Sunday, November 10, 2019

3D Printer / Wall Thickness, Detail Level and Accuracy

This isn't about the actual accuracy of the printer itself (which depends on build quality, design, components etcetera) but about the minimal detail level that the printer can handle, and when to use what wall thickness.

Again, I'm not an expert, but got lost on my way to redemption... 😅

Wall Thickness

My primary driver was to find optimal wall thickness, to a. speed up my prints, b. get vertical walls (horizontal wall thickness) as strong as possible, and c. smoothen vertical surfaces.

Everything else was a bonus...

Every printer is different. Every filament is different.

In the end you just might have to experiment a little with your settings...

The (test) model

(So tempted to put Doutzen Kroes in here 😇)

2 mm seems to be a very sturdy size, horizontal and vertical, so I used that as the maximal wall thickness. Then I ramped it down to 0.2 mm wall thickness in 0.1 mm steps. See the images below.

As 3D printers behave very differently horizontally and vertically I created a horizontal ramp as well.

The 3D model in Fusion360 (click to enlarge):


When you run this through your slicer software, it will try to use the capabilities of your printer as best as it can to reach the specified size.

When walls are too thin, the software may decide to either remove that wall, or use the minimal size your printer can produce. (This depends on your slicer software.)

I'm using Simplify3D (decided to buy it after all) and tried different settings on the model, then used the build-in simulation to have a look at the results.

Cura should produce similar results, but keep in mind that each slicer behaves differently.

Keep in mind that, although your slicer may suggest you can go down to a specific level that doesn't necessarily mean your printer can actually do that!

Z axis

Wall thickness

For a typical Cartesian (XYZ) printer the thinnest element you can print on the Z-axis is the thickness of a layer. So for horizontal elements the minimal wall thickness printable (not the same as usable) is same as layer thickness. However, single layers have little to no cohesion, so as a rule of thumb the minimal thickness of horizontal elements will be 3 times layer thickness. Usable thickness depends on the distance to bridge and the force the final object has to handle.

Detail level

Obviously, the minimal detail level is the same as the layer thickness. Assume you have selected a layer thickness of 0.1 mm, then you could print some fine details 'on top of' a horizontal surface in the next layer. So details of 0.1 mm in height are a possibility. But... more detail in Z-direction means a lot more printing time... You cannot print any details below layer height.

Also keep in mind that the printer uses a certain extrusion diameter. See the following sketch (click to enlarge):

You can't print anything like the red 0.1 x 0.1 mm square, as the extruder is already larger than that. (You can actually print smaller objects, as the printer can control the amount of extrusion, but this would lead to lower accuracy.)


Accuracy is a different story, and depends on two factors. First the accuracy of your printer itself, obviously. The second factor is layer height. If you set the layer height too large you won't end up with the right dimensions / detail level. This would depend on the software, which might be limited to multiples of the layer height for any dimensions on the Z-axis.

X / Y axis


Accuracy is defined by the mechanical aspects of your printer. Any accuracy set in the slicer software typically has little to no impact on the printed object, unless the slicer software tries to simplify the model to further speed up printing.

Detail level

The detail level is defined by the minimal horizontal (X/Y axis) step size of your XYZ printer. If your printer does make 0.1 mm steps horizontally, the best detail level will be 0.1 mm. However, keep in mind that the size of the extruder plays a role as well. If you extrude 0.4 mm of material, then yes, you can move the printer head 0.1 mm to the side, but the area affected by the step is defined by the amount of material extruded.

However, detail level in one dimension can be much higher. See the sketch below:

So although the extruder dimensions are larger than the step size, you can print more detail in one dimension, the other dimension is affected by the extruder size.

Wall thickness

If the extruder size is 0.4 mm, then you won't be able to print any vertical walls below 0.4 mm. (Actually, you can, I'll get back to that in a moment.) Typically your slicer software will remove any object of less than extruder size.

Deliberate 'stretching' of filament

You can instruct the printer to extrude less than the extruder width. This is accomplished by extruding as slow as possible and moving the printer head at sufficient speed. The 0.4 wide extruded material will then be stretched into something thinner. If this actually works depends on your printer.

