Layer shifting

[ApacheXMD]

I'm having a weird issue where partway through a print the layers will shift by about 1mm. I dont hear any motor making any weird noises so i dont think its really skipping steps. I also inspected the belts any they look fine.

Currently the shift appears to be in the direction of the x tower. But a few prints ago it seemed to be in the direction of Z tower.


I'm using some experimental magnetic arm joints so it may be that, but before I blame my arms I want to eliminate other potential causes.

My belts seem tight enough but maybe its backlash due to too loose belts?

I've tried bumping up motor current in configuartion.h but the problem still persists.

Anyone else seen this?

[cambo3d]

does your magnetic arm joint have a cup to seat the ball? or is it just a magnet sitting on top of the ball?

[ApacheXMD]

Just a small cup. Ive got some stronger magnets on order and will be printing some fatter cups.

I dont think that its the magnet shifting on the balls because the problem doesnt seem to be speed dependent. If the magnet were shifting, i would expect that it would manifest itself as only a momentary positional discrepency, as th magnet will center itself eventually. Im seeing a very apparent offset in the layers so I thought it was a motor skipping a step or hopping a tooth on the belt. Next step is to tighten up the belts some more...

[Flateric]

This sounds very much like the issue I had with my magnetic adventures. I have become sure that it is a result of the cup sizes being to small and causing an "overshoot" but only sort of. Hard to describe really. I think larger cups will help especially if you get stronger magnets since there is a notable strength to the edges of the magnetic field which show up in the print as a "shifting", "skipping", "hopping" or whatever it can be called. I got so frustrated with the whole thing I have taken a break from the entire problem to let it simmer on the backburner of my brain for awhile and see what I can come up with. Before going back into it fresher and not so frustrated.

I read everyone's progress with great interest on this topic. Berry used magic and love with a little voodoo and got his just perfect! (Damn him and his magic voodoo sugar plum Berry charms! Just joshing you Berry, nothing but love, you know it.)

[dsnettleton]

I've also had some layer shift issues which turned out to be a problem with the cheapskate bearings. Unplug the tower stepper motors from your RAMBo board, then try sliding them up and down on the rails. If they don't move smoothly, you've got a problem. In my case, the tension bearings were improperly adjusted so that they scraped the cross-bar fastening the rear bearing onto the fixture. Don't know if that applies to your setup at all, but I figured it was worth a mention.

[ApacheXMD]

Thanks for the suggestion. I'll double check my cheap skates tonight among other things. That brings up a good question: how tight should we adjust the eccentric cam? As it is right now, I can manually move the cheapskates with the arms attached, but its not super easily moved. Without the belt installed and with NO arms attached, the cheapskates drop easily under their own weight.

I replaced the experimental magnetic joints with the standard universal joints but the problem still persists. Belt tension is such that I can hear a low bass note when plucked. I am almost at a loss now..

I'll double check the cheapskates and recheck belt tension again, I'll run more systematic tests.

Ive also read on the reprap forum about EMI issues that might cause steppers to skip steps. But I am still leaning toward a mechanical issue.

An interesting thing to note is that all of my single wall hollow cubes have proven to be aligned. Those were run at a slow 10mm/s. Now wheni maintain the 10mm/s speed with a model with curves, the layers will shift. With the delta geometry it makes it hard to narrow down the problem, or whether or not it's even a problem with one or more towers or something else entirely.

Mayne its a bad stepper driver? I have tried a print with a 80mm fan blowing at the Rambo board with no difference.

I am open to any and all advice. It's getting a bit frustrating but Im trying to keep a level head and be systematic

[RedWood]

I'm have the same issue. Two hours into my build i started hearing the tip dragging over some internal perimeters. Not long after i found my layers had shifted one too two mm to the right. I suspect that the dragging had caused the stepper to miss a step or two. Is there a way to confirm this? Is it possible for the RAMBO hardware to monitor the amperage going to the steppers? Any spike in the amperage would be able to tell us if we're encountering any significant resistance.

