Layer shifting - help!

[neilcoggins]

Hi,

I am getting layer shifting in the X-axis only on nearly every print (anything from 2cm^3 upto 20cm^3) on my M201. I thought I had pinned it down to too-low voltage on the A4988 (was 0.76V) - I increased to 1.14V and it seemed to fix the issue for smaller prints. I then tried a decent-sized print (200x120x35mm), which worked perfectly for 11 hours. It then decided to lock up the comms completely (grr!), so I had to disconnect/reconnect/edit g-code/start printing again (no problems there). Got another 2 hours in, and then I got a layer shift of ~15mm in the -ve X-axis - 13 hours of print ruined completely.

I have checked for the X-axis binding (one LM8LUU bearing lost a ball in the box from Geeetech before it was even installed), but it seems to move freely (I've ordered some Igus Drylin bushes to be sure). I've tried various tensions on the X-axis belt - no change. I've reduced the travel rate to 2400mm/min in Repetier Host (down from the default 4800mm/min) - is this still too high? On the odd occasion, I've heard the machine "grinding" very loudly - I'm fairly sure this was the stepper not having enough current to move the carriage, and I don't think it happened with the last big print (although I was in the next room, so can't be sure).

I've reached the point where I think I should try a different stepper driver. I have obtained a few DRV8825 units (with 100R current-sense resistors), primarily because they're 4-layer PCBs and better able to cope with heat dissipation. (BTW can the GTM32 cope with 1/32 microstepping? If it's a binding issue, then the DRV8825 would potentially improve that by moving smaller steps anyway...)

Does anyone have any other suggestions for what I can try to fix this? I've not managed a decent print in 3 weeks / 2kg of PLA so far, and I am losing the will to live somewhat!

Neil.

[Mark]

have you tried to interchange the driver and motor with the other axis? could it be a bad driver and motor? and the bearing, better to change a new one since it lost a ball.

[neilcoggins]

I have tried using a DRV8825, but that didn't seem to work any better; I've also tried a StepStick which I had in my drawer, but that didn't improve things either. I'm running with the X and Y stepper drivers swapped at the moment, and it appears to beworking (only done 1 30-minute print so far though).

I've ordered replacement LM8LUUs and a set of Igus Drylins (if the problem is the carriage binding on the rails, the Drylins will put a stop to that, at the expense of wearing out a lot quicker). I'm going to write a test Gcode file to exercise the carriage without actually extruding anything. This should allow me to eliminate the possibility of the layer shifting being the head hitting the print (if it shifts when the hotend is 100mm above the heatbed and it isn't printing).

One thing that I have read about online is the possibility of the GTM32 overheating - is that possible, and if so, should I point a decent-sized fan at it for a long print?

I've not heard of many cases where a stepper motor itself went bad - I guess it couldn't hurt to get a new NEMA17 or two in as spares anyway! I'll post an update after my next big print (it's ~5h of a single part for an MPCNC machine, and will be done this evening).

[neilcoggins]

OK, so it looks like the problem is a combination of things...

The carriage was binding in the X axis because of 2 separate issues:

• 1. the damaged LM8LUU (for which replacements are en-route)
  2. the belt being too tight

I've loosened the belt by 3 turns of the adjuster wing nut, and I now get reasonable prints roughly 50% of the time (although the prints are always too small, but that's obviously a separate issue). I'll be replacing the bearings over the weekend to see if that resolves the final issues...

On the subject of prints being too small, I'm not entirely sure what it can be... I have an STL from Thingiverse that should have a 25mm hole through the middle of it (I've verified this in FreeCAD and Sketchup). Printing it, I get 24.2-24.4mm, depending on whether I slice in Cura (worse) or Slic3r (better). If I scale the STL by 102%, the Cura slice comes out just about perfect (which is what I have been doing, since I need several of these parts and every other print layer-shifts and has to be thrown away).

I can't see anything in the settings that would cause this inaccuracy. I haven't changed the X/Y steps/mm (both are still set to 80 in the LCD), but I don't believe the belt has stretched 2% in both directions. Am I missing something?

[Mark]

I think you need to calibrate the XYZ steps/mm, check this link, it says the same problem too:
http://www.geeetech.com/forum/viewtopic ... 26&t=17333

[neilcoggins]

Having loosened the X-axis belt, and tweaked the A4988s for both X and Y, I can now print 90+% of the time without layer shifting. The remaining layer shifts appear to relate to an LM8LUU binding on the XZ carriage which is probably related to the bearing having lost at least one ball before it was even fitted - I found one loose in the bag with the bearings before it was even assembled!)

I've got replacement bearings to fit to the carriage, but I'm going to print some replacement parts for the Z-axis first to address some other issues (mainly Z-axis backlash and Z-axis wobble). Hopefully once I have all this printed & installed, the layer shifting will be eradicated...

[stado65]

@neilcoggins

I almost eliminated all the layer shiftings (X and Y) on my M201 by dramatically reducing the print speed.
According to me, the speed values in the config file provided for Slic3r (Normal.ini) are far too high for this printer !
I use 40 mm/s for all speeds, except for bottom layer (20 mm/s) and for travel speed (100 mm/s). That's much better...

[neilcoggins]

@stado65

I think I have fixed the layer shifting completely now - I've done 2 things, so could be either/both:

• Added an 80mm fan blowing across the GTM32
• changed the A4988 on the Y-axis to a DRV8825

Works like a charm now... printed 4 parts @ 4 hours each with no hiccup yesterday, so fingers crossed!

I'm using Repetier's built-in Cura, BTW - mainly because I am looking to migrate to OctoPi (when the M201 will work with it, anyway!). I get pretty good results considering the issues I've had (although I'm now starting to get extruder issues - Triffid Hunter's guide will be on the screen again tonight I suspect).

[stado65]

Good news !!
I think the real thing is the relacement of the A4988 : with the DRV8825, you can reach higher current in the motor and it does not shift anymore. Personnally, I also think of replacing these 42 mm motors with 48 mm ones.

I use Octopi for all my prints with external Cura slicing. Apart from the communication issues, it works well and allows me to watch the printing over the internet with the webcam (and potentially stop it from anywhere if I see that something goes wrong).
With it, I have almost finished to print the MP CNC parts (1,6 kg PLA)

[elocm]

You should reduce the "acceleration" via menu on the LCD display from 5000 to 500. This reduce the inertia issue that overload the current driver and stepper (had the same problem on the X-axis...). Nevertheless the replacement of the driver is also an idea, but this should help in the intermediate. Do not forget to STORE the setting, otherwise it got lost after the net reset.