Why your 3d printed stuff doesn't fit together and how to fix it! 

One setting that's kind of hidden by default and makes an enormous difference on Cura is slicing tolerance: it can be set to "exclusive", "inclusive", or "middle", which basically means wether the slicer will place your filament path entirely inside the geometry, outside the perimeter, or average its placement. By default it's set to "middle" which means it will always make your holes slightly too small, and your pegs slightly too thick. Edit: after a quick google search as a refresher I see I learned about that feature on this very channel! What were the odds haha

Thanks for this! I can't remember the number of times I've had this conversation with myself: "Ok, the screw is 3mm, and 3.5 for the hole -should- be ok" 🙂

Honestly i am an engineer and i think your channel does a magnificent job and explaining concepts and discussing important topics for people interested in 3D printing, it was a HUGE help on my 3D printing journey.

As a D&T teacher (retired) I used to do a short exercise with 11 year old students where they were asked to mark, cut and file a square of 3mm Acrylic to some given dimensions. When they presented their work as finished I tested it in a go / nogo gauge that had +/- 0.25mm sized holes in it. Generally first time none of the samples fitted but after a few tries they got the understanding of what tolerance was. When I fit 2 printed parts together I usually print the inside part 0.25mm smaller and then use fine wet and dry paper to get a perfect fit. Fettling, as my engineer friends would call it.

I would recommend using “Hole Horizontal Expansion” in Cura or the equivalent for Prusa. Scaling the object to fit causes all other features to scale. Hole Horizontal Expansion allows you to adjust holes with affecting the rest of the model.

As a mechanical engineer who has also been 3d printing for 7 years you did a great job explaining everything that’s necessary to this topic! some times people go off on tangents that aren’t really relatable and it’s interesting but some people may find it confusing. It’s good you didn’t.

In Prusa slicer, a modifier can be added with the setting "External Perimeters First", so you could apply that setting only to the sections of the model that need higher accuracy.

I found that if you have a "big" peg to go into a "big" hole (like making something out of two colors of plastic that you want to fit together) you can make the inside thing significantly smaller and then give it a wrap of teflon plumbing tape to hold it in place.

I took a 3D printing class in my MET degree as an elective, this info is spot on and in more detail about tolerances than that course. We had learned about the expansion coefficient of the layers, but were just given estimated tolerances for common materials, for typ layer heights for the machines we used in the lab. We had a demonstration print exactly like you made @5:19 , this allowed us to see how tight/loose certain tolerances were in real life. great video! earned a sub from this guy!

I would suggest adding a feature to the pegs to indicate their X-Y print orientation. That would allow the user to diagnose X vs Y axis fit issues if they are present. Good job.

When I need precise tolerances on something I usually mess with the horizontal expansion setting. I have a workflow for this where I duplicate the parts meant to fit together in CAD, but I only duplicate the pegs/hole sections of the two parts while being mindful of their orientations. I then print them, adjust the horizontal expansion (or slicing tolerance) and try again. Sometimes I have to iterate this two or three times to get the perfect clutch strength for what I want. Interference fits are counterintuitively more easy to do this for, assuming you don't care about the parts being removable in the future. I just print one of the parts with very few perimeter lines (usually two) and a higher infill density to make up for the weaker shell. The perimeters will stretch into the infill when pressed together, permanently deforming things and making a rigid joint.

I print always "perimeters first" up to 45° overhang. Not only because of dimensional accuracy but it also gives you a nicer surface finish as imperfections (from the usually faster printed) inner perimeters don't transfer so easily to the outer wall.

Wouldn't it be a great way to integrate PCBway as a sponsor by letting them print this gauge model in their different printing technologies and comparing them? Would love to see something like this.

Tolerance tests are always a good idea. I did that for an important part of mine not too long ago and figured it out for my settings at the time and learned a lot just from that.

This is a nice gauge and certainly useful. I think one thing that's missing though is how to account for these offsets and clearances. You briefly mentioned resizing the part, but that's not going to give you dimensional accuracy: it will not only make the holes larger but it will scale up the entire part, causing the outsides to be even larger too. Instead, you probably want a setting like horizontal expansion which will keep the overall scale of the part, but will subtract a bit from the size of each edge, like shaving a bit off of every face with sandpaper. You can use your gauge to determine how much too large your pegs and how too small your holes are, then put that amount in for horizontal expansion and it will correct for it, while maintaining the same overall size. You should never have to change your steps/mm or scale your part to make clearances!