[elqisqeyano]

I'm have the same issue. Two hours into my build i started hearing the tip dragging over some internal perimeters. Not long after i found my layers had shifted one too two mm to the right. I suspect that the dragging had caused the stepper to miss a step or two. Is there a way to confirm this? Is it possible for the RAMBO hardware to monitor the amperage going to the steppers? Any spike in the amperage would be able to tell us if we're encountering any significant resistance.

Tighten up them belts, check cheapskates are firm but not stiff. Solved it for me.

[ApacheXMD]

Ok, I've double checked my belt tension and they are consistently tight; I tried tightening them a little bit more but it started to bind, so I backed it off. The cheapskates move uninhibited and smoothly with no noticeable backlash.

On some prints I've noticed the tip dragging a tiny bit over bumps on the print, but think that's a result of a skipped step rather than the cause.

Curiously, I've been successful at printing hollow cylindrical test objects with no shifting at all.

hollow cylinder

This was printed at 10mm/s, single wall 0.2 layer height

I mentioned earlier that hollow cubes printed ok for me too:

20mm cube

I made a test object that's a 40mm square, with an array of holes to test dimensional accuracy on the printer, but it prints out skewed.

test square

test square2

You can see the holes between layers don't exactly line up

I've designed a hot end platform with divots for steel balls for use with my experimental magnetic joints. Printing that has yet to be successful.

Platform_killed

This print I killed part way through when I saw the layers shifting. The holes don't line up.

platform side

This is another attempt. Looking at it from the side you can see the layers don't line up.

I tried adjusting axis accelerations with little affect, I went all the way down to 400mm/s^2 with no improvement with layer alignment. I'm not sure what else to check...

It's weird how the only things I'm able to print decently well are hollow single walled objects. Does anyone see a correlation or a pattern? I think I might contact SeemeCNC for help also.. So far I haven't had a single useable print that wasn't a hollow calibration object, and am at my wits end.

[foshon]

Is the other side of that platform side print skewed inward? If so the orientation of the skewing will tell you which axis has the issue. Maybe try a three pointed star and orient it to match the tower pattern.

[Polygonhell]

If the simple prints don't skew the odds are you're losing steps on rapid movement, because there is none for a single wall print you don't see it there.
Turn down your travel speeds and see if it goes away.

[ApacheXMD]

The platform that I printed out is triangular in geometry, and as I stated in the first post, the skew looks like it's pulling toward the X tower. The X tower cheapskates and belts are as near as I can tell the same as the other two. And yeah, where it skews outward, the opposite side skews inward.

After checking all mechanicals, perhaps there's a problem with a stepper driver. With the Rambo board, it's all integrated and I have no way of trying another driver without another board.

If the simple prints don't skew the odds are you're losing steps on rapid movement, because there is none for a single wall print you don't see it there.
Turn down your travel speeds and see if it goes away.

I'll try that. I thought I had travel speeds turned down but I see now that I had it set at 100mm/s. If that's the cause, what would be the solution?

[Polygonhell]

I'll try that. I thought I had travel speeds turned down but I see now that I had it set at 100mm/s. If that's the cause, what would be the solution?

First establish it's the issue. I'd just set it to whatever my print moves are running at to verify.
Then you have to determine if the issue is mechanical, move the carriages by hand and see if there is undue resistance.
If not you can turn up the stepper drivers a bit in the firmware, if that doesn't work, reduce Jerk and acceleration in the firmware/EEPROM.

[ApacheXMD]

Ok so I've tried lowering all speeds including travel speeds to 10mm/s and the issue appears to be alleviated. So I guess that points to somehow the motors not being able to move the carriages.

So I've reduced belt tension a little and the cheapskates seem to move with a little less resistance. I've also tried readjusting the cheapskate eccentrics and that also seemed to free things up a little more.

