Cause of "The Lean"

[cope413]

Hello Everyone...

I put my MAX kit together about 3 weeks ago. I successfully got a thin wall calibration cube to print nearly perfectly, but nothing else.

Every other print has a "lean" to it towards the X-Y towers. Aside from the lean, the layers line up cleanly and the prints look really good.

I've gone through everything I can think of. I've gone back and forth with Brian from Tricklaser and John. I've gone through almost every forum thread, though I may have missed something.

John told me today that he was confident that the pulley on the Z tower stepper was slipping. I pulled it apart, and the pulley was tight and hadn't moved. I did remove the set screws and re-set the pulley with fresh loctite.

I was very meticulous about my sanding, and I don't feel any play, but I did order the trick laser carbon arms to confirm it isn't the arms causing the issue. They haven't arrived yet, but here's what I've done/tried thus far:

Checked and used PTFE lube on all u-joints.
Calibrated extruder (My value was found to be 100.86 steps/mm)
Leveled bed
Shielded all stepper wiring with foil and heatshrink. Routed all wiring cleanly so there is no noise (after reading Apache's thread with a similar issue)
Calibrated delta radius - hot end moves consistently across the bed in all directions (slightly raises up right at the 140mm mark in all directions).
I've played with Jerk and Accel settings... Jerk from 2mm - 50, Accel from 100m/s^2 to 5000
Sliced the same STL with Slic3r (.9.9, .9.10), Cura, and KISS
Printed at 20mm/s speeds all the way up to 150. Slow print speeds generate less lean, but it's still noticeable. (picture of Emmett's gear was at 15mm/s)
Belts/Cheapskates all move freely. I may have had the belts a little too tight at first, but I've tried loose and tight with the same results.
Hot end is tightened firmly to the platform.
Tried printing off the center of the bed - all prints lean from Z tower towards X-Y towers - regardless of position on bed.
I have cardboard surrounding the printer, so there are no drafts or temp issues.
My IR thermometer shows correct temp on the bed (120C)


The carbon arms are scheduled to arrive on Wednesday, but it doesn't seem like they'll solve the problem.

I'm totally out of ideas, and pretty bummed/frustrated at this point, so any help is much appreciated.

[Cleveralias]

Sure does look like what I saw when I had a slipping pulley... I swore mine were good, too, but in the end the Z stepper was slipping. Did you file a flat on the stepper shaft? Might help while you're at it if you end up giving it another look.

Here's a few things I tried while chasing it that I didn't see mentioned:
Lower/raise stepper digipot and see how it changes the shifting;
Redid stepper crimps and connector;
Set to false "always check endstops" in config.h to rule out noise.

Hope that helps. Good luck.

[cope413]

My motors came with flats on the shafts. John told me this morning that he thought it was the pulley, so I did take it apart and re-assemble it. Sure didn't seem like it was shifting, but I suppose it's possible.

I did re-do the stepper crimps and connectors when I shielded all the wiring.

I'll make that change in config.h.

Recommendations for values to adjust on the steppers?

[cope413]

Tried another print after checking/resetting Z stepper pulley and making that change in config.h ... Still got the lean

Desperate for a solution...

[Polygonhell]

There are a limited number of possible causes.

loose pulley, which you think you've eliminated.
Missing steps on one axis, this can be caused by two things, the first is too much mechanical resistance on one of the towers, check them by moving the cheapskates by hand and feeling the resistance, if one feels different, figure out why.
It can also be the motor drivers on the Rambo overheating, point a fan at the board and see if it goes away.
It could be damaged belts, check the belt teeth, about 1/3 of the way up on the inside of the tower.
The final possibility is somewhat less obvious, but you should be able to see it happen if it's an issue, if your bot is not correctly calibrated, and if the head hits/scrapes the bed while printing the first layer, it can cause one axis to skip a significant number of steps, and everything printed afterwards is printed at an angle. Print the first layer painfully slowly and watch it, the first layer should be consistent and in line with the other layers in thickness.

[jetpad]

I started having a very similar problem about a month ago. The first couple layers would print correctly and then every one after that would shift very slightly in the same direction. At first I thought it was something mechanical and I tried everything to fix it.