If I'm needing exact tolerances I usually do a couple of test prints on a cut-down version of the part/tolerance in question. I've found that even if you think you have allowed for the issue you will still be surprised. To be honest given the relative low cost of modern 3D printers its actually surprising how good the tolerance/consistency of prints are so I'm not complaining!

I love your channel man it's highly underrated and should be a must for people that are getting into 3d printing

The wall print order was a mystery to me until I saw your terrific graphic and explanation! Thanks so much. Now I'm off to print that pegs-and-holes-tester!

Round holes are tricky, especially smaller ones. The extruded line will want to drag to the center of the hole a bit, depending on the diameter and speed (even with outer walls first). But I managed to do a perfect interference fit of a small square peg on the first shot - with zero extra tolerance, but chamfered the peg corners ~0.5mm as they aways overflow a bit on both the peg and the hole, even with a well tuned LA. It being PETG which is flexible helped and some force was needed, but it did fit.

Can't go wrong with GO/NO GO gauges especially when there are so many variables when 3D printing, excellent idea!

"Give up and use a circle." Yup, works for me. Nice informative and useful video, especially as I'm getting quite a few requests for much larger (3-4' tall) prints recently, thank you.

I always do the following: instead of a square or round solid peg, design it as a flexible two-pronged peg which can be compressed in one dimension. Think of it as two banana-shaped prongs, with their widest parts a bit wider than the hole. When inserted in the hole the two parts will bend slightly, and will provide good frictional fit with the other part. The larger you make the hole and banana-peg, the more friction you can produce.

What you mentioned applies to holes along the X/Y plane, facing into the Z (e.g. holes in top or bottom). However, holes in the sides suffer from yet other issues. First, a layer has a thickness and the wall at that layer will be basically straight - it won't follow a circular path. Therefore, the wall will be either too close or too wide as compared to the circle you are trying to make. By default, slicers will make the wall at the middle of whatever line/curve is crossing as you go up, which means it's both too close and too far. You can set it to print on the inside or outside instead, which will make it always too close or always too far, but neither is ideal. Really, you don't want round holes on anything but top/bottom if you need dimensional accuracy. In addition, the top of a hole has very shallow angles - approaching horizontal at the top. This amounts to a very sharp overhang. Slicers won't try to make this extremely shallow overhang. Instead, they will create a bridge across the top. Again, this means that either you have filament where you don't want it, or you are missing filament where you do want it. Another issue for top/bottom holes is the seam where the circle starts and ends. Again, you will run into issues at this seam. Either you will be missing a fit of filament or you will have some extra. The shape of the seam also depends on the direction that the extruder leaves when it joins up the seam - something you rarely have any direct control over. So you'll more than likely end up with bumps along the inside wall of your hole that make it too narrow. Getting a perfect fit for any shape is nearly impossible. The easiest way to get accurate and consistent round holes is to undersize them a bit in your design, then drill them out with a drill. We use the same technique in subtractive methods like milling because it suffers many of the same problems - you can't make perfectly square shapes and you can't make consistently round holes. When we need a perfect hole, we will drill (come directly down from above) rather than milling (circling around the perimeter). For any other shape, leave clearance, and provide relief in tight inside corners, such as using dog bones.

For me there is usually the option of manual scraping down a part, but in general if it really needs to fit, making a small test case really does help. It could be as simple as cutting away all the part except for the hole you want to test. then I print it using my usual guestimate and adjust where needed.

This just happened to me last night on a 5.5 hour print using PrusaSlicer's Cut feature. Had to go drill out the holes. So frustrating. Thanks for the video!

Regarding printing the outer wall first. I have that on by default on my profiles for about two years now. And on my test I even had better overhangs. Especialy curling up edges are a bit reduced that way. On various tests I never had a print with worse overhangs by enabling it, so i have it on by default and did that for the profiles of a friend as well who has no issues with it as well.

I’ve waited a long time for someone like you to make this video

By FAR my favorite printing channel! I wish this guy had a channel called “lost in history” where he did documentaries lol i swear i could listen to him talk about paint drying lol

Great content. And thanks for the gauge. I was thinking about creating one of these, but you just made it so much easier. I’m pretty new to 3D printing. But, when I use my Shaper Origin (handheld CNC), I can always create my male part first, the when I’m cutting the female, I start with an offset, try the fit, and add a few though each subsequent pass until I get a good fit. That gets me the answer for the project. It takes so much longer if you need to print the whole item before you know the answer. Thanks!

