UNSOLVED Duet 3 3.2 Beta2 + SBC jerking motion
Garfield last edited by dc42
There comes a time with printing that you can't identify whether a problem is
A : You
B : The machine
C : The Slicer / Gcode
D : The Filament in use
E : The day of the week
You get the picture here I hope.
I built a CoreXY based on the highest possible grade of everything, including linear rails, I drive it with a Duet 3. Enough background .... needless to say I have already eliminated the above list.
Recently I've encountered some real wierd issues, all since rolling into 3.2 Beta. I had a strange low frequency ripple 'wavy line' in vertical surfaces that was 10 to 12mm peak to peak, it was totally vertical and not skewed in any way although it was very very slight and you needed a critical eye to spot it (the camera saw it first - I didn't).
Every now and then the print head jerks violently almost like it had been snagged on something, I watched it, it wasn't, no belts were snagged, nothing was catching on the print, no reason for such a violent jolt - and thats what it is a jolt like it releases pressure or jumps back and forth really quickly, it is a very audible 'thud'. It doesn't happen in consistent places and sometimes is more violent than others and is so fast that I sure as heck can't spot exactly what is going on.
I've spent a lot of days trying to figure this and ultimately rolled back firmware from 3.2 beta to 3.1.1 - what do ya now - the problems went away - all of them. There is no mechanical or other issue here - it is 100% been introduced in the code some place. But how to track down what it is ?
fcwilt last edited by
Well beta code is expected to behave oddly at times - otherwise it wouldn't be beta would it.
I waited on the move from v2 to v3 until dc42, when asked, felt comfortable saying that 3.1.1 was good.
Garfield last edited by Garfield
A little obvious - I kind of know this - I run beta on purpose in the hope that issues found may assist the team resolve. I'm not asking for a fix hence the FYI - but I am making those that can fix aware so it doesn't end up in a release version.
This is an FYI. Partly in the hope that somebody with a better understanding of the code than me may have ideas that can be explored to detect whatever 'this' is or may be.
Moved to the beta firmware forum.
Would you be able to share your config.g?
Are you using the Duet 3 in standalone or SBC connected mode?
Can you provide an example of the ripple you mention?
Do you have an example gcode that was able to provoke the violent jolt?
Reporting a problem is helpful, but the more details the better.
SAtech last edited by
I have experienced this exact behavior. The print head jerks, almost as if it's skipping a step multiple times mid-print. I am running a duet 2 with SBC on a corexy machine with rrf 3.2-b2 as well.
@Garfield, are you running your Duet 3 with or without attached SBC?
Garfield last edited by Garfield
Duet3, attached SBC.
All updates are done via the console of the SBC.
I've attached the config.g but I don't think it adds much - this problem does not exist on 3.1.1 with the exact same file. I verified this behaviour by printing the exact same test piece using the exact same gcode. 3.1.1 does not demonstrate the behaviour.
I didn't keep any pictures of the ripple - I will have to reproduce - I didn't think this was anything other than my problem - reverting the firmware was a last resort, I wasn't expecting the problem to go away.
The test model gcode is too large to upload uncompressed. It is essentially a 5mm thick box, 75mm sides with a 75mm radius on one corner. Be aware though that this prints fine on 3.1.1.
Be aware - this is a gcode file that has been zipped - the extension should be .zip but I couldn't upload that.
I'm going to set up and try to run without the SBC (never done this so result could take a while)
There are some known issues with Beta2 used with SBC. I suggest you revert to Beta1 until Beta3 is available.
Appreciated Dave, I was about to report but have refrained since doing so would have been premature - and the kiss of death.
So far running direct with no SBC and 3.2 b 2 I've had a print running 30 mins or so with zero issues so far.
If you need me to do any testing let me know. I've created a completely new SD card to run local and can easily pop the Pi back on.
Said it was the kiss of death ....
So no jolts or odd behaviour - except now the print failed 1hr 14 into the print with Error 1 - Cannot read file.
File reads fine via the Duet GUI, can be downloaded, edited, viewed to the end etc etc
File reads fine with the SD card on the PC - scanned with notepad++ - no defects in file -
I will try again and see if it chokes at the same point.
Was just a glitch - file has printed fine now for hours ... all print defects are gone, the jolts are gone with the Pi removed from the equation.
tristanryerparke last edited by tristanryerparke
I had this issue with the Pi4 plus Duet3 on the 3.2 beta as well, went back to the stable to avoid it. Now after upgrading to 3.2rc1 I am still having this "jolting" issue.
I need to run the sbc as ethernet connection is not an option in my workshop.
@tristanryerparke Can you give some more details about your setup? config.g, etc.
Gantry style five axis machine. Nema23 on X,Y. Nema17 on Z,A,B. Duet 3 mainboard and one toolboard controlling the b axis and z probe.
Dual motors for Y axis, auto squaring with U axis.