I then tried printing with speeds set at 30mm/s including travel, and the issue returned.

I previously posted that I have reduced accelerations to 400mm/s^2, and now have reduced the jerk setting to 6mm/s and am trying to print again..

Edit: No go.. still shifting toward the X tower..

[Polygonhell]

Verify the XBelt isn't damaged, verify the set screw in the X pulley is tight, mine dropped out before I checked it when my prints were leaning.
What's the motor driver gain set to in the firmware?

[ApacheXMD]

Tightening the set screws was one of the first things I tried. They are tight with red locktite on both screws. Maybe I should grind a flat on the motor shafts?

By motor driver gain, do you mean the digipot settings? Currently theyre at 190 for the three towers. I wad trying a variery of settings but they dont seem to make a difference. I know that its possible to have motor driver current set too high and the driver will overheat and cause the motor to skip steps for sometimes brief moments. Thats why Ive gone as low as 135 and also tried aiming a 80mm fan at the rambo board during a print. But thst didnt cure anything.

I'm thinking of trying to swap x and y pins in firmware and see if the problem follows the actual tower or the driver port on the board. If it follows the tower then I can continue chasing mechanical gremlins. If it follows the driver then bad board maybe or firmware wierdness? Right now Im using polygon's 0.80 development branch. Maybe i should try seemecnc's version or polygon's 0.82. Poly: is the master branch working with the rostockmax?

[foshon]

Tightening the set screws was one of the first things I tried. They are tight with red locktite on both screws. Maybe I should grind a flat on the motor shafts?

By motor driver gain, do you mean the digipot settings? Currently theyre at 190 for the three towers. I wad trying a variery of settings but they dont seem to make a difference. I know that its possible to have motor driver current set too high and the driver will overheat and cause the motor to skip steps for sometimes brief moments. Thats why Ive gone as low as 135 and also tried aiming a 80mm fan at the rambo board during a print. But thst didnt cure anything.

I'm thinking of trying to swap x and y pins in firmware and see if the problem follows the actual tower or the driver port on the board. If it follows the tower then I can continue chasing mechanical gremlins. If it follows the driver then bad board maybe or firmware wierdness? Right now Im using polygon's 0.80 development branch. Maybe i should try seemecnc's version or polygon's 0.82. Poly: is the master branch working with the rostockmax?

This is a good idea whether it turns out to be your issue now or not.

[ApacheXMD]

Still fighting the issue here. John from Seemecnc emailed me back and suggested some settings for accel and jerk, but that didn't seem to help, I had already tried settings lower that his suggestions.

Just for poops and giggles, I turned acceleration values and jerk waaay down. 50mm/s^2 for all three axis and 2mm/s jerk. I limited max travel and feed speeds to 50mm/s as well. And the print seemed to be aligned!

So I'm going to keep playing with it slowly turning it up to see where I end up.

I might also end up doing a complete tower/belt/cheapskate teardown to see if I can find any bit of resistance, but the axis seem to be moving just fine to me.

Edit: one thing i noticed during the low acceleration test, is that the motions were very slow and the platform sped up and slowed down during each move. But on occasion during infill, the platform would come toward a perimeter and, instead of slowing down, it would hit the perimeter turn and continue its path with no apparent decelleration. This didnt happen all the time, maybe only 2 or 3 times per layer. It seemed to just ignore the acceleration settings in firmware every so often.

I dont know if what i observed has something to do with my alignment problem at large, but I do find that the layer shifts seem to happen after a series of short, rapid infill moves. Maybe acceleration settingd arent doing what I think theyre doing

[ApacheXMD]

Here's a video of my max with acceleration set to 150mm/s^2, jerk of 2mm/s and both feed and travel speeds set tp 50mm/s. You can see how most of the moves are slow to start and stop but every so often it seems to change direction suddenly without slowing.


