Welcome! Log In Create A New Profile

Advanced

Z/Layer Skipping & Gcode Repeating; High Speed Printing

Posted by Gunslinger 
Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 08, 2018 12:59AM
To summarize the problem: my printer on certain gcode outputs is skipping Z movements/entire layers and/or it is repeating the same Gcode 20+ times in the middle of a print.

The long version: my printer used to run ABS up to 90mm/s at 260C with Z lift enabled for travel. I always print at 100% line infill with only one perimeter wall, due to making parts used in mechanical devices and designing them myself. The only quality issues I ran into was large areas under extruding (fixed by raising the temp to 260C), poor cornering (fixed by >3mm fillets), and the infill extruding to much material when it U-turned (seemingly due to being at 260C and acceleration).

Due to the signs of a dying hard drive, I reinstalled Windows 7 and installed the newest versions of CraftWare (from 1.14 to 1.16) and Repetier-Host (1.4 or 1.6 to 2.05); my printer is a Folgertech Prusa I3. Neither CraftWare or Repetier-Host apparently kept the settings in a convenient directory when I backed up data, and I wanted to try new settings to address the above problems better. I reflashed the printer's firmware increasing its default X & Y accelerations and top speed. The printer currently is running up to 120mm/s for infill with 3+ perimeter walls, with the only issue being the corners are slightly oversized due to the sharp corners from inner perimeter walls (making the areas between corners slightly undersized, the original reason for a single perimeter wall). The machine hasn't undergone much maintenance aside from making sure the motors are getting ~0.45 volts. The gcode is ran by Repetier Host; the machine does not have an SD card. The machine does not have a cooling fan, but has not had this much trouble even for individual small parts before.

The problems: Repeating gcode [Solved, solution below]: The printer occasionally decides to repeat the same pieces of Gcode over and over in the middle of printing. The two times it was observed closely occurred on the same models and same layer when sliced using 99 perimeter walls. Both times, even with repositioning and reslicing the parts with fewer walls, the start of layers 5 an 6 would repeat the innermost concentric gcode over and over, 20 to 50+ times. This model being a long rectangular shape, it was extruding material from 10,10 to 10,40, then moved back to 10,10 and did it again. This resulted in an excess amount of plastic that the printer kept pushing around. It also appeared to occur a little on another part, but the printer pushed the excess plastic out of the way. It has stopped since reducing the settings to only 3 perimeter walls (the new layer starts with infill instead of a perimeter), but this means it could happen on smaller parts. CraftWare animates its extrusion path and does not show the repeating code. Looking at the gcode manually, there were no odd commands and both the X and Y coordinates were constantly changing >1mm. I tried downgrading to Repetier Host 1.6, but it did not resolve the issue.

Z/Layer Skipping [possibly still at large]: Following the previous issue, I've ran into a print job where Z inputs or entire layers are getting eaten somewhere. It was a 12 hour print, so I did not get to observe the mistakes. I've printed a complex object with no observable mistakes that was 34.02mm of its 35mm height. On this print job, objects that were suppose to be 20.75 mm in height printed to 21.06mm. On this same job, 4 objects were suppose to be 68mm in height, but only printed 59.23mm. I manually moved to 68 mm height with Repetier Host and measured it to be 68mm. There were 272 layers, 0.25mm each, accounting for 68mm. Half way through the print the X motor slipped on over-extrusions (possibly from repeating code) or due to Z shifts/inaccuracy, shifting the objects in the X direction. Very little warping occurred on the parts due to having a heat wall around them. A feature that is both above and below the shift measures to the dimensions it should have been printed. Above this feature and even at this feature's height, several features are cut significantly short. Normally holes look like O, but at one height all the holes are wide () shaped. A whole 1.5mm of layers just disappeared at one location; no observable over/under extrusion other than the layers being a fraction of a millimeter wide. The only signs of defect are slight over-extrusion and/or the printer plowing the print at the height the X shift occurred, which measures on the dot where the height should be at, 34mm. The printer finished the parts just fine, with the final feature and 9mm printing just fine. Prior to that 9mm, 6mm of 12mm of layers are missing. Features of multiple objects are at the same height, indicating layer-wide issues, no Z lift issues.

Layer Skipping Possibilities: Being a 12 hour print, its possible something may have started to overheat in the circuitry. The machine could be working faster than it can receive gcode, but this doesn't explain why its printed parts that are seemingly more complex (small movements) and why it could print the first 34mm and final 9mm just fine. The X shift may have been caused by CraftWare over-extruding areas of the part, resulting in localized bumps in the Z direction, but still does not explain the Z skipping in later layers. This issue may exist with other slicers as well. I attempted using other slicers to see what new features they had; the reason I swapped to CraftWare from slic3r a while back. A separate install/use of slic3r (not Repetier's native one) failed to move the Z axis at all when it printed (it tried printing into the glass), and Cura's finishing gcode would not execute "G1 Z21" as it does in CraftWare (Cura's infill/perimeter overlap was sloppy too).