;CNC Mode M453 ;Brake Pins M950 P2 C"out8" M950 P1 C"out7" ;Drivers M569 P0.0 S0 I1 M569 P0.1 S0 I1 M569 P0.2 S1 I1 M569 P0.3 S0 I1 M569 P0.4 S0 I1 M569 P121.0 S1 I1 ;Driver Mapping M584 X0.3 Y0.1:0.2 Z0.0 A0.4 B121.0 P5 M584 U0.2 P5 ;Current M906 X2000 Y3750 Z2000 U3750 A1750 B1500 I100 ;Microstepping M350 X16 Y16 Z16 U16 A16 I1 ;Steps Per MM M92 X80 Y80 Z80 U80 A46.1062140618 B26.6666 ;Instantaneous Speed Change M566 X500 Y500 Z500 U500 A500 ;Max Speeds M203 X13000 Y13000 Z13000 U6000 A40000 B40000 ;Max Acceleration M201 X750 Y750 Z500 U750 A2000 B2000 ;Disable Idle Current Reduction M84 S0 M917 Y100 ;Probe M558 P9 C"!^121.io0.in" H200 F400 T2000 I1 ;Limits M208 X0 Y-1360 Z-145 A-72000 B-101.4 S1 M208 X1242 Y0 Z0 A72000 B230 S0 ;Servos M950 S3 C"io4.out" M950 S4 C"io5.out" ;Endstops M574 U1 S1 P"!io6.in" M574 Y1 S1 P"!io1.in" M574 X1 S1 P"io8.in" M574 Z1 S1 P"io7.in" M574 A1 S1 P"io0.in" ;Head Endstop M574 B1 S1 P"121.io2.in" ;Tool 0 M563 P0 S"Probe" ;Tool 1 M563 P1 S"1/4 Flat Brush" ;Tool 2 M563 P2 S"5/8 Flat Brush" ;Allow Movement M564 S0 H0
RC2 is now out, can you test with that?
@Phaedrux Just tried with RC2, no luck and jerks happen in the same places on same file.
@tristanryerparke I'm just noticing in your config some rather large spacing between some values. Is that intentional? Can you send M98 P"config.g" to check for errors? Can you send some of those commands like M208 or M201 etc by themselves to see if all of the values for all axis are being set as expected?
@Phaedrux yes the spacing was large just for readability/comparing values for axes.
I have removed the spacing so config.g now looks like this:
;CNC Mode M453 ;Name M550 P"tristan-painting-machine" ;Brake Pins M950 P2 C"out8" M950 P1 C"out7" ;Drivers M569 P0.0 S0 I1 M569 P0.1 S0 I1 M569 P0.2 S1 I1 M569 P0.3 S0 I1 M569 P0.4 S0 I1 M569 P121.0 S1 I1 ;Driver Mapping M584 X0.3 Y0.1:0.2 Z0.0 A0.4 B121.0 P5 M584 U0.2 P5 ;Current M906 X2000 Y3750 Z2000 U3750 A1750 B1500 I100 ;Microstepping M350 X16 Y16 Z16 U16 A16 I1 ;Steps Per MM M92 X80 Y80 Z80 U80 A46.1062140618 B26.6666 ;Instantaneous Speed Change M566 X500 Y500 Z500 U500 A500 ;Max Speeds M203 X13000 Y13000 Z13000 U6000 A40000 B40000 ;Max Acceleration M201 X750 Y750 Z500 U750 A2000 B2000 ;Disable Idle Current Reduction M84 S0 M917 Y100 ;Probe M558 P9 C"!^121.io0.in" H200 F400 T2000 I1 ;Limits M208 X0 Y-1360 Z-145 A-72000 B-101.4 S1 M208 X1242 Y0 Z0 A72000 B230 S0 ;Servos M950 S3 C"io4.out" M950 S4 C"io5.out" ;Endstops M574 U1 S1 P"!io6.in" M574 Y1 S1 P"!io1.in" M574 X1 S1 P"!io8.in" M574 Z1 S1 P"io7.in" M574 A1 S1 P"io0.in" ;Head Endstop M574 B1 S1 P"121.io2.in" ;Tool 0 M563 P0 S"Probe" ;Tool 1 M563 P1 S"1/4 Flat Brush" ;Tool 2 M563 P2 S"5/8 Flat Brush" ;Allow Movement M564 S0 H0
Running 3.2 stable now and the issue is still happening. M98 P"config.g" runs fine with no errors.
Here is video of one of the clunks:
As you can see the system is not losing steps, it just stops for a fraction of a second without accelerating/decelerating.
Maybe I should make a new thread for this issue?
Let me know.
@Phaedrux I believe this is the same issue as the what the OP had.
I did read the notes for 3.2, A and B axes are rotational and should be defaulting to that within the first M584 command, correct?
I will try now explicitly defining them as rotational.
Also something interesting:
I just started messing around with the M595 command to see if I could fix the clunks.
So far I have tried S=5, S=10, S=50, S=100 and S=1000.
All of these made the clunking way worse!