3D printing guides: Calibration and why you might be doing it wrong

Things you should and shouldn’t calibrate!

Good day, ladies and gentlemen! I’m Tom and today i want to talk about printer calibration and especially about the ways that you can screw things up.

Now, first of all, let me take a step back. The RepRap project and therefore pretty much the entire hobby-grade 3D printing market started out as a dive into the unknown. Nobody had done it before. It was basically the idea of “hey, this is something that is absolutely doable now that the patents are gone” without anybody really being afraid of the complexity of the entire project. So people started building these monstrosities of 3D printers that were the RepRap Darwin and later the Sells and the first Prusa Mendel, and as they ran into problems, they came up with impromptu solutions to them. Many of those solutions are still being passed around today, even though they were only really intended as kludges to get your print quality and your parts from unusable to really bad, compared to todays standards. But they still worked for that one case, so they must be good, right? Well, not so much.

One of these kludges that i would like to finally put to rest today is the process of “calibrating” your steps per millimeter setting for any axis other than the extruder. Now, the usual process for that involves printing a 10mm test cube and then adjusting the firmware’s X, Y and often also Z steps per millimeter values until that cube measures exactly 10 point zero zero millimeters, of course depending on how hard you press your calipers together. Now, if that doesn’t already sound wrong to you, well, let me bring in a car analogy. I know some people hate those, but you know, haters gonna hate.

Say, you drive a car. Most of us do. Any you notice that it always seems to pull to the left. So, # obviously, you adjust the steering, because that makes your car drive in a straight line again and you can tweak it until you barely even need to touch your steering wheel anymore when cruising along the highway. And that’s all fine and dandy, until you also notice that your car now tends to spin out in corners and completely misbehaves when it’s raining. Well, turns out adjusting the steering isn’t all that smart when you’re driving around with a flat tire.

And that’s exactly what you’re doing when trying to calibrate a 10mm cube to the exact right size. Not only is getting the filament diameter and extrusion multiplier right the much more straightforward option, but you’re also forgetting that our 3D printers aren’t really precision instruments that can even print that accurately. Just the mechanical part – ripples, slop, backlash, that kind of stuff – can easily throw you off by a tenth of a millimeter. And if you compensate for that one tenth of a millimeter that the 10mm cube is off, well, guess what. The next time you print something that is larger, say a hundred millimeters aka four inches, you’ll have overcompensated the entire thing and instead of a tenth of a millimeter, you’ll end up with an entire millimeter off the dimensions you were going for.

So what i’m saying here is basically: Use the ideal, calculated steps per millimeter. Timing belts and threaded rods, which are the things that usually drive the X, Y and Z axes of our printers, are made to pretty tight tolerances, since they need work with every other bit of equipment that uses the same profile. Which means, that the worst case scenario for the ideal steps per millimeter setting is it being off by much less than half a percent. To figure out the settings that you should ideally use, head over to Prusa’s calculator, which i’ve linked to in the video’s description. It’s awesome, i’ve linked to it before, but it’s just so, so good.

Of course, if you’re not using belts or are using an extremely large printer, it might actually be a valid idea to calibrate the steps per millimeter for X and Y since, like i said, half a percent is going to make a noticeable different in larger parts. I’ve uploaded some files to youmagine, which you can use to properly calibrate those settings without the results being skewed by for example the extrusion multiplier being off by a bit.

So now that we have that out of the way, what are the things about your printer that you actually need to empirically calibrate? / Well, it’s not much, actually. There is, of course, the extruder steps per millimeter setting and the extrusion multiplier, which i show you / here how to get those right. Then there is the entire topic of print speed, jerk and acceleration settings, but those usually just boil down to how much of your print quality you’re willing to give up for faster speeds. Protip: If you just can’t seem to get a print right, slow it down. Like, print at half the speed and you will usually end up with much a better print. And even if not, you get a better look at what else is going wrong. For the basic speed setup thing, i made a long version, which you can watch here.

Other than that, there really isn’t much that you need to calibrate per se. Of course, for many slicer settings there is only a certain range of values that makes sense, but i wouldn’t consider that printer calibration. Rather, it’s you who’s learning how far you can go with those settings. Give an experienced 3D printer operator any machines that’s mechanically sound and well-built and they will make it print extremely well. Also, most slicers nowadays will give you pretty darn nice prints without # any tweaking or / calibrating beyond simply mashing in the basic data about your printer and deciding how your part should be printed. There are, of course, the settings about print temperature, retracts and so on, but most of those either work pretty well with the default settings or need to be set depending of what filament you’re using. So, also, not really something that you need to calibrate # your printer for, since it’s rather a property of the filament.

So to sum things up: Don’t try too hard to calibrate every last bit of precision out of your printer. The technology and especially the software aka slicers is still under heavy development and is continuously being improved. For example, the newest experimental build of Slic3r, the one with the “three” in its name, so Slic-three-r, added a feature to compensate for fitting errors without messing everything else up. Basically, exactly what the cube calibration is trying to do, just with the difference that that is how it should actually be done. Right now, that feature is still in beta, but by the time you watch this video, it might already be in the stable builds.

So that’s it for today, i hope you learned something. If you did, please leave a like, that really helps me out – mostly because that means that i see people are actually enjoying my videos.

Also, please leave a comment if you have any questions left – i usually don’t reply to private messages because those really only help out a single person. If you have an issue you need help with, please at least post publicly in the comments or, preferably, in a forum or in the Google Plus 3D printing community. That way, that knowledge gets publicly documented and other people who run into the same problem can at least Google for it.

I’m probably going to do another Q&A soon, so any question might get answered there as well.

One more thing: I’m going to be at Hackaday Munich on Novemeber 13th, so if you’re there, too, come say hallo, i’ll probably be building and hacking stuff. Maybe not entirely 3D-printing related, but close.

So, as always, thanks for watching, see you next week.

The Prusa calculator

Youmagine files and instructions

RepRap Darwin image by Adrian Bowyer, at 0:31
RepRap Prusa image by Josef Prusa, at 0:34

 

Music by Kevin MacLeod


You can support me without spending a single penny!