Then I discovered a few of my EEPROM settings had spontaneously changed but the problem still wasn't fixed after I changed them back. At that point, I didn't trust anything to do with the EEPROM settings so I moved all the values into the firmware, recompiled it to disable reading from EEPROM and my problem disappeared.

[ApacheXMD]

I had a similar problem. My fix was to reroute my extruder stepper motor wiring far far away from the endstop wiring. Somehow it was triggering the endstops and causing a shift in my prints.

My troubleshooting thread

[cope413]

Yes, apache, I went through your whole thread because I thought for sure it was the same issue. I shielded all my wiring and made sure that noise wasn't a factor. Unfortunately, it didn't fix the issue.

Polygon, I will do as you suggest, but I've gone through those steps. Are those really the only possible culprits?

I've had two fans on my Rambo board from the beginning

My belts aren't damaged, the cheapskates move equally freely. But I will disassemble and check again because I don't have any other ideas.

I saw a thread about the laser cut parts not being properly aligned. I didn't check mine before I assembled. Could that cause the issue I'm seeing?

[cope413]

Also, I haven't tried adjusting motor current... Is that worth trying? Suggested values to try?

[Eaglezsoar]

I saw a thread about the laser cut parts not being properly aligned. I didn't check mine before I assembled. Could that cause the issue I'm seeing?

I believe that if you assembled without checking the alignment of the top and bottom plates that could cause your problem.
Cambo3D was the first to notice this alignment problem, hopefully he'll read this and give you his opinion on whether this
could be the cause of your "lean".

[kbob]

Your limit switch posts (the long screws sticking up from the carriages) could be at different heights. It would have to be really severe to make that much lean.

Your towers might not be equidistant. Again, I'm not convinced that could cause such a pronounced lean.

When you move the effector to X=0 Y=0 Z=1, is the nozzle directly above the middle of the bed? If something were geometrically skewed, it would be off center, I think.

Is the lean repeatable? Is it affected by how fast/slow the carriage moves? If you just move the carriage by typing G-code, does the effector move vertically or diagonally? If you start at different points in the X-Y plane, does that change?

(I don't have answers, just asking questions to help characterize the problem. I guess I'm trying to understand whether it's geometry, firmware bug, electrical issue, or what.)

Did you replace one of your motors with a 0.9 degree stepper, or one of your pulleys with a 16 tooth? (-:

And here's a real long shot that would match your symptoms: Is one of the Cheapskates upside down?

[lordbinky]

Before you fix it, I suggest getting any Tower of Pisa prints out of your system while it's easy.

[Batteau62]

It's your build surface clipped securely? I had some custom clips that let a print get knocked out of register. Shot in the dark

[cope413]

kbob, the lean is the same no matter where I start the print in the X-Y plane.

It is less noticeable at slower speeds. The emmett gear I took photos of was printed at 20mm/s. The other part was at 50mm/s. You can see the difference in the severity of lean.

Cheapskates are all correctly oriented.

Nozzle is centered with that gcode command. I'm fairly confident in my calibration of the nozzle,bed leveling, and delta radius calibration.

In the process of taking the printer apart to check the alignment of the laser cut parts, all motor pulleys, tower heights, and to confirm wiring is routed to avoid any possible noise. I will also be installing the TL carbon arms that are arriving tomorrow.

Thanks for your help/suggestions. I will update as soon as I get it back together and test.

[Polygonhell]

oops

[626Pilot]

I wonder if it would help to solder 0.1uF capacitors across the limit switch wires to help cancel out any current getting induced in them. They're used to dump any line noise where power is going into a microprocessor. I learned today that these switches are normally CLOSED, meaning that current is flowing through both leads all the time, except when something is hitting the limit switch. Not a big surprise that they could be interacting with current being sent to the stepper motors.

My bet is on the towers not being aligned properly, to where the aluminum extrusions are touching the inside edges of the notches they sit in and the tower braces up top are fully engaged with the build plate and equally tall. There are a lot of degrees of freedom involved in that and it can be a real bear. I was thinking of tying string around all three towers down below and up top, with some kind of ratcheting device to tighten the string down and thereby bring the towers as close together as possible. There is also a .stl someone designed of tower clamps that are connected with threaded rod and tightened down, but I don't have the link.