Resolution: Currently I'm attempting to print the same objects, but with only half the objects to print. I'm fairly certain the repeating issue is software related, but the second issue could be overheating and/or memory related, being that I rarely do prints that are >6 hours. The only printer change not Windows/software side was re-leveling the bed, increasing the Z motor voltage from ~0.25v to ~0.45v (due to slic3r's and Cura's failures to move the Z axis), and putting on a new nozzle twice due to clogging (cheap filament and/or occurred from over-extrusion while testing new settings). The only other major issue I currently have is CraftWare randomly deciding to fill in holes every once in a while.



Edit:
Repeating Issue: Solved: CraftWare was responsible for the repeating Gcode. It even shows in CraftWare's animated movements, but due to the stacked extrusion paths, only a few pixels of the animations were showing, making it not obvious. I resliced the same parts using the 99 walls setting, and this time on layer 1 they repeated the Gcode. After repeating the same section 20+ times, CraftWare then broke the loop it was stuck in and wrote the normal Gcode. Snippet of some of the repeating Gcode below, with repeating code bolded (only E values changed). Layer 1 returned to that one position and a second position 40+ times in a row. Layer 1's code length amounted to ~25% of the total code length for a rectangular shape that got larger than it over 40 layers. CraftWare's mistake seemed to occur sometimes when the central most infill/path was only one extrusion width wide (odd number of concentric walls), hence why 99 walls made many issues on a part that V's out, nearly every other layer was an odd number of concentric walls [See image].
G1 X76.616 Y10.588 E19.0766
G1 X76.618 Y10.474 E19.0836
G1 X76.620 Y10.360 E19.0907
G1 X76.622 Y10.246 E19.0979
G0 X76.593 Y11.846 F10800
G1 X76.595 Y11.731 E19.1043 F4800
G1 X76.597 Y11.617 E19.1108
G1 X76.599 Y11.503 E19.1173
G1 X76.601 Y11.388 E19.1239
G1 X76.603 Y11.274 E19.1306
G1 X76.605 Y11.160 E19.1373
G1 X76.607 Y11.046 E19.1441
G1 X76.609 Y10.931 E19.1509
G0 X76.607 Y11.060 F10800
G1 X76.607 Y11.046 E19.1517 F4800
G1 X76.609 Y10.931 E19.1586
G1 X76.611 Y10.817 E19.1654
G1 X76.613 Y10.703 E19.1724
G1 X76.616 Y10.588 E19.1793
G1 X76.618 Y10.474 E19.1864
G1 X76.620 Y10.360 E19.1935
G1 X76.622 Y10.246 E19.2006
G0 X76.593 Y11.846 F10800
G1 X76.595 Y11.731 E19.2070 F4800
G1 X76.597 Y11.617 E19.2135
G1 X76.599 Y11.503 E19.2201
G1 X76.601 Y11.388 E19.2267
G1 X76.603 Y11.274 E19.2333
G0 X76.613 Y10.717 F10800
G1 X76.613 Y10.703 E19.2342 F4800
G1 X76.616 Y10.588 E19.2412
G1 X76.618 Y10.474 E19.2482
G1 X76.620 Y10.360 E19.2553



Workaround: CraftWare does not offer Concentric infill as an option, but I'm currently printing the same part successfully using 99 top/bottom infills using the Concentric setting in place of using Wall Perimeter and the Line Infill. The only downside is CraftWare caps the setting at 99, so I can only print objects up to 49.5mm (0.25mm layer height) with 100% Concentric top/bottom layers; any more layers have to use Line Infill with 1 Wall, since CraftWare apparently doesn't like a Wall Perimeter greater than 1 wall. I changed the 68mm tall model to be printed in a different orientation due to the Z axis errors and to better strengthen some areas, so I got it down to 50mm, meaning only two layers will have the Line infill. I'll check the CraftUnique forums for a workaround of the 99 top/bottom layers cap.

Edited 1 time(s). Last edit at 03/09/2018 01:17AM by Gunslinger.
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 08, 2018 04:53AM
That's a very odd one...
-I'd say temporary data loss, due to memory failure.
-PCB's / drivers gone to hot.

