Calibration, then under extrusion

I calibrated slic3r yesterday as per these instructions. Worked exactly as explained and I thought I'd try a test print (the octopus from thingiverse). Printing at 1 shell I noticed there were holes. A results of under-extrusion? I only let it print for about 3-4mm then cancelled the print. The measured shell thickness was now 0.22mm, whereas after the calibration it measured 0.4mm. What happened? The parameters of both the calibration piece and the test piece were identical.

Material: ABS (esun)
Print feed: 50
extruder temp: 225
HBP temp: 95
Nozzle dia: 0.4mm

---

Sebastian

Flash Creator X
Creality CR-X
Anycubic Photon

I have come to the conclusion that the thin wall test is not at all accurate or correct. I think the reason is that the correct extrusion amount is when the extrusion width is accurate when constrained on both sides (as in solid infill). When one or both sides are unconstrained, they will bulge out and produce an extrusion that is on average the correct width, but is thicker at its widest point. The thin wall test measures the thickest width of the extrusion, when the correct value would be the average width (between the ridges and valleys of the layers). Thus adjusting the extrusion to make the wall thickness correct will result in under-extrusion.

A better way I have seen is to print a cube with 2 or more solid top layers and an infill of 95%. After 20 or so layers have printed (and it is printing infill), pause the print and check the infill under a magnifier. You should just be able to see very small gaps between the lines. Now complete the print and examine the top solid layers. You should not see any gaps between the lines.

If there are no gaps on the infill, you are over-extruding. If there are gaps on the solid layers you are under-extruding.

Ignore the first few layers, because the bottom layers are affected by inaccurate bed levelling and homing.

Dave

Thanks, Dave. I'll try that. What do you set as extrusion multiplier then?

---

Sebastian

Flash Creator X
Creality CR-X
Anycubic Photon

Quote
sfinke
Thanks, Dave. I'll try that. What do you set as extrusion multiplier then?

Using the thin wall test I used to set 83% (0.83), but the new method is OK at 100% (1.00) and gives better results (except on overhangs where it seems that under-extrusion reduces curl-up).

Dave

Here are some results of last night's efforts:




On the left is a solid 20x20x10 calibration cube; on the right is a single shell calibration piece as per the Slic3r extruder calibration method. Both were sliced by Slic3r.

All relevant setting were filled in as required. All extrusion diameters were set to 0.4mm. I then printed the single shell item. Shell thickness cam out at exactly 0.4mm. Great. Looks very smooth and clean. Left extrusion multiplier on 1 as the shells were on size. Next I printed the solid cube as a test and well, you can see the result above.

I followed the calibration instructions very closely. I tried 3 brands of filament. All filaments were measured in multiple places (10 to 15 measurements over a two metre length per spool). The bed is levelled. My thinking is that if the extrusion multiplier is spot-on for a single shell print then it should work for any print. The height of both prints is right (10mm). But the sides are telling a different story i.e. very rough in the solid cube, smooth in the single shell.

Interestingly they are both over-size in X and Y. I always scale up my models by 0.7% to compensate for shrinkage. In Makerware the cube then comes out at 20x20, however in Slic3r the same model comes out at 20.15x20.15. This shouldn't happen either.


Some possibly helpful info:

•FFCX
•Makerware 2.4/Slicer 1.2.9 (both x64)
•Material: ABS (Esun, Flashforge, local brand)
•Feed: 50mm/s
•Extruder temp: 225
•Layer height: 0.1mm

I feel am missing some simple step but cannot see it. Thanks.

