Extruder motor stops during print…

ofloveandhate

another failure, on my wifi, 1.20. we can safely say:

either neither of my duet boards are defective, or both are
if this is a firmware problem, it exists in 1.19.2 and 1.20

i additionally believe that

we can rule out my wiring as a problem. otherwise, when i switched E0 to E1, the extruder would not have moved without a restart. however, if my wiring is causing this fault, it should be firmware detectable… ideally

one problem i have observed is that the firmware is unable to detect if the extruder motor comes disconnected. you can unplug, and replug, no errors. should this not be observable?

ofloveandhate

is this the same problem as i am experiencing?

https://community.ultimaker.com/topic/4733-print-stops-or-fails-halfway-through/

ofloveandhate

it seemed so at first, but no, my fans work fine. that post was about heat creep, not the motor stopping.

dc42

Does extrusion stop at the same point in the print each time?

The stepper drivers on the Duet can in theory detect when a motor wire is disconnected, but in practice we found it gave false positives, so I disabled that facility. If I can find a way to filter out the false positives, I will enable it again.

ofloveandhate

another thought i had – maybe i should put a fan on the duet? is the board overheating?

https://www.duet3d.com/wiki/Connecting_and_configuring_fans

ofloveandhate

no, it is not the same point every time.

ofloveandhate

here's a snippet of what my gcode looks like:

G1 X-9.773 Y20.202 E0.0240
G1 X-8.049 Y19.083 E0.0845
G1 X-7.228 Y18.592 E0.0393
G1 X-5.829 Y17.806 E0.0660
G1 X-2.005 Y15.805 E0.1774
G1 X-1.749 Y15.664 E0.0120
G1 X-0.741 Y15.031 E0.0489
G1 X-0.182 Y14.575 E0.0297
G1 X0.225 Y14.087 E0.0261
G1 X0.746 Y12.742 E0.0593
G1 X0.847 Y12.526 E0.0098
G1 X0.864 Y12.365 E0.0066
G1 X0.745 Y10.941 E0.0588
G1 X0.342 Y9.133 E0.0761
G1 X-0.212 Y7.194 E0.0829
G1 X-1.216 Y3.878 E0.1424
G1 X-1.686 Y2.198 E0.0717
G1 X-1.997 Y1.061 E0.0484
G1 X-2.338 Y-0.306 E0.0579
G1 X-2.657 Y-1.687 E0.0582
G1 X-3.138 Y-4.143 E0.1029
G1 X-3.300 Y-5.092 E0.0396
G1 X-3.519 Y-6.570 E0.0614
G1 X-3.767 Y-8.725 E0.0892
G1 X-3.921 Y-10.510 E0.0736
G1 X-4.038 Y-12.900 E0.0984
G1 X-4.060 Y-14.016 E0.0459
G1 X-4.047 Y-16.295 E0.0937
G1 X-3.986 Y-17.935 E0.0674
G1 X-3.894 Y-19.398 E0.0602
G1 X-3.771 Y-20.858 E0.0602

so it is a sequence of short extrusion moves the whole file. no long linear extrudes, just this.

ofloveandhate

here are my four attempts at 'seepferdchen'

ofloveandhate

another thought: should i prefer absolute or relative extrusions? i have only been using relative through this whole process.

ofloveandhate

i am starting to think there is a bug in the reprap firmware, in the part that processes relative extrusion distances. i printed the same model, same supports and all settings, last night. only change was that in simplify3d, i switched to absolute extrusion.

ok, there was another difference that i should have not let happen, and that was that i printed this one without the https://www.thingiverse.com/thing:2567240 bottom corner plates.

i am re-printing the same file now. if it fails again, then we know that the relative / absolute distances had nothing to do with my problem. if it succeeds, we have stronger evidence that it is a firmware bug.

ofloveandhate

i am now convinced that there is a bug in the firmware dealing with relative extruder distances

another success with absolute extruder distances, print time was 36h 21m

video:
https://youtu.be/7BCaq2k_oJk

pic:

a little problem at the top. caused due to top layers being set to 0 for that process in simplify3d. (6 processes used total, to reduce the amount of plastic used by varying the infill %)

dc42

Most Duet users - including me - only ever use relative extrusion distances.

