Problem printing circles

Phaedrux

0.4 width no thin walls:

G1 Z5.000 F10800.000
G10 ; retract
G1 X111.743 Y156.008
G11 ; unretract
M204 S800
G1 F2400
G1 X111.537 Y155.539 E0.02393
G1 X110.750 Y153.516 E0.10139
G1 X110.053 Y151.460 E0.10139
G1 X109.445 Y149.376 E0.10139
G1 X108.929 Y147.268 E0.10139
G1 X108.506 Y145.139 E0.10139
G1 X108.176 Y142.994 E0.10139
G1 X107.939 Y140.836 E0.10139
G1 X107.797 Y138.670 E0.10139
G1 X107.750 Y136.500 E0.10139
G1 X107.797 Y134.330 E0.10139
G1 X107.939 Y132.164 E0.10139
G1 X108.176 Y130.006 E0.10139
G1 X108.506 Y127.861 E0.10139
G1 X108.929 Y125.732 E0.10139
G1 X109.445 Y123.624 E0.10139
G1 X110.053 Y121.540 E0.10139
G1 X110.750 Y119.484 E0.10139
G1 X111.537 Y117.461 E0.10139
G1 X112.411 Y115.475 E0.10139
G1 X113.371 Y113.528 E0.10139
G1 X114.415 Y111.625 E0.10139
G1 X115.541 Y109.769 E0.10139
G1 X116.747 Y107.965 E0.10139
G1 X118.031 Y106.214 E0.10139
G1 X119.389 Y104.521 E0.10139
G1 X120.820 Y102.889 E0.10139
G1 X122.321 Y101.321 E0.10139
G1 X123.889 Y99.820 E0.10139
G1 X125.521 Y98.389 E0.10139
G1 X127.214 Y97.031 E0.10139
G1 X128.965 Y95.747 E0.10139
G1 X130.769 Y94.541 E0.10139
G1 X132.625 Y93.415 E0.10139
G1 X134.528 Y92.371 E0.10139
G1 X136.475 Y91.411 E0.10139
G1 X138.461 Y90.537 E0.10139
G1 X140.484 Y89.750 E0.10139
G1 X142.540 Y89.053 E0.10139
G1 X144.624 Y88.445 E0.10139
G1 X146.732 Y87.929 E0.10139
G1 X148.861 Y87.506 E0.10139
G1 X151.006 Y87.176 E0.10139
G1 X153.164 Y86.939 E0.10139
G1 X155.330 Y86.797 E0.10139
G1 X157.500 Y86.750 E0.10139
G1 X159.670 Y86.797 E0.10139
G1 X161.836 Y86.939 E0.10139
G1 X163.994 Y87.176 E0.10139
G1 X166.139 Y87.506 E0.10139
G1 X168.268 Y87.929 E0.10139
G1 X170.376 Y88.445 E0.10139
G1 X172.460 Y89.053 E0.10139
G1 X174.516 Y89.750 E0.10139
G1 X176.539 Y90.537 E0.10139

0.45 thin walls

1 Z5.000 F10800.000
G10 ; retract
G1 X137.971 Y90.726
G11 ; unretract
M204 S800
G1 F2400
G1 X138.447 Y90.516 E0.03302
G1 X138.452 Y90.514 E0.00031
G1 X138.456 Y90.512 E0.00031
G1 X140.471 Y89.729 E0.13714
G1 X140.476 Y89.727 E0.00031
G1 X140.481 Y89.725 E0.00031
G1 X142.528 Y89.030 E0.13710
G1 X142.532 Y89.029 E0.00031
G1 X142.537 Y89.027 E0.00031
G1 X144.613 Y88.422 E0.13713
G1 X144.617 Y88.421 E0.00031
G1 X144.622 Y88.420 E0.00031
G1 X146.722 Y87.906 E0.13713
G1 X146.727 Y87.905 E0.00031
G1 X146.732 Y87.904 E0.00031
G1 X148.852 Y87.482 E0.13713
G1 X148.857 Y87.481 E0.00031
G1 X148.862 Y87.480 E0.00031
G1 X150.998 Y87.152 E0.13713
G1 X151.003 Y87.151 E0.00031
G1 X151.008 Y87.150 E0.00031
G1 X153.157 Y86.915 E0.13712
G1 X153.162 Y86.914 E0.00031
G1 X153.167 Y86.914 E0.00031
G1 X155.324 Y86.773 E0.13713
G1 X155.329 Y86.772 E0.00031
G1 X155.334 Y86.772 E0.00031
G1 X157.077 Y86.734 E0.11058
G1 X157.495 Y86.725 E0.02655
G1 X157.500 Y86.725 E0.00031
G1 X157.505 Y86.725 E0.00031
G1 X159.254 Y86.763 E0.11096
G1 X159.254 Y86.763 F10800.000
G1 F2400
G1 X159.666 Y86.772 E0.02617
G1 X159.671 Y86.772 E0.00031
G1 X159.676 Y86.773 E0.00031
G1 X161.833 Y86.914 E0.13713
G1 X161.838 Y86.914 E0.00031
G1 X161.843 Y86.915 E0.00031
G1 X163.992 Y87.150 E0.13712
G1 X163.997 Y87.151 E0.00031
G1 X164.002 Y87.152 E0.00031
G1 X166.138 Y87.480 E0.13713
G1 X166.143 Y87.481 E0.00031
G1 X166.148 Y87.482 E0.00031
G1 X168.268 Y87.904 E0.13713
G1 X168.273 Y87.905 E0.00031
G1 X168.278 Y87.906 E0.00031
G1 X170.378 Y88.420 E0.13713
G1 X170.383 Y88.421 E0.00031
G1 X170.387 Y88.422 E0.00031
G1 X172.463 Y89.027 E0.13713
G1 X172.468 Y89.029 E0.00031
G1 X172.472 Y89.030 E0.00031
G1 X174.519 Y89.725 E0.13710
G1 X174.524 Y89.727 E0.00031
G1 X174.529 Y89.729 E0.00031
G1 X176.544 Y90.512 E0.13714
G1 X176.548 Y90.514 E0.00031
G1 X176.553 Y90.516 E0.00031

