I do a lot of 3D printing, and often I spend a lot of time trying to improve print quality. It seems like a game of whack-o-mole, you might knock an issue out for a while but a new one will pop up soon. I wanted to make this log so that I (and maybe others) could keep track of changes I made that improved quality, given specific issues. I find that a lot of the "guides" online are actually completely useless, since they don't take into account the wide variety of things that can go wrong in 3D printing.
My Standard Test Print
I needed something a little bit better than cubes and Benchys, and something more specific to what I usually print. Specifically, I need support material to be as easy to remove as possible, reliable, and clean. I also print a lot with lines thinner than the nozzle width, and need them to come out continuous. Overhangs, surface quality, and dimensional accuracy are my other requirements.
This model didn't take very long to design, but it works very well for what it is:
The base is 50mm square, and that's how I do quick checks of accuracy. For actually tuning a printer, I will print a series of squares and check each dimension at different sizes. When you change the extrusion multiplier, you can easily get a ~0.3-0.5mm difference in the exterior dimensions of your part.
The stepped boxes are for testing wall width, they are thickest and the bottom and gradually decrease in wall thickness. The overhang is just that, an overhang to test supports. I find that cylinders are usually where you will see extrusion issues the most, and small cylinders specifically show heating problems too. The pyramid shows surface quality, and the fact that it's decreasing in area as it goes upward may also show cooling issues or flow consistency.
Examples of Poor Quality
Fixing Gaps - October 2022
Data: Overture, Matte Black filament (Old, but dried for 12hrs). Filament is thin, ~1.71mm. 190/60, P1
Turning the fans off resulted in great layer adhesion but horrible quality. Top surfaces were pillowed all the way to holes. 50mm test piece was almost unrecognizable. Cylinder and overhang support test came out wrinkled and potted. The pyramid came out surprisingly good. Nozzle is old.
190/60 P1 re-run, much better results. Fan full speed. Accel turned down to 500. Extrusion at 90%. Speed at 84%. Significant gaps in exterior walls still. Idea: Print inside walls first and see if that helps. Significant pillowing gaps on top surface still. Idea: Disable Cura bridging settings in case it's doing weird stuff over the infill.
200/60 P1 re-run, slightly better top surface (after bridging disable). 10C higher temperature seems to have helped. Idea: Up the extrusion 13% (103%). I upped it to 110% to see extra results.
200/60 P1 re-run, turning up extrusion to 120% (speed 84%) improved quality once more. Pillowing still significant on top surface. Idea: 110% extrusion, increase top line width but increase speed and extrusion. Hopefully the speed bridges the gaps more smoothly, and the larger line width provides a wider base for each subsequent line.
200/60 P1 re-run, extrusion set to 110% and having upped the top/bottom speed didn't help much. Also set the top line width to 5.5mm. Idea: Shrink line width, slow speed. Turn off monotonic ordering.
200/60 P1 re-run (84% spd): Line width 0.3mm, 30mm/s top/bottom, monotonic ordering off. Result: No real improvement. Idea: Return speed and line width to normal, turn monotonic ordering on, disable bed heat. Increase min layer time.
200/33 P1 re-run (84% spd): That did not improve the top surface much. Idea: Bite the bullet and increase infill from 25% to 30%. Changed from tri-hexagon to triangles.
200/60 P1 re-run (30% infill now): significant improvement. "Good enough" print quality, probably. The printer still didn't do a great job bridging the infill, but after a couple top layers the gaps were covered.
Lessons: Extrusion multiplier can have a great effect on surface print quality. Leaving the fans off can cause huge issues. Printer P1 probably also needs greater cooling, since improvements were still seen when increasing to the max. Bed and hotend temp didn't make a huge difference with the overture black matte filament.