[Polygonhell]

My bet is on the towers not being aligned properly, to where the aluminum extrusions are touching the inside edges of the notches they sit in and the tower braces up top are fully engaged with the build plate and equally tall.

I think the towers would have to be miles off to get the sort of result he's seeing, you would probably be able to look at the machine and say that's not straight.
The fact it's less pronounced with slow motion implies that it's a stepper skipping because of a mechanical issue or something slipping.

If the issue were software or electrical, speed would likely make no difference.

[kbob]

My bet is on the towers not being aligned properly, to where the aluminum extrusions are touching the inside edges of the notches they sit in and the tower braces up top are fully engaged with the build plate and equally tall.

I think the towers would have to be miles off to get the sort of result he's seeing, you would probably be able to look at the machine and say that's not straight.
The fact it's less pronounced with slow motion implies that it's a stepper skipping because of a mechanical issue or something slipping.

If the issue were software or electrical, speed would likely make no difference.

If one motor skipped steps (or its pulley slipped), wouldn't it make the firmware's concept of X-Y plane tilt more than it would make it move sideways? The prints look like the layers are staying horizontal.

[Nylocke]

This is a guess, you could double check your firmware to see if the Z tower acceleration (I'm thinking acceleration might be a possibility because it kind of sounds like it could be caused by the z tower not being able to catch up with the other towers, and you getting less lean from a slower print seems to make it more sound, because it doesn't have as high a velocity to accelerate to) and max velocity are the same as the other 2 towers, though kbob does have a point, it should make the print slanted rather than offset, though repetitive loss of steps or the like in both directions maybe could cause this? this is normally a symptom of a cartesian printer with poor axis lubrication or not enough power to the stepper or the like, I wouldn't even think a delta could pull something off like this...

[Polygonhell]

If one motor skipped steps (or its pulley slipped), wouldn't it make the firmware's concept of X-Y plane tilt more than it would make it move sideways? The prints look like the layers are staying horizontal.

Yes and when i look at the pictures at the start of the thread, the top of the print doesn't look horizontal to me.

[bvandiepenbos]

hey cope413

I think it is motor losing steps. Try increasing your motor current.
Also, run from SD card not from computer (USB communication can be a issue)

Your part printed fine on my machine.
1.75 ABS 210c, Glass with hairspray 75c, .2 layers, .3 fill, 60 feed, 4.8mm retract/35 speed/6mm lift/.1 extra on restart, .35 J-Head, Slic3r 9.10b, Repetier firmware v.90c

[PartDaddy]

Polygonhell, you sure know your stuff...

1. After reading through this post, I would look at wiring. This seems like it could be noise related, which means it can be tricky to pin point.

2. The next thing is cutting your acceleration values. Cut them in half or more just to see.

3. Put a fan on the stepper drivers on the RAMBo. The EZStruder requires more current. I've found it necessary to add a fan on the driver to keep it from over-heat shutdown. The symptom was exactly like the extruder skipping. So your skewed print might be skipping due to a similar situation.

[JohnStack]

Polygonhell, you sure know your stuff...

1. After reading through this post, I would look at wiring. This seems like it could be noise related, which means it can be tricky to pin point.

2. The next thing is cutting your acceleration values. Cut them in half or more just to see.

3. Put a fan on the stepper drivers on the RAMBo. The EZStruder requires more current. I've found it necessary to add a fan on the driver to keep it from over-heat shutdown. The symptom was exactly like the extruder skipping. So your skewed print might be skipping due to a similar situation.

Woah! Ok. What size fan? 12 Vdc?

[626Pilot]

What size fan? 12 Vdc?

There are mounting holes for 40 and 50mm fans on the RAMBo mount. I don't know what screws to use, though. There's ample 12V power next door.

[bvandiepenbos]

What size fan? 12 Vdc?

There are mounting holes for 40 and 50mm fans on the RAMBo mount. I don't know what screws to use, though. There's ample 12V power next door.

screws are M3
or use 4-40 screws.

I like to use t-nuts, but regular nuts nuts work fine also.

have the fan blowing against back of RAMBo board, not drawing away from it.

If you already have fan on back of board and chips are still getting hot, add another fan blowing across chips on front of board, even a tiny 25 or 30mm fan is sufficient.