Default settings

The extruder on the 3D20 is 0.4 mm. The default extrusion width was set to 'manual' 0.4 mm.

I'm not sure if the 3D20 can handle a 0.2 mm nozzle (no, it cannot). Other printers can.

Low quality / 15% infill / 2 bottom layers / 2 top layers / 2 shells

Low quality sets the primary layer height to 0.30 mm.
Result: 20 layers. 6 mm / 20 = 0.3 mm per layer.

Poor quality but obviously quicker prints. With only 2 bottom and top layers probably not usable for actual use. Simplify shows the following result for layer 15:

  • Vertical walls 0.4, 0.3, and 0.2 are not printed.
  • At 1.5 mm the 'second' perimeter / shell is no longer printed.
  • At 1.0 mm there is no longer any infill, 1.0, 0.9 and 0.8 will be weak.
  • It takes 6 steps to go down from 2.0 to 0.2 floor thickness.

Horizontal walls of 1.6 to 2.0 mm might actually print faster than 1.5 to 1.1 mm. A second shell is a single, direct movement and does not involve patterns.

Layer 5 shows the infill pattern:

Medium quality / 20% infill / 2 bottom layers / 2 top layers / 2 shells

Medium quality sets the primary layer height to 0.20 mm.
Result: 30 layers. 6 mm / 30 = 0.2 mm per layer.

At least 50% slower, but it produces usable results (though it's better to keep top and bottom layers to 3 if you actually want to use this.)

This is layer 23:

  • Vertical walls 0.4, 0.3 and 0.2 are not printed.
  • It now takes 9 steps on the horizontal ramp.

The infill pattern at layer 7:

Medium quality / 25% infill / 3 bottom layers / 3 top layers / 2 shells

Vertically the same resolution, but with an additional top / bottom layer and a little more infill this is good enough for most simple objects for 'practical use'.

Layer 23 looks like the previous print, as expected:

The infill pattern is a little denser (barely noticeable). Here is layer 7:

Medium quality / 30% infill / 3 bottom layers / 3 top layers / 3 shells

Adding an additional shell layer triggers infill on 2.0 and 1.9 wall thickness:

The denser infill only affects thicker walls (you could think of the 'bottom' part of the model as just another thick wall):

High quality / 30% infill / 3 bottom layers / 3 top layers / 3 shells

Medium quality sets the primary layer height to 0.10 mm.
Result: 60 layers. 6 mm / 60 = 0.1 mm per layer.

Layer 44:

The vertical resolution is twice as high, but nothing else appears to change.

High quality / 30% infill / 3 bottom layers / 3 top layers / 3 shells / extrusion width auto

The default extrusion width setting for the 3D20 was 'manual 0.4 mm'. Changing it to 'auto' does change the print.

On manual: (click to enlarge)

On auto: (click to enlarge)

As the flow can now be adjusted, the 'inside wall' on 2.0 and 1.9 mm can be thicker, thus no infill is required.

High quality / 30% infill / 3 bottom layers / 3 top layers / 3 shells / extrusion width 0.3

Now you might try to adjust the extrusion width to something smaller than the actual extrusion diameter. This just might result in a higher resolution print, but there's no guarantee whatsoever. The extruded PLA is now thinner than the extrusion nozzle, and may move in unpredictable directions.

Assuming it would work, you would probably want to increase the infill extrusion a little. By default it's 200% of the outside wall. Now let's say I adjust the extrusion diameter for the outside wall from 0.4 to 0.3 (by 25%). Infill was at 200% (200% x 0.4 = 0.8) so to keep the resulting infill the same I would change the parameter to 260%. (0.8/0.3 = 266).

It just might work... but probably not on a cheap Dremel 😄 but if it would work, it would look something like this (it actually did work on the Dremel, mostly):


I will do a test print later today (it got a little late / early 😳) to see if my conclusions are right, but for now...