First thing I would try is print a bit slower.
Second thing I'd try is swapping computer. While printing do not multitask. In the past I had very strange behaviour of my printer due to multitasking. It disappeared completely when I left that PC alone. One of the things I always do these days, before installing any printersoftware, I strip all unnecessary tasks and also all unnecessary Windows services. + on my printer-laptop there's no network/internet at all. The one I'm using now myself is a small eeePC only 1Gb ram on board, 2 printers installed(UScool smiley, no problems at all. Designing/drawing parts and slicing too is done a separate computer.
You don't have to to do all these things I mentioned here, but swapping PC is a important step. So doing you eventually could eliminate the computer itself from the problems you're trying to eliminate. If you do, do install yuor 'old' printer software, so you can eliminate that too.
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 08, 2018 07:41AM
You're missing the forest for the trees. I suggest you start paying less attention to the details of which host/slicer creates which bizarre behavior in which prints and start looking at the big picture.
You're experiencing similar problems with multiple host programs and slicers. What's common? The computer, the USB cable, and the printer's controller board.

Your problems lie with one or more of those.

It has been demonstrated that USB can't keep up with fast printing. Why are you still trying to use USB for fast printing?
Windows is unreliable. Put linux on the host computer if you must use a host computer.
Has the printer ever been calibrated? Have you ever set the acceleration and jerk, or are you just using default values?


Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 08, 2018 08:42AM
Although there's nothing wrong printing with Windows, I'd give it a t try myself.
Problem is I do not know where to start, never used Linux..!
But that's another problem, sorry for OT message.
Back to topic...

A common mistake is to use very long USB cables. I always use properly shielded and very short USB cables. Had lots of trouble with longer cables in the past.
And indeed as in that last reply : acceleration and jerk should be dealt with. But not for these problems imho. But if so please the_digital_dentist do explain. I'd like to understand why.
The thing is, this printer seems to get stuck in a repeating loop once in a while.

Thinking out load here, You never had this problems before in the past. And now you have these very strange issues..?(never heard of it before as a matter of fact) Did you by any chance relocate that printer...?

Edited 1 time(s). Last edit at 03/08/2018 08:43AM by chriske.
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 08, 2018 09:39AM
If the Z axis motor isn't moving the extruder up or the bed down, it can look like the gcode is stuck in a loop with the extruder moving in the same pattern on the same layer over and over.

Check the Z axis mechanically- are the screws properly fastened to the motor shaft? Is the cable between the motor(s) and the controller board in good shape? Is the current set to a reasonable value? Is the driver working properly?


Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 09, 2018 01:11AM
Repeating Issue: Solved: CraftWare was responsible for the repeating Gcode. It even shows in CraftWare's animated movements, but due to the stacked extrusion paths, only a few pixels of the animations were showing, making it not obvious. I resliced the same parts using the 99 walls setting, and this time on layer 1 they repeated the Gcode. After repeating the same section 20+ times, CraftWare then broke the loop it was stuck in and wrote the normal Gcode. Snippet of some of the repeating Gcode below, with repeating code bolded (only E values changed). Layer 1 returned to that one position and a second position 40+ times in a row. Layer 1's code length amounted to ~25% of the total code length for a rectangular shape that got larger than it over 40 layers. CraftWare's mistake seemed to occur sometimes when the central most infill/path was only one extrusion width wide (odd number of concentric walls), hence why 99 walls made many issues on a part that V's out, nearly every other layer was an odd number of concentric walls [See image].
G1 X76.616 Y10.588 E19.0766
G1 X76.618 Y10.474 E19.0836
G1 X76.620 Y10.360 E19.0907
G1 X76.622 Y10.246 E19.0979
G0 X76.593 Y11.846 F10800
G1 X76.595 Y11.731 E19.1043 F4800
G1 X76.597 Y11.617 E19.1108
G1 X76.599 Y11.503 E19.1173
G1 X76.601 Y11.388 E19.1239
G1 X76.603 Y11.274 E19.1306
G1 X76.605 Y11.160 E19.1373
G1 X76.607 Y11.046 E19.1441
G1 X76.609 Y10.931 E19.1509
G0 X76.607 Y11.060 F10800
G1 X76.607 Y11.046 E19.1517 F4800
G1 X76.609 Y10.931 E19.1586
G1 X76.611 Y10.817 E19.1654
G1 X76.613 Y10.703 E19.1724
G1 X76.616 Y10.588 E19.1793
G1 X76.618 Y10.474 E19.1864
G1 X76.620 Y10.360 E19.1935
G1 X76.622 Y10.246 E19.2006
G0 X76.593 Y11.846 F10800
G1 X76.595 Y11.731 E19.2070 F4800
G1 X76.597 Y11.617 E19.2135
G1 X76.599 Y11.503 E19.2201
G1 X76.601 Y11.388 E19.2267
G1 X76.603 Y11.274 E19.2333
G0 X76.613 Y10.717 F10800
G1 X76.613 Y10.703 E19.2342 F4800
G1 X76.616 Y10.588 E19.2412
G1 X76.618 Y10.474 E19.2482
G1 X76.620 Y10.360 E19.2553



Workaround: CraftWare does not offer Concentric infill as an option, but I'm currently printing the same part successfully using 99 top/bottom infills using the Concentric setting in place of using Wall Perimeter and the Line Infill. The only downside is CraftWare caps the setting at 99, so I can only print objects up to 49.5mm (0.25mm layer height) with 100% Concentric top/bottom layers; any more layers have to use Line Infill with 1 Wall, since CraftWare apparently doesn't like a Wall Perimeter greater than 1 wall. I changed the 68mm tall model to be printed in a different orientation due to the Z axis errors and to better strengthen some areas, so I got it down to 50mm, meaning only two layers will have the Line infill. I'll check the CraftUnique forums for a workaround of the 99 top/bottom layers cap.