Did you turn off "Allow zeroing of extrusion distance" in S3D? You should do that when using relative extrusion, although AFAIK it doesn't matter if you leave it enabled.

ofloveandhate

zeroing was indeed allowed in s3d, both in my relative and absolute prints.

brad at ultibots indicated that he thinks people mostly use absolute, where you think it is relative…

i have > 70 hours of absolute distance printing with no errors now, so am growing in confidence that i was experiencing something problematic with relative. i think some stress-test kinds of gcode files are merited. i was thinking that cylindrical circles or ellipses with ever-smaller faces might be useful, and mix in the zeroing.

are there known smallest distances that the extruder should be asked to perform? i read the code in the reprap firmware, and nothing immediately jumped out at me as risking underflow or rounding error. i was however concerned in GCodes.cpp at lines 2044 or so,

const float moveArg = eMovement[0] * distanceScale;
float requestedExtrusionAmount;
if (gb.MachineState().drivesRelative)
{
  requestedExtrusionAmount = moveArg;
}
else
{
  requestedExtrusionAmount = moveArg - virtualExtruderPosition;
  virtualExtruderPosition = moveArg;
}

that in one case, the value of virtualExtruderPosition is adjusted, and not in the other. is it assumed to be always 0 if relative? where is it adjusted?

i just want to know exactly how to manifest this problem, as i am not content with absolute distances as a workaround.

garyd9

Taking a WAG based solely on the code you posted:

If doing relative extrusion, the "virtualExtruderPosition" variable is meaningless (simply because it's not needed: It's a variable only used to calculate requestedExtrusionAmount if and only if absolute extrusions.)

If relative, then the value of "E" is the amount that is extruded in this step.

If absolute, then the value of "E" (moveArg) is the amount that is extruded in this step and all the previous steps combined. So, in order to know how much THIS step extrudes (requestedExtrusionAmount), you have to subtract "all the previous steps combined" (virtualExtruderPosition) from moveArg. At that point, "all the previous steps combined" (virtualExtruderPosition) needs to be updated to reflect the added amount THIS step extrudes (requestedExtrusionAmount.)

That could be written any of the following ways, and all would be accurate:

virtualExtruderPosition = virtualExtruderPosition + requestedExtrusionAmount;

virtualExtruderPosition += requestedExtrusionAmount;

virtualExtruderPosition = moveArg; // because moveArg == (virtualExtruderPosition + requestedExtrusionAmount)

ofloveandhate

gotcha.

do you have references to any other places people have had problems with relative extrusion and zeroing on?

ofloveandhate

the drama continues.

i printed a surface called "tobel", and it failed, even with absolute extruder distances. we can rule out relative as the culprit.

added a heatbreak out of lexan, with a fan blowing directly across the duet.

the print still failed.

here is what it is supposed to look like:

i think we can rule out heat as the source of the problem. because the two prints failed at such a similar place, i am hopeful that i can develop a short gcode file that fails every time i run it. then, hopefully, someone else can also try it, and fail.

ofloveandhate

my journey continues. i have re-wired the extruder motor, used a heat shield and fan to cool the duet board, and played with various slicer settings. nothing has worked – i continue to experience failures.

is there an alternate firmware i can load on the board?

whosrdaddy

Think about it, if this was a "firmware bug", this forum would be swamped with threads similar as yours, but it isn't ;).
Having read your thread @ultibots, I think you are having ordinary nozzle jams.
You use 185 deg, which seems to me a bit on the cold side, I use 205 for my PLA prints and never had any issue.

Can you share the STL for above model? I Can do a test print if you want…

Cheers,
Kris

ofloveandhate

the absence of other bug reports is not evidence of no problem. to the contrary, my persistent problem is evidence of a problem. its not conclusive, i agree, but this problem has happened to me at least 19 times, and has persisted despite nearly 3 months of publicly documented problem-solving efforts, including swapping motors, re-wiring, heat shielding and actively cooling the duet, swapping duet boards, upgrading firmware, and twiddling nearly all possible slicer options (which shouldn't matter anyway because no gcode should do this).

what is left? power supply / fluctuations, too low a temp, a firmware bug, some machine setting like acceleration or max speeds (again pointing to a firmware bug), or ??? i appreciate you helping me explore the remaining options, because i am about out of ideas.

i do agree that few people with the problem makes it harder to tell whether this is indeed a firmware bug. but i point you to the famous case of the intel floating point arithmetic bug: https://en.wikipedia.org/wiki/Pentium_FDIV_bug . "1 in 9 billion floating point divides with random parameters would produce inaccurate results." to put it succinctly,

absence of evidence is not evidence of absence

furthermore, another d300vs user commented in my thread on the ultibots forum that they experienced a very similar problem. so, i am not alone in experiencing this problem. was their problem mine? we'll likely never know…

sure, i'll try a higher temp. i must try everything, because 1) right now, i can only print trinkets on my machine, 2) i am a scientist and mathematician so i have to know what the solution is, and 3) i genuinely hope that it is not a firmware bug, but instead a simple matter of my temperature being too low. we'll see...

