Print accuracy issue - circle.

Hi, I'm trying to print a ring - it's ~145mm across, with a 7mm width between the outer and inner circles. However the actual print results in a mean width of 7.37mm, a minium width of 6.76 and a maximum width of 7.88mm. A 1.12mm variance. Essentially not fit for purpose. Here's a ring on a peice of paper with measurements taken with digital calipers at regular intervals.
[imgur.com]

It's consistiently this bad, I've printed 7-10 of these, they all practically identical, with the exact same deviations from the target 7mm width in the exact same places.

Initially I was hoping it was just a slicer pathing issue, but no amount of tinkering makes a difference, I tried moving the print around to different locations on the bed, didn't really have any effect.

I tried re-greasing my rods in the hopes of smoothing things out, but again - no difference. I have no idea what's going wrong here, and was hoping someone here might have a few thoughts.

Here are some close ups of an aborted print with just the perimeter, it gives a much better look at the problem. I think some of it could be down to bed leveling, but my current bed is a lost cause, however, notice in image 1 below. That almost perfectly flat line. That's not exactly ideal for a curve...
[imgur.com]
[imgur.com]
[imgur.com]
[imgur.com]
[imgur.com]


Finally, here's my set up.

Machine Information:
i3 type cartestian machine,
e3d titan aero,
duet wifi
12v.

Slicer Information:
0.2mm layer height,
205 C hotend temp,
60 C bed temp,
Print speed 60mm/s
Travel speed 120mm/s
1.75mm filament, no flow modifiers
(tried in Cura and Slic3r)

- Are the rods/x/y axes square to the bed and to each other? Try getting it square before relying on auto bed leveling
- No skipping steps? Are you using stealth chop? try turning it off
- Did you setup and calibrate motor current? Maybe the micro stepping motion is choppy/jumpy. Tuning motor current will help a great deal.
- Could you upload a gcode sample? This would rule out gcode errors
- Are the circles based on actual stl data or did you use a brim or skirt? Brim and skirt calculation isn't very accurate.
- Did you measure backlash/compliance of the x and y axes? Are the belts tight? Are the pulleys and idlers locked in place? Any kind of movement, resonance or play will translate into defects in the print.
- Do the linear guides bind? Do they move freely and smoothly when the belt is off the stepper motors? Any impediment or resistance will stretch and bend the belt/pulleys/idlers/frame etc. and result in backlash/compliance.

---

--
Kind regards
Imqqmi

NFAN CoreXY printer:
[reprap.org]

The printer is as square as I can to verify with the tools I have to measure. It's a p3steel frame, 3mm thick steel all around, it's delightfully rigid.

I don't think the Duet Wifi supports stealthchop. But I'm not sure if it's skipping steps, it makes reasonable looks prints when you aren't measuring the dimensional tolerances.

I did calibrate all the steppers when I set up the machine, but only for steps/mm the currents I have set are fairly arbitary if I'm honest.

The ring was made in Fusion 360, exported with a high poly count. I've attached the gcode, from cura - should be reduced to a single layer. I've attached to my original post.

These last two points.

I'm not fully sure what to look for or test for with regards to backlash, could you eleborate on this point. Though my belts are as tight as I can get them, pulleys and gears are secure.

I can manually move X and Y with relative ease, they seem to put up a consistient resistance along the entire path of their travel.

Check XY squareness by printing a square or rectangle and measure the diagonals. If they are identical, the axes are square.
What is the size of the bed? Can you post photos of the machine showing the belt connections to the bed and the extruder carriage?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I'll have to do a test print after work for the cube. But here are my belts [imgur.com]

The belt paths look OK.

Have you ever calibrated steps/mm in the X and Y axes?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Looking closer at the pictures it would seem that the width of the perimeters vary. This could be due to an uneven bed. Is it heated? It could be warping by heating it up due to thermal expansion.

If you have a dial indicator you can set it up against the x or y axis, then move the axis toward the dial indicator like 2mm, then note the distance indicated, then move it back 2mm. subtract the difference and you've got the backlash.

---

--
Kind regards
Imqqmi

NFAN CoreXY printer:
[reprap.org]

Steps are calibrated. I could post my Duet's entire config.gcode file if helpful?

Currently
M92 X80 Y80 Z400 E418.5

(200[steps per rev] * 16[microstep])/ (2[pitch] * 20[teeth])

@imqqmi - yes it's a heated bed, yes while the bed is level (measured at the corners with a bl touch to within 0.05mm) , the bed itself isn't flat, I have a replacement on the way - but I don't think that fully accounts for the issue, take a look at [imgur.com] again, that's clearly a flat line where there should be a curve. That happens in that part of the print regardless of where I move the print on the bed, I've tried offcentring it to give it a bias towards one of the corners. Same result.

- And I don't have a dial indicactor, I'll look into getting one maybe..

edit: the travesty that is my heatbed; [i.imgur.com]

Edited 1 time(s). Last edit at 10/01/2018 03:32PM by Andywm.

So if I understand correctly the nozzle itself moves those parts in a straight line? Does this effect propagate to the center? Ie if you create multiple circles within each other the same flatness effect occurs? Also does this happen if you cut the bottom part and move it down with multiple copies across the y axis?

Could it be due to some kind of translation errors? Do you use auto bed leveling? Have you tried turning it off? It's probably hard to print on such a warped bed but maybe you can just print the top 1/3 of the circle?

How do the stepper motors sound?

Did you measure the thicknesses of the circle extrusions, by how much do these vary?