full gcode

I haven't done a test print yet, but if you try setting your extrusion width to 0.4 all around and turning off detect thin walls and print again, and it hasn't fixed it, I'll print your gcode and try to duplicate it.

OBELIKS

The layer change extrusion reset should not matter. It's relative extrusion. I will try without.
As for the walls, I have thin walls disabled and if you look at the top you will se that perimeters are 0.45mm, the same as model. So it only makes 1 line.
And as I said, after setting resolution to 0.1mm judder is mostly gone. Now I only have problems with geometry.

Phaedrux

@obeliks The profile in the gcode you posted has thin walls enabled. And I've shown I can duplicate getting slic3r to generate the odd extrusion amounts by toggling it along with setting the wall thickness.

OBELIKS

Oh, you are right. Sorry about that. I will test tomorrow.
But that should not trigger on normal perimeters as are on the picture a couple of posts above.

Phaedrux

@obeliks I'm not so sure about that. I've seen detect thin walls do some crazy things.

I think the reason that reducing the resolution makes things a bit better is that detect thin walls is basically pulsing the extruder at the start of a segment and then dragging the blob of plastic out to try and get a thinner extrusion width than you've actually specified. By reducing the resolution you now have fewer and longer segments so it has to pulse less.

On a slightly related note, just the other day I noticed that someone has a pull request outstanding for Prusa Slic3r to completely redo how thin walls are handled. It looks promising.

OBELIKS

Can you check both gcode files, and see how many lines you get for the second ;BEFORE LAYER CHANGE
Because for me it doubled with thin walls disabled.

Phaedrux

@phaedrux said in Problem printing circles:

full gcode

https://www.dropbox.com/s/ormosogim0yaeq2/sliced gcode.7z?dl=1

They can be found here.

OBELIKS

Thanks. I'll check tomorrow.

OBELIKS

I will also try the solution from this thread: https://forum.duet3d.com/topic/6078/core-xy-movement-issues/57

OBELIKS

So in additon to the fix above I also erased firmware and uploaded 1.21 Nothing changed.
Turning off thin wall detection in Slic3r did help with the juddering, but not completely.
Cricles still have flatspots on top and bottom, and printer still sounds rough.

So I need to find some time to wire the old board back in and test the steppers.

Phaedrux

@obeliks In addition to turning off thin wall detection did you also change your extrusion width to 0.4?

OBELIKS

Yes, and it is a different model anyway. So that is not really needed.

OBELIKS

OK, we can stop blaming DuetWifi. I took the time (and anger at some relatives) and I switched to the old board. The uneven circles are almost the same.
Left is DuetWifi and right is Anet board, same gcode
alt text

Now that I have the old board on, I will do another test with Cura and S3D. But that will wait for tomorrow.

sigxcpu

That is backlash in Y axis (if this is the printing position). There is also a smaller backlash in X axis, too.

OBELIKS

I hope this is true.
I really need to reduce weight of the bed assembly. I ordered PCB bed without ALU and thinner glass. That should save me around 500g, and if this goes well I will replace the bed frame for ALU also, at the moment I have inox. That would mean at least 200g.

sigxcpu

Unless you have very shitty belts that strech like crazy, there is no reason for the weight of the moving things to generate backlash. I have it too in Y on my CoreXY, but I am sure it is not the axis weight which is pretty thin.
If you suspect the weight is an issue then you can play with motors currents and/or acceleration and see if there are any changes.

timcurtis67

@sigxcpu said in Problem printing circles:

Unless you have very shitty belts that strech like crazy, there is no reason for the weight of the moving things to generate backlash. I have it too in Y on my CoreXY, but I am sure it is not the axis weight which is pretty thin.
If you suspect the weight is an issue then you can play with motors currents and/or acceleration and see if there are any changes.

I agree with this. @deckingman probably has the heaviest print head axis combo of everyone on this forum and he doesn't complain about backlash.

Ian, do you want to comment about this?

OBELIKS

My bed assembly is in the region of 1,5kg

deckingman

@timcurtis67 said in Problem printing circles:

Ian, do you want to comment about this?

Not really ............................

But since you asked...... I have 4kg split between two gantries so roughly 2 kg per XY mechanism. "No name" 6mm GT2 belts. Max X axis speed 50,000, max Y axis speed 35,000, accel 1,000 for both, jerk 600 for both. I have experimented with high print speeds up to 300mm/sec and my default non-print move speed is 350mm/sec. No sign of backlash but everything is pretty rigid. It'll print OK at higher accelerations but with the mass being over a metre above the floor, the printer tends to start to rock about a bit at high speeds.

As an aside, I don't get ringing either - one advantage of high mass

OBELIKS

Hah, I have a i3 style printer. So it is moving a lot.
I will investigate the whole mechanics of X and Y, something is wrong there, that I am certain.

Aaaand, he deleted the post