Some notes on 3D printing

25 Sep 2018 - tsp

Because I’m currently having some exchange with a few people about 3D fused filament fabrication (“3D printing with filament wire”) I’ve started collected some (unordered) notes about some topics or problems that are often arising.

PS.: Comments are of course highly welcome.

Misconception about simply “buy and print”

There is this huge misconception that you simply can buy and print an FFF machine. At least the consumer ones all require a massive amount of calibration and preparation. At least if you want to achieve good results - no matter how much instrumentation there is. This is different with DLP and SLA printers of course but they are harder to handle in your own home and the resin is not stable for long term storage.

You will invest a huge amount of time in calibrating, trying out, modifying, etc. settings and even the printer itself. That’s the current state of buyable printers (I’d happily be conviced that there exist consumer products that don’t require that but I’ve never ever seen one that doesn’t require a huge amount of tuning and tinkering).

Which material to print

If you’ve no experience with 3D printing I’d highly suggest printing PLA. It’s just the easiest material. Yes, it’s biodecomposeable (you won’t really see that even when using it ouside) and yes, it’s glass transition temperature is that low that you can’t print parts that will be used in the automotive area (or which carry high stress and are used in direct sunlight - one can print garden equipment with PLA successfully tough) but it really is the easist to print and it doesn’t require an heated bed. And it required way less tuning and has a much more forgiveable temperature range that you can use.

If you want to print parts that can sustain a higher temperature range you could choose between ABS and PET-G. I’d personally recommend PET-G because it’s more UV stable and easier to print than ABS. You could also try Nylon (beware that it only adhers to some surfaces like cork and requires a totally different heat and cooling profile than other filaments) - you can even print with Nylon from gras-string trimmers effectivly.

Just a quick note about FEP. Even tough I really like FEP (Fluorinated Ethylene propylene - really similar to Teflon) for some applications (even though it’s really hard to get) don’t even think about using it in an normal consumer printer. That stuff get’s really nasty if slightly overheated (toxic gasses, etc.) - so you really need an air extraction and filtration system. And it’s expensive (really interesting whenever you’ve something to do with agressive chemicals or liquid oxygen though).

Importance of filament diameter

You often hear stories about problems with filemanet not being exactly eg. 1.75mm in diameter but having deviations of about 1.75 +- 0.005mm. That’s not that important (it’s important for fine tuning if you’re printing really fine details but if you’re using an 0.4mm nozzle for example - forget that tuning … as long as it reads 1.75 +- 0.01mm that’s totally ok).

Trusting temperature readings

Never try to compare temperature readings of various printers. First they may use different thermocouples. Then they may use different positions of the temperature sensor and may have different masses on the heated block. Also the temperature of the area around the printer with witch the filament get’s moved into the hotend plays a huge role. The effective temperature that the filament reaches is determined by many factors like it’s temperature when pushed into the hotend, the length of the melt zone, thermal capacity of the heatblock, feedrate and drain rate at which cold material enters and hot material leaves the melt zone, distance between melt zone and nozzle, etc. Use comparisons only as a rough hint and tune temperature yourself.


You want cooling. Never try to use your hotend without cooling (you could melt the filament up into an area where it’s not supposed to melt and clog your nozzle) - and you’ll also want cooling for the area directly below the nozzle. That is crucial for good bridging performance. The more (controllable) cooling you can provide directly below the nozzle directly affects bridging performance. Having one or two dedicated fans and airguides just for cooling (if not enclosed inside an heated cage) is really reasonable and helpful.

Delta vs. Cartesian

There is the common misconception that delta printers provide way much more print volume or provide better performance. This is generally not true - and they are much harder to calibrate. Then there is the drawback of having position dependent resolution with delta printers, their printing heads provide much less space for additional tooling (only hotend, thermocouple and bowden tube should go there - sometimes there is enough space for additional tooling like an leveling sensor).

Mechanics: Repeatability

If your machine does not fulfill a basic repeatability test (using your dial gauge to check how good your printer can return to an arbitrary selected “origin” after moving around and how huge the drift is) you will never ever get good print results. Fix your mechanics. Check if all belts (if you have some) are tightened, check if your ball bearings are not damaged and check that your guide rails are really parallel if your printer uses them.