---

--
Kind regards
Imqqmi

NFAN CoreXY printer:
[reprap.org]

I noticed a package at my door as I came in, new bed.

Currently installing it, but yes. I do use auto leveling, I have tried turning it off. With respect to translating the circle on the bed. I moved it off centre to the lower left corner, the "flat edge" was in the same 3 o'clock position on the circle. I didn't measure around the circle with calipers like I did with the others though, hmm.

Multiple concentric circles was going to be my next test. I'll finish installing and calibrating this bed to see if I can rule out the mountain range I've been printing on up until now. Then I'll try that.

edit:

also the steppers sound healthy. They're practically in-audiable with the TMC drivers.

Edited 1 time(s). Last edit at 10/01/2018 04:14PM by Andywm.

New bed is in, took a little longer than I'd hoped to properly level. The bed is not the issue.

I'll do the concentric test tomorrow, but I did notice something while printing a first layer. That flat line is on both sides of the circle, it's slightly less pronounced on the left, I guess it was just hidden by the extrusion width before.

Have you looked at the gcode in a viewer that renders it? How about the stl file? Are the flats part of the file?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Pretty much the first thing I checked, no - the model is perfectly circular, and the gcode output is a smooth curve.

Going to do a few prints on it now, concentric circles, xyz cube to see if I can get any more information.

Edited 2 time(s). Last edit at 10/02/2018 03:30PM by Andywm.

If the flatness is happening at equal and opposite parts of the circle it could be an issue of flex in your belting system. I ran into a similar issue with custom designed pulley mounts in my printer, they didn't have enough rigidity so when the belts would change direction I would get deformities in my prints that most notably presented themselves when printing circles. I'm not sure if that is the issue with your printer but it is something worth considering/investigating. Also, tweaking belt tension might not hurt. I also was able to print calibration cubes without any measurable issues when I had that problem but I think that was due to the smaller scale of the cube..

Edited 1 time(s). Last edit at 10/03/2018 11:10AM by obelisk79.

@obelisk79 - pretty sure that's the problem. I fashioned a tensioner for the X axis out of 3 cable ties and the tolerance dropped to +-0.15mm, witha mean width of 7.05mm.

Thanks.

My woes are not quite over though, I've somehow introduced what looks like a severe a z banding issue into my prints in the process of fixing this.

More pictures might help. A full-view shot of the printer, and maybe a couple of more direct pictures of your Z-guides, lead screws (and how they are mounted on the ends).

Edited 1 time(s). Last edit at 10/03/2018 08:12PM by obelisk79.

Shots of the printer, each z axis as a whole, then a few close ups on the lead screws. Plus a shot of a print held against one of the screws to demonstrated the periodicity.

[imgur.com]


You'll have to ignore some of the oddities, the random bolt that's taped to one of the screws, and the cardboard guides at the top.. These were added in my attempt at troubleshooting the problem.

I'll investigate further after I get back from work, my current idea is that trapizodal nut on the left looks uncomfortably loose compared to the one on the right.

edit: that's not it, I replaced my lead screws and the nuts with some spares I had lying around, and recalibrated the entire Z and X axes. The problem persists.

Edited 1 time(s). Last edit at 10/04/2018 07:36PM by Andywm.

Completely remove the constraint at the top of your leadscrews just let them wiggle around how they want and let your z-guides handle the rest.

"You'll have to ignore some of the oddities, the random bolt that's taped to one of the screws, and the cardboard guides at the top.. These were added in my attempt at troubleshooting the problem."

Those were added after the problem appeared. They are gone now anyway. I've also tried completely rebuilding Z with completely different lead screws and completely different trapazodial nuts. No effect.

Not really sure what to try next.

does everything move completely smooth without the leadscrews installed, is there any abnormal play in your zbearings? without the top ends of the screws constrained I don't understand why you would still be having z-banding

Edited 1 time(s). Last edit at 10/05/2018 06:06PM by obelisk79.

Just took the lead screws out. Motion is exceptionally smooth up the guide rods.


Something has happened in the process of me installing the new bed to cause this, initially I thought maybe I tugged a wire on the stepper connectors, or have somehow damaged a stepper - but it's impressively consistient with the banding. That and two sides of Z never fall out of sync, the X axis is perfectly level...

I wonder if the extra tension I added is some how causing this. The printed parts for my x motor and x idler are made of PLA, maybe it has some degree of give.. Idk, I'm just wildly speculating now.

Edited 1 time(s). Last edit at 10/05/2018 08:29PM by Andywm.

Ah, bloody hell. I've figured it out.

Issue 1: the flats on my circle; belt tension!
Issue 2: the z banding;

That new bed I installed... I didn't PID tune it, the temp was bouncing +-5 C with about a minute period, I guess the expansion and contraction was just enough to mess up the print in a periodic way. I didn't even consider this possibility, I had more or less dismissed the bed as a potential cause.. But it's all that was left, so in a last ditch attempt to save my sanity - I disabled the heater. And perfect layers. A sight for sore eyes.

I've had this printer fully to bits twice now, I've replaced the lead screws with spares, I've done the most excruciatingly precise calibrations I've ever carried out, I've messed around with dozens of slicers settings, gone through about 10 iterations of my config.g file... All because I was too lazy to run a bloody PID tune, *sigh*. That's a week of my life I won't be getting back...

Thanks all that offered help.

Well, great lessons learned! Glad to hear you've figured it out.

---

--
Kind regards
Imqqmi

NFAN CoreXY printer:
[reprap.org]