Fixing Inconsistent Extrusions Systematically

This is my Cetus MK3 that Tiertime sent me more than 2 years ago and I do feel a bit bad that I never reviewed it. The Cetus was praised by many at that time for great print quality and all linear rails construction. Unfortunately, I never really liked it. Well, let’s say I liked the idea of a minimalistic machine but their control board running proprietary firmware, an oversimplified slicer, noise and only okay print quality drew me away from it. I always felt that it could do better when seeing it on my shelf, so I thought I’ll bring it back to life by simply replacing the control board and this way fixing all of my issues. I’ll cover this endeavor in an upcoming video, so if you want to learn how you can simply revive an old 3D printer, put in my case a Mellow FLY E3 with Wi-Fi, silent stepper drivers and RepRap firmware with Input Shaping in it make sure to be subscribed.

New electronics with custom adapter board and MELLOW FLY RRF E3

In this article though I’ll show you how I was able to fix my extrusion inconsistencies because they were not solved by the new mainboard I put in it. In the end the solution was obvious, but it took me a while to get there. The fix was super simple, hence I thought I make a quick video about it.

CETUS MK3 printing

But let’s start at the beginning. What do I mean, when I talk about inconsistent extrusions. Well simply put – surfaces that should look regular and smooth have this rough texture. One layer seems to be thicker than another or they just don’t line up properly which makes prints look especially bad if you inspect at them at the right angle.

Rough texture on prints

After trying this and that over the last two years and never making it better I finally took the time and approached it systematically. I thought about how the phenomenon might be generated and what the causes for it might be. The inconsistent extrusions or layers that I saw on my parts might be from an inconsistent layer height, where thinner layers are squished more to the side than thicker ones. Then there could be play in one or more axis which would cause layers to not line up properly. And finally there is inconsistent flow from the hotend.

Let’s quickly talk about reasons and solutions for all of them. I can’t go into super great details, because then this video would end up being hours long but if you are interested in a specific topic or even encountered one yourself on your machines or if I forgot something, then leave a comment down below.

First – inconsistent layer heights: This might come from a not smoothly running z axis, if maybe rollers are too tight or too loose. The Cetus uses linear rails, which I even disassembled to make sure that everything runs fine. Z-wobble is another phenomenon that can happen if you have a crocked z-axis. Z-wobble though shows a very regular pattern in the z-direction, which wasn’t the case on my parts, plus, the Cetus uses a belt driven z-axis which usually doesn’t show these artefacts. As I’ve shown in a previous video, a badly tuned heated bed can also cause a pattern very similar to z-wobble but since the Cetus doesn’t use a heated bad, this can’t be the reason for my problems either. Then there are the, in my opinion controversial Magic layer heights as some call them. This basically means that you should choose a layer height that is a multiple of a full step of the z-axis motor to avoid problems if micro stepping doesn’t work perfectly. That didn’t make a difference on my machine, maybe also because the TMC2209 drivers installed do a pretty good job with microstepping in StealthChop and Spread cycle mode. I even measured the positioning accuracy of the z-axis with a dial gauge at different layer heights which all showed that the tolerance is below 1/100th of a mm. Nothing made a significant difference in print quality why I concluded that my problems are probably not caused by inconsistent layer heights.

Next – play in the x, y and extrusion axis. This is something that is more common than one might think and usually caused by the grub-screws on the belt pulleys coming loose over time which causes play between the shaft of the stepper motors and pulleys. This can cause wired artefacts on prints, mainly with layers being slightly shifted from time to time. Well, the Cetus uses press-fit and glued pulleys on all motors so this wasn’t obviously my problem. Similar to loose pullies, you might also have a hotend that wobbles around, either because the bolts holding it are loose or the nozzle isn’t tight. I checked both on my machine, where everything was properly tightened.

Let’s finally get to inconsistent flow from the hotend and this is probably the broadest topic. First, if your filament diameter is constantly varying of course, extrusions won’t be constant either. Nowadays this is usually not a problem anymore, but I still tried different brands and rolls which didn’t change a lot. Then there are the print settings themselves. Especially during speed changes you can often see under extrusion when the printhead accelerates and over extrusions when decelerating just because the extrusion system always lacks a bit behind. This can quite simply be checked if you just print a model really slowly where the Hotend has enough time to be line with the movement. If you run a more current firmware, linear advance or pressure advance also try to minimize that phenomenon. Though, regardless of how slow or fast I printed or how I tuned pressure advance, the pattern stayed the same. At that point there wasn’t that much left, and I looked at the extruder itself. The Cetus uses a super simple direct extruder without a big heatsink or complex gearing and pre-tensioning system, which works better than one might expect. But due to the lack of pre-tensioning solution you can’t really control how much the filament is pressed onto the extruder gear. If it slips or strips the extruded material won’t be proportional to the rotation of the extruder motor. I tried different extrusion temperatures and extrusion speeds which didn’t change anything and even the imprinted pattern on the filament looked fine. I even changed, and removed the Bowden tube that feeds the filament to the extruder because I thought that this could cause addition drag in the system but no change in the results.

More or less the only thing left was the feeding system itself. The Cetus uses a custom extrusion gear that is directly glued onto the shaft. I manually moved the motor as fast as possible with some G-Code and there it was. The extrusion gear wobbled on the shaftnot because it was loose but because the bore in the extrusion gear was off-center. This causes a serious problem, because that makes gear rotation and filament feed not linearly proportional anymore. If the gear is wobbly the extrusion radius constantly changes. Over one rotation of the gear the extrusion amount is not constant but varies causing extrusion inconsistencies.

wobbly gear

Unfortunately the extrusion gear on the Cetus is custom and glued, so not easily replaceable. I tried grinding it down but due to the amount of wobbliness I basically removed all the teeth on one side rendering it useless. Because I was out of options the only thing left was trying to replace the gear. Removing it was simpler than I thought and by carefully using a blowtorch and two flathead screwdriver I was able to remove the existing one.

removing the gear

I still had a bunch of old extruder gears lying around but the ones currently popular on many Bowden style printers are 11 mm in diameter, in comparison the stock 10 mm one of the Cetus. So by replacing the old gear, I couldn’t use the old idler anymore and had to design a new one using Fusion 360 with some reference dimensions plus a photo. The first version looked pretty much like the old one, also relying on the flexibility of the part for pretension. I tried different materials and noticed that this isn’t an optimal solution so I added a small pre-tension spring that should make everything more reliable. I printed the first idler from PLA out of laziness but that got too soft due to the heat of the stepper motor. I’ve now been using a PETG part for a while and that worked flawlessly. STL and CAD files are all available in the description.

Stock, V1 and V2 idlers

I’m usually not a big fan of modding a printer until there isn’t anything left of the original machine but with this simple replacement the print results now are just shy from perfect! Of course, these results are now also with the updated mainboard but I’m quite certain that the electronics update mainly helped to make it basically silent and way easier to use with RepRap firmware.

Now I don’t want you all to rip out the extrusion gears from your various printers and blame bad printing quality on them. Though, it might still be worth the time, checking your extruder gears if your print results look strange. And maybe there is another point on the list that I talked about that might be the reason for your problems. This was honestly the first time that I thought about the possibility that the extruder gears might play such a significant role for print results. I’m not sure how widespread this problem is but maybe there are more badly manufactured extruder gears out in the wild. So, if you have ever encountered something similar on your machine, please leave a comment down below!