bad quality on S3D, good quality on Cura

[tomaski]

for last couple of days I am trying to find the cause of an issue I am having when printing with S3D: the prints that are sliced using S3D, come out as if my printer was not calibrated, while exactly the same model sliced with Cura prints just fine.
As an example: a simple 25mm test cube (0.2 layer height, 0.4 width, 3 shells, 0% infill, no top layers, no bottom layers)
sliced with Cura: print quality is not perfect but is good enough to be usable. Dimensionally is spot-on: X+Y+Z are 25mm +/- 0.05 mm and walls are 1.2mm +/- 0.02
sliced with S3D: print quality is tiny bit worse, Dimensionally less accurate and walls.... walls are like a mini bunker 1.4mm thick +/- 0.2 mm

When printing 9 calibration squares, I noticed that the first layer looks as if the nozzle was too far (lines are separated and more round-ish). If I move nozzle closer to bed, so that the first layer lines are nicely squished and bonded, the rest of the print looks more like an accordion than a cube and dimensional accuracy is nonexistent (22mm height, 1,86 mm wall thickness)

Flow on both slicers is set to 95%. Test subjects were printed using the same filament, the same temperatures and similar speeds. What can be the reason that S3D is so much off?

[S3D-Jake]

Check out our guide on correcting issues with dimensional accuracy

• https://www.simplify3d.com/support/prin ... -accuracy/

Also, KeyboardWarrior's tip of the day on steps/mm has a lot of good learning about correcting for these types of issues as well.

• viewtopic.php?f=8&t=1937

[greybeard]

I shouldn’t be surprised about this, given that common-sense is not a part of today’s society but, I was born in the 1940’s...

Neglecting what can be Googled about the size of Human Hair, I measured my own.

Head hair = 0.16mm = 0.006 inch
Beard hair = 0.2mm = 0.008 inch

Do people really think that plastic printing can be accurate enough to talk about dimensions of prints less than the ability of the printer to print when the machine is built to, at best, greater than 0.2mm Accuracy ???

Considering Belts, Gears, Backlash, Hysteresis (mechanical and electrical), Print Process, Speed....etc

And, do you really need the Printed Part to be dialed down to that kind of tolerance??? OMG!

[tomaski]

@greybeard no, of course I wasn't expecting it to be as precise as if CNC-milled (that'd be laughable, to say the least )
No need for a snarky, down-looking comment.

Obviously, 3D printers do have some degree of calibration that can be applied - otherwise how would you print a bolt+nut or a box or some other fitting parts.

My point (apparently misunderstood) was that AFAIK most of the calibrations are slicer agnostic: if I ask printer (calibrated one) to extrude 50mm of filament it pushes 50-ish, not 80 mm. So some "accuracy" is to be expected?

So wanting a wall that is 3 shells thick (with a 0.4 wide layer) , why does Cura deliver a wall that is 1.18-1.22 thick but S3D one is 1.38-1.42?

Since I am relatively new to this, all that I wanted to know is this: Is this normal / expected and I need to re-calibrate the e-steps and flow every time I switch slicer or that shouldn't be happening and indicates something wrong with my S3D settings?

[greybeard]

A worst-case of: Difference = Best-Worst

Difference = 1.42 - 1.22 = 0.2mm

That's the thickness of my Beard Hair. And that's, Best and Worst. What if the values are not at those limits. How about the Average?...

How about, 1.42 - 1.18 = 0.24mm = 0.009 inch. ? Thickness of a couple of Hairs...

Cura coding and S3D coding are not the same coding routines... Do you really care it Cura and S3D are different by the thickness of a 'Hair' I sure hope not because you're setting your (3D printing experience) to be less than successful...

My expensive CNC Mill is good to only about 0.001 inch when tuned/adjusted

[airscapes]

Well that is why they are different programs.. they are not the same. Not sure what they have added to and broken in the current version of S3D, I am still using 4.01, but some of the things that control the wall thickness accuracy are:

Filament thickness and how much it deviates through the spool. You must measure and set actual average thinness in the slicer
Extrusion multiplier and how you have it calibrated it. I have found using a single wall square will end up being under extruded.
Extrusion width, are you using manual or auto (I prefer manual)
Perimeter over lap percentage
starting Z offset to a small degree

Digital caliper repeatability
Extruder drive, is the tension set correctly (hardware)
Over all printer hardware accuracy
Nozzle size (extrusion width in S3D) vs model wall size. When I create a dimensional model I want accurately I make sure the wall thickness is equal to my extrusion width of the nozzle I expect to use to fabricate it.
Speed the temp can also cause issue with accuracy to a small degree.

This is a mix of art and science and unfortunately there are software bugs in the mix as well.
Do not compare slicers, find one you like and Learn it inside and out.

[tomaski]

Thank you @airscapes for your very insightful post. I had extrusion width set to auto. Changing it to manual produces walls that are almost spot-on 1,20-1,21 thick

Filament thickness and how much it deviates through the spool. You must measure and set actual average thinness in the slicer

tests were run on a quality filament

Extrusion multiplier and how you have it calibrated it. I have found using a single wall square will end up being under extruded.

For that purpose I have always used the abovementioned cube: 25mm cube, 0% infill no top, no bottom, 3 shells for walls. Used average thickness of all 4 walls as a base for extrusion multiplier calculations.

Extrusion width, are you using manual or auto (I prefer manual)

that was it!

Perimeter over lap percentage

I have it at 40% (it's the Outline Overlap, right?)

starting Z offset to a small degree

I have it dialed in pretty well

[arhi]

What can be the reason that S3D is so much off?

They run under different assumptions and calculate stuff differently. One is not better than the other, they are just different.

Major difference is how they "expect" the overlap between lines to happen with 1.0 flow multiplier and extrusion width. (V1 is s3d)