Keep in mind I was primarily interested in horizontal wall thickness. Everything else is stuff I learned whilst trying 😁

With medium quality, 3 top, 3 bottom, and 2 shell layers, nozzle manual to 0.4 mm, Simplify3D's results suggest the following:

Horizontal wall thickness (vertical surfaces)

1. From practical experience 2 mm is very stable, but it does not have the strength offered by additional infill. If you need maximal strength switch to 3 shell layers. Note that a 2 mm wall still may 'tear' at the point where it connects to another horizontal surface.

2. 1.6 mm appears is the optimal wall thickness if you want to avoid infill 'puckering' and resonance patterns. If you have proper inner / outer-wall binding it is going to be strong, but if you suffer from a little under extrusion the different layers may start to 'peel'.

3. 1.5 could be stronger than 1.6, but the outside surface will be rougher. If your printer (firmware) doesn't like frequent direction changes it could print slower than 1.6, in spite of the additional pass at 1.6 mm.

4. Do a test print with 1.5 and 1.6 mm wall thickness and see what you prefer. It differs between printers and the object to be printed.

5. 1.0 mm prints a lot faster than 1.1, and is 'hollow' thus flexible. If you need rigidity go for 1.1 or up.

6. For very thin elements try 0.8, 0.7 or 0.6 mm. At these widths objects are inherently unstable due to extruder size.

7. 0.4 and down won't even slice, let alone print. Forget it. (They're still visible when it comes to surface details though.)

8. Use at least 2 shell (perimeter) layers for MQ, consider 3 for HQ. When forcing extrusion of 0.3 mm from a 0.4 mm nozzle consider 4 shell layers.

Vertical wall thickness (horizontal surfaces)

1. From practical experience 1 mm is a about the minimum. It is quite flexible.

2. At 2 mm the object becomes sturdy. For horizontal parts of the model I would try to use 2 mm or thicker unless there are reasons not to.

3. In MQ (0.2 mm layers) using 3 layers top and bottom results in a better surface. With 2 layers holes are likely to appear.

4. In HQ (0.1 mm layers) use at least 3 layers but consider 4.

5. When combining HQ and forcing extrusion of 0.3 mm from a 0.4 nozzle consider using 4 or 5 layers.

6. Many sources on the internet suggest a higher number of top and bottom layers. If you run into uneven surfaces or even pits and hollows, increase the number of top and bottom layers. I found 4 layers of 0.1 and 3 layers of 0.2 workable and use 5x 0.1 and 4x 0.2 for stuff that needs a nicer surface. Experiment...

Surface details

When printing objects with (vertical) surface details anything below 0.2 mm is probably lost or inconsistent, though it may add 'texture' to, for example, model railroad scenery or role playing games miniatures.

Surface details are not the same as accuracy, so although your printer may be inaccurate the surface detail might come out fine.

Test prints

1.5 versus 1.6

Here is an example of 1.6 (top) versus 1.5 mm (bottom). Note that the same complete hollow box 130x130x35 mm would take 5 hours with walls of 1.5 mm but only 3 hours with walls of 1.6 mm. With a little over extrusion (1.02) the individual lines were well fused, so unless dimensional accuracy is important I'll go for 1.6 from now on.

MQ versus HQ 0.4 versus HQ 0.3

Three test prints, from top to bottom with decreasing detail level. The top one is using an extrusing width smaller than the nozzle, ie. 0.3 instead of the nozzle size 0.4. This actually works, but the printer accomplishes that by keeping the speed of the printing head rather high.

I'll have to print some miniatures or railroad scenery models or something similar to see if it is worth the effort. Anyway, the test prints...

Note that the very first layer appears in HQ sometimes causes troubles, and I'm not the only one experiencing this... An option to modify (over)extrusion for the first layer would be nice, to 'squash' the plastic into the building plate, but there is no such thing. (It would be nice if Simplify3D would add such an option. I know their concept and approach is correct, but sometimes a 'messy work around' saves the user a lot of hassle.) There is a 'Variable Settings Wizard' which might be abused to get the proper results but I haven't tried this yet 😑

All prints came out as expected, and confirm my earlier assumptions and conclusions. Ah... all I now need is a cigar...


This guy seems to know a lot more about 3D printing than I do...

No comments:

Post a Comment