Constant Clogging
Clogging is one of the most annoying print issues since it often results in a half-complete print that's just a waste of filament. One of my printers keeps clogging, and the normal suspects don't seem to be at fault this time. I've changed the nozzle (twice, it seems to help a bit). I cleaned the whole extruder, taking the dust off of the heatsink fins to help with cooling and cleaning the filament bits off of the gears to improve grip. I also tried a different filament, and I removed the guide tube I had from the roll to the direct-drive extruder. I also checked the hotend temperature, and it seemed OK with a 210C set temperature being about 210C on the measurement device. I've also tried printing with different layer heights and without retraction, to see if it's a flow or retract related issue. Both of those approaches probably should be more thoroughly documented and explored. Usually the printer would clog after the first few layers, often halfway through the print.
Anyway, at this point I think I have 3 main suspects:
The spring tension on the feed gears is too low
The hotend geometry is worn to the point that filament is hard to push through it.
The extruder body cooling is insufficient, causing the filament to start to melt prematurely.
The first thing I did was to try and make a little gcode file for testing the extruder and hopefully trying to get it to clog. "Make it break" is often the most important step in troubleshooting since it forces you to think about the problem from the machine's perspective essentially. The gcode file I made is supposed to simulate a fairly complex print with lots of retracts and travel moves. This is done by using pauses in between some retracts and at the end of all un-retract + extrusions. I scattered a few extra random pauses to hopefully simulate travel moves.
Now that I have this file, I will try different retraction settings (I'm using klipper's firmware retract in this file), speeds, and temperatures. Hopefully I can identify some variable that consistently ends up with a jam/clog. After a bit of testing, just about every run of the file did not result in a jam of the hotend. I did get it to clog a few times, but it didn't seem related to any of the variables I was changing. I tried some real prints afterward with the same results as before - constant clogging and failures. The print quality is actually pretty good on this machine, it just refuses to print consistently. I also tried increasing the filament extruder tension by jamming a pair of tweezers in the lever to make it press the gears more together.
When pulling out the filament of the hotend after a clog (when it's possible to do so...) the end seems swollen, so maybe somehow that is causing a clog? I've also been dealing with this problem for the past a bit, and especially now it seems after a new nozzle. So I'm going to try a different nozzle now and see if the problem persists. The first one I tried is a "POLISI3D" brand, nickel-plated brass 0.4mm nozzle.
After changing the nozzle I saw no difference in performance. So, the remaining components that haven't been replaced are the heat block, the extruder gears, and the feed tube heat break. The heat break itself seems like the most likely culprit, and indeed in the past when I've had trouble with all-metal hotends jamming it got fixed when I replaced the heat-break. I don't think the problem is caused by excessive heat creep, I've tried a lot of feed speeds (more filament flow = better cooling), I've turned down the temperature, and the same hardware was working fine before. The extruder cooling fan seems fine, and was blowing great when I took it off and put my hand over it. So my suspicion was that after many many hours of operation that my all-metal heatbreak had worn out a bit and was rough, causing filament to catch on it too much.
Rather than simply replacing the heatbreak (they are almost impossible to replace on these Biqu H2s I have - the set-screws self destruct on the slightest touch), I decided to try and "season" the thing by essentially oiling the filament as it gets fed into the extruder. I did this by putting a bit of avocado oil on a paper towel and clamping it around the filament. I also noticed that my guide tube + feed encoder was causing quite a bit of drag and temporarily removed it. My guide tube is just a bowden tube, which is technically too tight since its purpose isn't to be a stiffly constrained path but just to make sure the filament isn't yanked on by the extruder moving around. But again, that was working just fine up until about a month ago. But anyway, surprisingly, the avocado oil seems to have helped! Maybe even have completely solved the problem, since I haven't had a jam yet. I've only done 2 successful prints so far, but they look amazing and I haven't had that kind of print quality for a while from this machine. I will tentatively add the feed encoder sensor back inline after another print or two and see if all is well after that.
Lessons: The surface finish (and type) must play a massive role in the performance of an all-metal heatbreak. With a worn (or potentially low quality) heatbreak, introducing a small amount of a high smoke point oil like avocado oil can improve performance and fix clogging issues. I should have tested other filament types like PETG and ABS in the problematic extruder, since I found online some people saying it was easier to print with them than PLA in some all-metal heatbreaks.
Comments