---

Sebastian

Flash Creator X
Creality CR-X
Anycubic Photon

You should not scale-up your models. Shrinkage does not affect FFF prints in the same way as it affects injection moulded parts. To illustrate, consider the following thought-experiment. Imagine that the plastic has a huge 50% linear shrinkage (=87.5% shrinkage in volume!). A 100mm cube that is injection moulded would thus shrink down to 50mm per side. Now imagine FFF printing that same cube. The nozzle lays down a line of molten plastic 100mm long for a side. When it gets to the end of the line, do you really think that the line of plastic will have shrunk back and be only 50mm long? No, what happens is that the plastic progressively shrinks as it cools, but as the length is being maintained by the nozzle position, the shrinkage will affect only the width and height of the extrusion, not its length. The printed cube will be 100mm per side (near as dammit). The shrinkage will affect the gaps between the lines of solid infill. Now imagine what happens to the Z dimension. As more layers are added, do you think the nozzle will get further and further away from the last layer until the top layer is printed in the air 50mm above the model because of the massive shrinking? No, again what happens is that the previous layer will shrink to leave a gap of half the layer height, but the following layer will fill the entire gap - again resulting in a thinner width rather than a lower height. In fact, the combined stretching effect of the XY shrinkage and the bigger gap due to Z shrink will result in an extrusion width that is only 12.5% the width laid down - your part will be the correct size but full of gaps.

The end result is that with FFF printing, shrinkage results in the part retaining its dimensions but being a lower density than it would have been with no shrinkage. The shrinkage is effectively translated into (bigger) holes in the part (the gaps between infill lines). Compensation for shrinkage is thus made in FFF printing by increasing the extrusion factor rather than by increasing the size of the part, though usually the extrusion error is significantly greater than the shrinkage so there is no point in making such small adjustments.

Shrinkage can however cause a printed part to warp (your photo shows warping on the solid cube). That is a result of uneven shrinking, caused by a relatively large volume of plastic not all cooling at the same rate. In the solid cube of your photo, the outside of the cube cools faster than the inside - I can clearly see a circle on the base of your print that demarcates a hot core that resulted in the colder edges lifting. There are various ways of minimising warping. To prevent the base or corner curl-up which I can see on your photo, you need to coat the build plate with something that adheres to the 1st layer more strongly. This is sufficient to prevent mild warping such as you have - the uneven shrinking still creates an upward pull but it is not strong enough to pull the plastic off the bed so instead stretches the plastic slightly, keeping the Z dimension accurate. Try coating your bed with wood glue or solvent cement used to join plastic pipes. I use ABS dissolved in acetone. For worse lifting, layer separation, warping in the XY direction or curl-up of overhangs you can cool the plastic as quickly as possible after it has been extruded so that shrinking occurs entirely on the thin extrusion when it is able to be translated to a stretching effect rather than developing a large block of hot plastic which will behave more like an injection moulded part and pull the rest of the part out of shape. That can be accomplished by a cooling fan with air outlets directed close to the nozzle - but a downside is that if the extruded plastic is cooled too quickly it will not bond (fuse) very well to the previous layer or infill creating a weak part that readily delaminates. You can design for minimum warping by trying to avoid large solid volumes of plastic (adopt the same design considerations as are necessary for moulded parts). Avoid sharp corners as well. A heated bed helps a lot, but will only stop warping on prints that are not too tall. Possibly the best method is to put the entire build volume inside a heated chamber so that the whole model remains evenly hot. The chamber should then be cooled slowly after the print is complete to give inner areas time to cool at the same rate as outer areas. That method will result in the whole part shrinking similar to injection moulding - but it will be far less because the temperature change is not so great (it is not being cooled from anything like as hot as its molten temperature).

The vertical lines on your photo are caused by a different mechanism (there are two causes of that effect that I know of). The easiest way to reduce or eliminate them is to print more slowly (usually only the outer perimeter needs to be printed more slowly).

[edit] - I forgot to mention the most common cause of warping - draughts. Even a very slight draft caused by opening a door, or a slight breeze from the chimney of an open fireplace can cause huge warping that would otherwise not occur. This is because the draught causes a very rapid cooling of the outside of the part. When I was unable to eliminate draughts from a room, I put a plastic bin-bag over the printer to keep it draught-free - but be careful that this does not result in your printer overheating.

Dave

Edited 1 time(s). Last edit at 07/10/2015 09:46AM by dmould.

Hope to revisit the Slic3r calibration this weekend...

---

Sebastian

Flash Creator X
Creality CR-X
Anycubic Photon