Layer shifting

[ApacheXMD]

Its in that same github discussion.

https://github.com/repetier/Repetier-Fi ... t-17202043

Motion.cpp is modified with new float definitions for dx and dy.

Later in the thread, repetier said made some other change though that fixes the problem too but i havent looked at what he did

[Eaglezsoar]

Polygonhell, do these fixes he is talking about need to be incorporated into your version of Repetier?

[cambo3d]

so this fix solved your layer shifting problem?

looking at polygons firmware and the main branch there seem to be more than just changing those lines of code. in the motion.cpp

i'll wait til polygon gets this update in his version.

[ApacheXMD]

so this fix solved your layer shifting problem?

I didnt get a chance to do a test print but i think it is a related problem. More testing is forthcoming.

[Polygonhell]

I'll pull the change over when I get a chance, might not be until the weekend though.

[Flateric]

So this is a worthwhile change then poly? Worth the reflash of firmware?

[ApacheXMD]

Did anybody try the gcode i posted? Are you guys seeing the same phenomenom? Becasuse if you are, and printing fine, then I guess it puts me back to square one.

[Polygonhell]

So this is a worthwhile change then poly? Worth the reflash of firmware?

Probably though I'd follow the if you don't have issues don't change it rule of thumb.
As far as I can tell from the discussion in the bug, and I've only been vaguely following it, there is an issue where calculation of the speed in one place in the delta code is incorrectly signed and this leads to a case where you can get a sudden change of direction and the acceleration/jerk settings are not respected. I haven't looked at the code or the fix, and the last post I saw in my email suggested there was a fix, but it hadn't been tested by anyone with the issue.
Minimally this could affect print quality, worst case it could result in missed steps, though you'd know if you were getting those.

I currently don't have a running delta printer, so I can't trivially pull the code and test it.

[ApacheXMD]

So although motion has improved, it has not corrected my layer shift issue.

Notice the squirrel's head, it's shifted over to the left.

Look at the left edge of the squirrel's tail, the layers are misaligned all the way up.


Now I'm really at my wit's end. Motion is smooth, and there isn't any mechanical binding, acceleration is already lowered, jerk is lowered. Motors ohm out equally at 1.1ohms for every coil, belts have been tightened and loosened, motor currents have been set high and low, magnetic joints have been swapped in and out, and I am getting tired of this. I must be doing something wrong and I don't know what.

Bad motor? Bad rambo motor driver? I don't know.

[Polygonhell]

Clearly you are losing steps in what ever way that's leaning on the bed.
Have you tried swapping the bad axis onto the extruder 2 steeper output to rule out the electronics?

[ApacheXMD]

No i havent. Would that just be a change in pins.h? Anything else i would need to change?

[Polygonhell]

Yes just swap the extruder 2 step dir and enable pins with whatever axis, the swap the connector into the socket.

[Flateric]

Has anyone considered the possibility that this shifting is being cause by noise?

I saw a very similar situation with my original CNC machine I made with homemade drivers.

The fix was super easy. Twist all of your wires in twisted pairs and optimally add a ferrite coller and run the wires through this a few times as well.

Also check you wire routing, Try to avoid passing the stepper wires directly over and or touching anything AC (anything near the AC cord is the most potential for electrical noise) and away from each other (other steppers wires).

Obviously you cannot avoid some proximity routing in some cases, but doing your best to avoid when possible goes a long way towards occasional missed/unwanted steps.

You may be surprised and this could cure your problems.

[ApacheXMD]

Yes just swap the extruder 2 step dir and enable pins with whatever axis, the swap the connector into the socket.

Thanks Poly.

So here's the results of the pin swap extravaganza. In openscad, I created another test pattern so I could better see differences between each change. It's just three 10mm cubes arranged in an equilateral triangle, with faces facing each tower.

The setup

The center group is the control group. Each stepper motor on it's default driver. Lower left is X motor on extruder1 pins,
lower right, Y motor on extruder 1 pins, and top group is Z motor on extruder 1 pins.

Control group

Control group side

This is the control group.

Z swap

Z swap side

This is the Z motor swapped onto the Extruder1 port. Looks pretty much the same as Control.

Y swap

Y swap side

This is the Y motor swapped onto the Extruder1 Port. Obvious lean away from Z tower.

X swap

X swap side

This is the X motor swapped onto the Extruder1 Port. General waviness.

So how does one interpret these results? I'm not too sure. Seems like since the Z motor swap didn't change a whole lot, the Z motor driver isn't the culprit. But since changing both X and Y messed it up even more compared to Control, I'm not sure where that leaves me.

[Polygonhell]

I don't have a good answer for what's happening, I would want to try different electronics to rule that out, but I'm in a different situation than most, I have spare boards I could try.
Have you spoken to SeeMeCNC about this?