here is a link to a folder of gcode files, all of which have failed for me at least once. they are set up for 1.75mm filament with a 0.4mm diameter tip on a delta bot with the origin at the center. if i can get even one other person to run my gcode, it'll do a long way to helping rule out a firmware bug. repeated success on anothes duet?
my problem. another person gets a fail? clearly not my fault, points to firmware. another strong piece of evidence would be what happens if i use another firmware, so looking to install something else and give it a try.

http://danibrake.org/gcode/failing/

the 40% file takes like 5 hours, and i have had it fail on me 3/13 times. the 60% takes like 14 hours, and has failed 1/1 times. the 100% takes even longer, and has failed 4/4 attempts. if you want a sure fail, try the 100. the model is also posted to the same location.

i am still on firmware 1.20, per forum recommendation. here is my config file:

; Configuration file for UltiBots D300 V-Slot Delta 3D Printer
; version 1.0 for RepRapFirmware version 1.18.1

; Communication and general
M111 S0                             	; Debug off
M550 PD300VS				; Machine name and Netbios name (can be anything you like)
M551 XXXXX				; Machine password (used for FTP)
; *** If you have more than one Duet on your network, they must all have different MAC addresses, so change the last digits
M540 P0xBE:0xEF:0xDE:0xAD:0xFE:0xFF 	; MAC Address
; Wifi Networking
M552 S1					; Enable Wifi by default

M555 P2                           	; Set output to look like Marlin

; *** Uncomment if you have the optional PanelDue display
;M575 P1 B57600 S1			; Communication parameters for the PanelDue

G21                                 	; Work in millimeters 
G90                                	; Send absolute positional coordinates...
M83                                 	; ...but relative extruder moves

; Axis and motor configuration
M569 P0 S0				; Drive 0 goes forwards (X tower)
M569 P1 S0				; Drive 1 goes forwards (Y tower)
M569 P2 S0				; Drive 2 goes forwards (Z tower)
M569 P3 S1				; Drive 3 goes forwards (extruder 1)

M574 X2 Y2 Z2 S1			; Set endstop configuration (all endstops at high end, active high)

; *** The homed height is deliberately set too high in the following - these will be adjusted with delta auto-calibration
M665 R216.411 L379.00 B140 H434.314 X0.101 Y0.295 Z0.0	; Set delta radius, diagonal rod length, printable radius and homed height
;H461.071
M666 X1.79 Y0.20 Z-1.99			; Endstop offset adjustments, these will be adjusted with delta auto-calibration

M350 X64 Y64 Z64 E16 I1			; Set microstepping to 32 for X, Y and Z and 16 for extruder stepper with interpolation
M92 X800 Y800 Z800			; Set axis steps/mm
M906 X1000 Y1000 Z1000 E400		; Set motor currents (mA)
M201 X1000 Y1000 Z1000 E1000		; Accelerations (mm/s^2)
M203 X20000 Y20000 Z8000 E20000	; Maximum speeds (mm/min)
M566 X400 Y400 Z400 E400		; Maximum instant speed changes mm/minute

; Fans
M106 P1 T50 S255 H1			; Set hotend heatsink FAN1 thermostatic control at 50°C

; Thermistors
M305 P0 T100000 B3950 R4700 L54 H-97	; Kapton bed heater thermistor
M305 P1 R4700 T100000 B4725 C7.06e-8	; E3D V6 Semitec GT-104 thermistor cartridge

; Heater configuration
M307 H0 B1							; Heater 0 (bed) use bang-bang control		
M307 H1 A512.9 C267.0 D9.0 B0		; Heater 1 (hot end) use PID

; Tool definitions
M563 P0 D0 H1                       	; Define tool 0, the extruder
G10 P0 S0 R0                        	; Set tool 0 operating and standby temperatures
M92 E700	                       	; Set extruder steps per mm

; Z probe and compensation definition
; Change "H25" to "H3" AFTER commissioning your printer
M558 P4 X0 Y0 Z0 H5 I1 			; FSRs with JohnSL board Z probe behaves as a switch and is not used for homing any axes
G31 X0 Y0 Z-0.2 P500			; MUST READ: http://www.sublimelayers.com/2017/05/fdffsd.html

T0					; Select tool 0, the hot end
M501					; Load config-override.g

thanks for helping me, i genuinely appreciate your time

whosrdaddy

May I ask why the steps for the extruder (M92) are missing?
Also the M92 I1 parameter is a bit pointless as it only works for 16x microstepping…