Importance of correct temperature

Correct temperature is unquestionable important but it’s less important on the first hand than one would imagine. If you do not talk about stringing and bridging performance PLA can print over a really wide range of temperature as long as you’re not clogging your extruder. I’ve had a misconfigured machine that perfectly printed PLA at 350 degrees celsius (don’t try that if you don’t have a full metal hot end though because the teflon liner will burn if you do - and then you’ll have clogging in your hotend / have problems retracting after the hotend cooled down; and it will leak, have problems during loading, etc.).

Importance of the first layer

The first layer is important. If you don’t get the first layer right abort the print and tune until you get a good first layer. Never expect errors in the first layer to be compensated later on. That will not happen. It takes time to calibrate for a good first layer. Print an test object that’s only one layer high and look at some images how too high/low/perfect layers look like - there is a good one at . Bed Leveling can be done to 1/100 for a perfect result - don’t be impatient.

Bed leveling

Because the first layer is so important (see above) leveling your bed is important. Just don’t buy a printer without a probing mechanism or you will regret that. If you’re using a contact probe (either conductive or mechanical) remember that your nozzle has to be really clean (no spare plastic from the last print) to get correct results. If using an inductive probe it’s a good idea to set the trigger point at an height at which the nozzle is a few mm above the bed (!). This allows the probing to progress (with a cold nozzle) even in presence of some remaining material or dirt at the nozzle. If you mount the probe “perfectl” and the nozzle get’s dirty AND are using the probe also as endstop (a common setup) you may end up driving your nozzle into the bed and forcing it down until something gets damanged (bearings, nuts, motors, etc.). Let the system probe multiple grid points and average each point. That takes time (one some small and slow printers up to 45 minutes) but it’s worth the time.

If you leave space between the trigger point and real zero you are also capable of fine tuning your zero position (do that for example with the image reference references above) more easily without dismantiling your printer or print head all the time.

Bed material, Requiring a heated bed

If you start printing with PLA on painters tape and have adhession problems - it’s definitely not a problem with surface temperature. You can print PLA on painters tape without adhession problems and without a heated bed - it even adhers so good that you may have to re-build the surface with new tape every time you start a new print. If you’re printing other materials like Nylon, ABS or PET-G on the other hand you really require a heated bed - and you’ll want to buy cork (Nylon) or Kapton tape (ABS, PET-G). You really want to have Kapton tape in any case earlier or later for other things so that will be something handy to have available.

Stepper drivers on z Axis

Try to use separate stepper drivers for motors if you’re using multiple steppers for the z axis (typical cartesian printers have on on the left and one on the right side) - especially when using the most common “silent” drivers because they tend to overheat / do not provide that much current budget to not skip steps - which would lead to the axis being out of sync which on the other hand changes relation between your probe and your nozzle which completly destroys result from bed leveling and may even lead to damage if the probe is also used as end stop.

Because keeping the axis in sync it’s also adviseable to use separate endstops for both axis so they can return into synchronized state after they’ve been out of sync. They may get out of sync as soon as stepper drivers are put into a power down mode whenever you’re using microstepping - or when somebody has played around during the drivers had been powered down.


If you see periodic wobble during your print along the z axis - it’s nearly always misalignment between your guides and trapezoidal spindles. In that case do not try to force the spindle to be aligned with your guides if you’re not modifying all relative distances (motor position & guides have to fit the distance of your driving nut and bearing) - if you really want a quick fix let the spindles some clearance on their top end but always under all circumstances ensure that the guide rails are stiff and are not forced to bend by the movement. Having tilted spindles just changes the height your axis moves per revolution (of course it also leads to uneven speed because the spindle then looks oval but that’s most of the time neglectable).

Stiffness (Mechanics)

Yes. Stiffness of the machine is one of THE key elements to it’s accuracy. If you don’t have a stiff frame and stiff connections to the guide rails / etc. you will never achieve something that’s accurate. When you try to reach layer heights of about 1/10 of a mm you will be tuning the repeatable positional accuracy to about 1/100 or 5/1000 at least - and the machine has to reach that accuracy also under heavy load.

Over-Expectation of accuracy of 3d printed parts

The reachable accuracy of 3D printed parts is not compareable with the accuracy you can reach on your lathe or mill. It’s possible to solve some problems (like getting parts tight, etc.) with tricks especially developed for FDM machining (you have to design your models that way) but you will never get parts that are compareable accurate to milled or turned parts.