[ApacheXMD]

I don't have a good answer for what's happening, I would want to try different electronics to rule that out, but I'm in a different situation than most, I have spare boards I could try.
Have you spoken to SeeMeCNC about this?

Yes I have. John from Seemecnc suggested some lower jerk and acceleration settings as well as limiting max speeds, all of which I've tried. John did mention that he'd try to send me another rambo to try, but I havent heard from him since his last email last friday. I just gave them a call and left a message.

[ApacheXMD]

So i was able to talk to steve on the phone, and he ran down the list of possible causes. And I told him I've tried just about everything. His last suggestion is to do a fresh flash of the seemecnc repetier firmware one last time. So I did just that, and did a test print. But the issue still remained. I really didnt expect a different result.

So I decided to try Flaterics suggestion and try twisting the stepper motor wire into twisted pairs for each coil. Removing the crimped pin sockets from the connector body is a simple operation if you use a tiny flat blade screwdriver to push into the slot on the back side of the connector. What you're doing is pressing down on the barb thst holds the pin socket in place.

After twisting each pair, i reassembled and retested.... with the same result. No improvement. But it was worth a shot.

I then decided to play with motor currents. I had previously tried varying the digipot settings from 135 to 200 without much difference. So I had the current setting at 195, and measured the driver chip temperature with a thermocouple. During a dry run print at speeds of 30mm/s, drivet temperatures hovered around 200F. I then dropped current settings to 100. And during the same dry run print, temps were in the 110F range. A significant temperature difference.The motors sounded different too, with a lower pitched sound. I did a real test print with current at 100. But still saw layer shifting. The shifting however seemed to be different though, now kinda moving away from the Z tower. I'm not sure what that means.

I had previously tried point a 80mm fan at the rambo board during a print but that didnt help either.

So that wad it for last nights trouble shooting. I'm still hoping for SeemeCnc to tell me that they are for sure sending me a replacement Rambo to try out. Im running out of ideas.

Oh one more thing I tried briefly. I tried loading seemecnc's version of Marlin firmware, but i couldnt get it to flash correctly or something because the Lcd didnt work, and sending a g28 just made the motors vibrate and make funny noises.

So thats it. If anyone has any other suggestions Im all ears and am willing to try anything.

[ApacheXMD]

Problem has been solved!

As far as I can tell, it was the endstop wiring picking up noise. During my build, I ran all of my endstop wiring up a the middle of a single 8020 extrusion. And in a stroke of genius I decided mount the extruder on the top platform, which meant running the extruder motor wires up the same column. Well it was a dumb idea because when I rewired the extruder motor to run off an external cable, all my layer shifting issues disappeared! Lesson learned, keep those endstop wires a safe distance from stepper motor wires!

Thanks for all the help guys, turns out it was EMI afterall.

[cambo3d]

what's weird is that it does it randomly. With steppers motors running constantly during your prints. I would think you might see it throughout the print.
I always use shielded wiring; not just for the stepper motors, but for all the wiring where its feasible.

[ApacheXMD]

Yeah it wasnt a constant shift. I can't really explain it but the only difference now is the extruder stepper wiring. I do have the extruder on a shielded wire cable now. I'm just glad this issue is solved. Now I'm battling abs warpage and curling.

[Flateric]

Problem has been solved!

As far as I can tell, it was the endstop wiring picking up noise. During my build, I ran all of my endstop wiring up a the middle of a single 8020 extrusion. And in a stroke of genius I decided mount the extruder on the top platform, which meant running the extruder motor wires up the same column. Well it was a dumb idea because when I rewired the extruder motor to run off an external cable, all my layer shifting issues disappeared! Lesson learned, keep those endstop wires a safe distance from stepper motor wires!

Thanks for all the help guys, turns out it was EMI afterall.

Ding Ding, I win a.......gold star?

[ApacheXMD]

Flateric, your suggestions did help. It got me thinking about emi and noise and though merely having twisted pairs didnt help my problem, it led to an intense google search of steppers and noise. That led me to researching shielded wire and steppers; some saying use it, others saying not necessary. That led to a post of someone saying sometimes it's not so much keeping outside noise from affecting stepper wires, but keeping noise IN! That prompted me to remember how I routed the extruder wiring and endstops in the same path in the extrusion (which I initially thought was providing a shielding effect). I had read before how end stop switches can pick up noise from AC sources, but didnt think too much of it. I thought since the end stops were wired to Normally Closed (NC) to ground, noise pickup would be negligible.

Anyway, it wasnt a light bulb going off nor was there a "eureka" type moment. It was just one more thing that I tried. But thanks everyone for all your help.
I was at the San Francisco makerfaire this weekend and saw a ton of 3d printer that made me drool, but this forum is a big reason why I'm happy to have bought a Max.