[youtube]http://www.youtube.com/watch?v=Wkk6kOJrLsc[/youtube]

This particular was better with small layer shifts, but it's still present. So decent print at 50mm/s^2 and failed print at 150. Default acceleration is what 3000mm/s^2? Something is definately wrong with my setup. Just gotta figure out what. I think a complete teardown may be in order

[foshon]

Here's a video of my max with acceleration set to 150mm/s^2, jerk of 2mm/s and both feed and travel speeds set tp 50mm/s. You can see how most of the moves are slow to start and stop but every so often it seems to change direction suddenly without slowing.


[youtube]http://www.youtube.com/watch?v=Wkk6kOJrLsc[/youtube]

This particular was better with small layer shifts, but it's still present. So decent print at 50mm/s^2 and failed print at 150. Default acceleration is what 3000mm/s^2? Something is definately wrong with my setup. Just gotta figure out what. I think a complete teardown may be in order

Have you checked to make sure eeprom matches what you want it to be? I fought a retraction speed issue for three or four print starts becuase of that.

[ApacheXMD]

Have you checked to make sure eeprom matches what you want it to be? I fought a retraction speed issue for three or four print starts becuase of that.

Eeprom is where I've been setting the acceleration/jerk/max speeds. I know the settings are changing because the low accel values are very noticeable.

I took apart the arms and the belts tonight to chase down any resistance that might be hidden in the cheapskate carriages. I didn't think there was any before tearing it all apart, and unsurprisingly, the cheapskates move freely and without resistance or backlash. All three carriages seem to move as perfectly as I can probably get them. In the following video I test each tower immediately after releasing the belt with not further adjustment. So this is how the cheapskates were during all the shifted layer prints. I give each a wiggle to show that there is no play.

[youtube]http://www.youtube.com/watch?v=Is8MEd08Ylc[/youtube]

While I had the belts off, I took off the pulleys from the steppers. All set screws were still tight (they should be since I checked them about 4 times since building the printer), and there was no evidence of pulley slippage. No unusual wear marks where the set screws contact the shaft. Looked perfectly normal.

Next step is to reattach the belts, and make sure there's no belt rubbing on the bearing spacers or the extrusion. There really is no adjustment if the belt isn't running perfectly center in the slots of the towers, besides the top idler bearing. But I'll do the best I can to center the belts in the idlers and towers.. The hunt continues!

[dsnettleton]

Have you checked your wiring? If you have a loose wire or bad crimps leading to your X tower stepper motor, it could be shorting out. I just fixed a problem with my wiring and it turned out to be pretty obvious once I had a look up close. Erm... Also, I jiggled the wires in the middle of a print. That made things very obvious...

[ApacheXMD]

I did check the wire. It looks ok to me. I bought a crimping tool to do all my molex crimps so they should have a solid connection.

I also checked my motor coil resistance and they wer all consistently 1.1ohms on all coils on each motor.

[ApacheXMD]

Still working on the problem.

Saw this yesterday:

https://github.com/repetier/Repetier-Firmware/issues/99


Can some one try out the sample gcode mentioned in that thread? It does some rough things to my printer.

Code: Select all

G1 F15000 Z20 X-50

G1 X0 Y10
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y9
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y8
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y7
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y6
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y5
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y4
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y3
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y2
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y1
G1 X50 Y0
G1 X-50 Y0

G1 X0 Y0
G1 X50 Y0
G1 X-50 Y0

Fortunately a user posted a fix, and repetier is looking into fixing it properly.

Went from this:[youtube]http://www.youtube.com/watch?v=zO4rnhNJ0Yc[/youtube]


To this:[youtube]http://www.youtube.com/watch?v=nPojwTWTCq0[/youtube]

Seems to make a huge difference. Still need to test print.

Am i the only one experiencing this issue here? I kinda find that hard to believe, since we're all using the same firmware and hardware.

[Eaglezsoar]

Where is this fix you are speaking of?