -0.18
5: 0.19
(inner 3)
6: -0.12
7: 0.06
8: 0.10
(center)
9: 0.06

I feel like the zigzag of the deltas around the edge should mean something...

It does. Try bumping up your steps/mm value by 1% and see what happens.

[adarcher]

It does. Try bumping up your steps/mm value by 1% and see what happens.

Do you know the reasoning behind that? I've got some mech-engineering in me, so I'm curious. I'll try to find the time tonight or tomorrow night to try this.

[adarcher]

It does. Try bumping up your steps/mm value by 1% and see what happens.

Lol, that got my hopes up. Forgot what that felt like

It got rid of the zipzag of the probe points, but didn't fix my overall problem. Ah well.

My motors look fine. My belts look fine. The carriages have a little dust in them, but if it was an uneven wheel, wouldn't I see the problem elsewhere as well? Plus, they feel steady to me.

Can my wires be cross contaminating somehow? Should I be isolating the cables? The original instructions kinda had them bunched up in a rats next under the there...

[adarcher]

I haven't fully given up yet.

I'm either slipping on the carriage or skipping steps.

I took everything apart and looked at the bearings on the Y tower. They all didn't spin as nice as some random left overs I had laying around, so I replaced them. No change.

I did notice that my belts are favoring one side of the bearings and I have a feeling that it's rubbing the edges... which would explain the noise. Any way to get the belts to stay in the middle of the bearing? I can slide them away by hand, but they quickly jog back over.

[adarcher]

OK. I'm going to call this mostly solved. Now only the first layer is only ever-so-slightly off.

I was sitting here watching it fail yet again while sitting next to an old ramps board. The steppers have heat sinks on them. My rambo doesn't. It never did. For giggles I stopped the print, added some spare sinks with some thermal paste.

OK, so here's the question I have now. Why now? Is my rambo actually stressed almost to hell and these heat sinks are just a bandaid?

[DeltaCon]

Stepperdrivers can get very hot when they get overcurrent. A sink can help preventing overheat. But in reality, if the steppers on your rambo cannot be touched without burning your finger, you have too much current on the motors. That does not explain however, why only the lower layers would shift. On the contrary: heat has to build up before it gets a problem, so only the lower layers would not be hurt by this effect. So my guess is that you have two unrelated problems.

[adarcher]

Stepperdrivers can get very hot when they get overcurrent. A sink can help preventing overheat. But in reality, if the steppers on your rambo cannot be touched without burning your finger, you have too much current on the motors. That does not explain however, why only the lower layers would shift. On the contrary: heat has to build up before it gets a problem, so only the lower layers would not be hurt by this effect. So my guess is that you have two unrelated problems.

Thanks for the reply!

I was thinking that too, but it's hard to not notice the drastic change: I smoothed out the shift by updating the belt tensions and whatnot so it went from a clear 10 degree angle on a 5x20x20mm box to just the first/second layer being off by .2mm. I should have posted pictures. I was thinking that the overheating was making it easier for the steppers to skip with the added stress of the first few layers possibly dragging on those initial boogers that sometimes appear. Also, by the time I've heated up my bed and hot end, the steppers are hot--no actual movement going on yet. Part of me wonders if there is a rogue current going to them somehow/somewhere that only is relieved by actual stepping

I only had about an hour to play with it last night, so I didn't get to fully test it past 4 prints to mark it as a consistent improvement.

I'm going to look into some kind of bearing centering fix because the rubbing (now that I see and hear it more) is really driving me crazy. Even if this isn't a fix for my problem, it'll help with my sanity... hopefully.

I've done so much to fix this issue that just fixed other things, it's comical to me now. I can have an amazing first layer everywhere on the bed. I've reduced my angle and offset calibration values to close to zero for all towers. I've tightened every major structural screw/nutt/bolt. Lots more to do though apparently.