Skipped Layers: Giving it more thought, the Z axis issues can't be any kind of motor slip, due to no signs of partial layers or over extrusion. Since I sliced those models with 3 walls, and there are thin sections on the part, its possible the X slip occurred due to one of those thin sections producing repeating Gcode and making a bulge of plastic. As for the skipping layers, I'm still out of any guess other than CraftWare issues; although normally CraftWare's problem is deleting an entire layer (visible in both CraftWare and Repetier Host's preview), but even that still moves the total amount of Z. In this scenario, the Z movements weren't even performed, and all the Gcode between it and the next layer appeared to have been skipped.

Printing the same model as before, but with only half the objects in 6 hours, yielded 66.45mm; 6 layers short. It is not the Z steps/mm, because a feature at 28.5mm measures 28.42mm, and a feature at 22mm measures 21.92mm. A feature at 48mm measures 47.29mm, implying 2-3 layers missing inbetween. The finishing top layer is smooth meaning the nozzle was close to the surface, so its not underextrusion occurring, but the Z axis not reaching its target height. It'll be hard to reproduce the error and test it easily, since its not showing up till its 3 hours into the print and doesn't occur on shorter parts. I have a few more tall prints to do, so I may see it again. I already overwrote the faulty file, since CraftWare misprints often and/or I make small improvements voiding the older stl/gcode. If the problem ceases, it most likely was tied to CraftWare's problem with >1 Perimeter Wall.


Responding to Questions/Comments:
Dedicated PC: The skipped layers on the first print occurred after I went AFK, closing most things like youtube videos/streams. For the first 6 hours of that print, I was even playing games that peaked one CPU core. I made sure to disable turning off the hard drive or any other Windows sleep other than turning off the monitors, which I've had as a setting for a long time. If the problem returns, I'll try printing with the laptop, although its plagued by Windows 10 breaking everything. Might pull out an old PC and give it one of the potentially busted hard drives to work off of.

Fast Printing: Unless Repetier Host has some built in detection to know that the printer isn't keeping up, there's no reason entire layers got skipped. If the machine's Z motors were skipping, there'd be over/under extrusion, and/or half completed layers due to the Z motors moving constantly with Z Lift enabled. If Repetier Host is expecting a "Completed" response when the printer has finished a layer, it is possible Repetier is missing that "Completed" and might skip the layer to prevent the machine from wrecking. Else, the machine seems to be skipping all Gcode for entire layers, indicating a problem with Repetier or CraftWare, or that the printer hangs after receiving a Z command, and waits till Repetier sends it another Z command to unfreeze it (except it looks like 4 layers back to back failed on the first print).

USB Cable Length/Position: The printer hasn't been relocated, and its cable is 6-10ft. The only changes to the cable since reinstalling Windows was possibly the USB port on the PC. The only other signals near the cable would be monitor cables and the other USB cables and printer cables where the cable plugs in.

Edited 1 time(s). Last edit at 03/09/2018 01:15AM by Gunslinger.
Re: Z/Layer Skipping & Gcode Repeating; High Speed Printing
March 13, 2018 04:41AM
Update/Solved (hopefully):
Skipped Layers: The skipped layers weren't being skipped, but were layers about 0.05mm to 0.1mm. They appeared no differently than how layers do around thin walls and holes, but the defects were around the entire object, just not near/inbetween those thin walls or holes. My last three prints were at the appropriate height, if not slightly too tall (.15mm to .3mm taller). Assuming what I did solved it, the problem was either plastic dust settling on the threaded rod was causing the Z motor to skip, and/or decreasing the Z speed and acceleration 30% (it was the same RPM/acceleration as the X/Y motors, now less than). My guess is it was primarily the plastic residue, since the skips/slips were occurring at a height of 30 +- 10mm, and were not impacted by Z lift. Since my Z lift was 0.3mm, nearly identical to the layer height, the dust particles were probably being plowed around every layer height change, but posed no issue to the Z skip since the particles were already moved. I was expecting more severe over-extrusion issues and/or the printer to stop printing, due to the extruder teeth tearing up the filament. For the time being, I'm assuming it as solved with more occasional cleaning and relubricating of the threaded rod.
Sorry, only registered users may post in this forum.

Click here to login