If you see ghosting (repeating of features like insets as “shadows” along a printed sidewall) - turn down your speed, reduce mass of the print head or raise stiffness along your printhead and bearings. The first one is always the easist. There are only two areas where speed really matters and that’s briding (not that important like one would imagine) and retraction (retraction and travel speeds). Vibrations are the death to all precission - so just raise printing speed as long as you don’t see any negative effects.

Dryness of filament

PLA gets brittle if you store it in an high humidity environment. It’s possible to bake filament at about 60-80 degrees celsius for some hours if you want to reduce humidity captured inside the filament. But most of the time humidity does not really affect printing performance with PLA (if you print with Nylon that’s an entirely different story). When starting you don’t have to be paranoid about getting PLA too humid as long as you don’t submerge it into water oder put it out into the rain - it just gets a little more brittle but prints well. Keep that for fine-tuning your environment.

Every printer and slicer is different

Yes, every printer has it’s own characteristics. There will be deviations between parts printed on different printers. And there will be even much larger differences when it comes to slicers. For exampel there are two slicers (Cura and slic3r) who require totally different corrections for inside diameters of holes - you really require CAD files that you can modify to fit your printer (easily done with parametric systems like openjscad or openscad, much harder to do with other CAD systems).

Don’t calibrate distances with test prints

Always measure them. Inside and outside diameters of printed objects are influenced by way much mroe parameters than just mechanical distances - temperature, slicer calculations, etc. Calibrate by measured distances until the moving distance of all ways matches the commands. Then tune other parameters.

Expensive CAD programs

You don’t require them. There are open tools for every approach. For the programmatic CSG approach there are tools like openjscad (which i prefer), openscad, etc. For people liking a more traditional CAD approach there is FreeCAD and for artists even Blender can export files that can be used for 3D printing. Expecting traditional designs to work for 3D printing

3D printed parts really have to be designed for 3D printing. You can use the same engineering ideas as for other manufacturing methods - but not everything works the way it works when building with other tools - on the other hand 3D printing allows designs that would not be possible with other methods. You really have to modify designs so they work with 3D printing.

Using your computer to control the print

Prints will take a long time - it’s really adviseable to use a separate system (can for example be a solution like octoprint for the RaspberryPi or a whole separate computer) to control your printer. Also from the standpoint of stability and responsiveness it’s a good idea to separate printer control from your everyday machines.

Fire hazard

If you’re not totally sure that your printer will not burn down and you’ve not taken precautions in case this happens - don’t run your printer unattended. Yes I know that this is hard if you want to run a 72 hour print but it’s better to re-shedule a print than having a burnt down house.

Most common causes for fire are partially broken cables (when not using proper wires for flexible connections, not having strain relieve, not using collets at the wire ends, using too small cable diameters, etc.) or having underdimensionalized parts like the MosFETs, using PCBs with too small traces. Then there is the possibility of thermal runaway in case the thermocouple fails - this can happen because it gets destroyed, disconnected or simply because a screw loosenes and it falls out of the measurement position - in this case an unprotected thermal regulation system just heats with full power - and may reach temperatures that the system has never been designed for. There are various countermeasures (limiting the maximum current, adding fuses, adding a second thermoelement and having software based thermal runaway protection - the latter one is for free). Oh and do never forget strain relieves (yep said that twice).

Although there are solutions (like octoprint) that allow remote observation (for example via webcam) and smoke detectors also always remember that it’s not much of a help if you know that your printer is burning if you cannot get there to extinguish that fire. So it’s not sufficient to have a webcam and temperature alarm set up - at least not if the printer is not in a different part in the same house or flat.

Entangled filament

This is nearly always your own fault. Think about it - how do you think a manufacturer could wind up entangled onto a spool? Just do NEVER ever let the end of your filament go loose or hanging around without it being either fixed in the two holes on the outside of the spool that are exactly there for that, squeezed in some tooling or being in the extruder and hotend or the bowden tube. Then you will never get entangled filament.

Bowden tube lubrication

Even though these are usually made out of Teflon it’s a really good idea to lubricate them with suited oil (use white oil - the one for sewing machines) in small doses. This reduces friction and the oil burns in your hotend as long as you don’t use too much. Oil helps to keep the bowden tube clean and frictionless - which can make a difference between a successful and totally failed build.

This article is tagged: Hardware, DIY, 3D printing

Dipl.-Ing. Thomas Spielauer, Wien (

This webpage is also available via TOR at http://jugujbrirx3irwyx.onion/

Valid HTML 4.01 Strict Powered by FreeBSD IPv6 support