My favourite place for 3D printing knowledge. Thank you. Cheers J

For anything other than cylindrical hole fitment, I always put a radius/fillet on the edges of the male part. Works every time

Very informative video. Thank you for taking the time to educate those of us who needed it.

I do have a .6 mm nozzle and would love a .6mm nozzle version! I use .6 because Im not concerned about the lines or texture, but I am extremely focused on how smooth that line is, I can see all the septs in the motors and vibrations and such. I also like building real large models and general fixturing pieces

I regret that I have but one upvote to give to this video. This solved probably 90% of my dimensional accuracy problems immediately.

Loved your movie theatre analogy. Another thing to consider is the very common elephant's foot that occurs on the first layer and how it can affect clearances.

Most of the parts I print are cases I design myself for various sorts of circuit boards, and my preference is for an interference fit so I can just snap the lid on. With my printer (a heavily modified Ender 3) and preferred filament (Duramic PLA plus) I have found that a 0.2mm clearance works consistently. To avoid the issue with the corners, I always put either a 0.5 or 1mm chamfer on all external corners (ie the outside of a peg, not the inside of a hole) which both gives a better fit, and at least to me gives a part that looks more professional.

This is an interesting topic & a great video. I sometimes print parts for use at work (automated packaging machine builder) & have access to secondary processes (mill & lathe) to bore out holes accurately for fits. So the tolerancing is taken care of by those processes. If it's just a clearance hole for a screw to pass through, the machining will be very quick & easy generally. I sometimes trap nuts with parts I print, & have found going oversize by 0.2mm is generally enough for my prusa mk3, petg filament. I've noticed that when the part is mid-print, it is closer to the actual size, & then shrinks after it cools. I have to add the nuts mid-print while it's still on the heated bed so iv not needed more clearance than 0.2mm

No, you're spot-on. This whole fit issue has already been resolved for machine shop work, and while the tolerances used in milling and lathe work don't apply here, you could use that as a model, or framework. Btw, this is the first time I have heard anyone involved in 3D printing talk about fit and tolerances. Well done!

This is literally where I spend most of my time, and you'd think I'd have learned and adapted more than "Design, print, reprint with bigger holes when it doesn't fit" but I haven't - until now. @lostintech to my rescue - again, but not for the last time I'm sure.

nice gauge bro, will certainly give it a try. and my way of dealing with this is the lower flow at the outer wall. my usual setting is something like 2mm shell thickness with 0.7mm line width(from a 0.4mm nozzle) the 2 inner pass of wall is set at 112 flow(varies by materials, 112 is for pacf) while the outer most pass is like 92. the higher flow on the inside is to have higher extrusion pressure and presumably higher strength, and the outer most wall is just to give a smush so its mostly dimentionally correct. at the end of the day i find it very helpful to have a rubber mallet and a tiny file on hand, which can solve like 90% of fitting problem.

what is always ignored is the radius of a curve affects the how much excess material is trying to move into the available space , the smaller the radius the more material to the available space, try and print a 1mm hole and you probably won't get a hole at all.

Its so worth the work to get that interference fit. It's so satisfying when 2 parts snap together perfectly and holds there.

Thanks for the video and the explanation. I did print it and tested it out, it definetely is a great way of controling your dimensional accuracy. Liked and subbed!

If you're going for a friction fit, and making the design your self. Just add a small slope (2 degrees or something like that) to one of the 2 parts. Once the head of the peg is inside the hole, the rest can be pushed in with a bit of force. Unless the negative clearance is more then the material can compress.

The concept of right size middle hole is valid since it really shows the right size of the hole as it was designed. Before your test an extrusion e-steps test and tuning should be needed in order to know that the sizes are real. Since then the test is useful. You should also need to compensate the elephant foot tuning the horizontal expainsion setting since the first layer will be 2-3 decimals smaller than the rest of the hole.

Orca and some other slicers have a feature called "Precise wall" which addresses the wall expansion, and X-Y hole expansion & X-Y contour expansion, which expands/reduces the perimeters of walls and shapes

One other thing you need to keep in mind when building things that have thin walls between features is that you have to think about your line width and multiples of that. If you design a wall to be 1mm thick but have a 0.4mm nozzle, you are either going to print a 0.8mm thick wall (2 lines) or a 1.2mm thick wall (3 lines) depending on how your slicer feels like rounding things that day. Theoretically you can under or over extrude to make up the difference, but to my knowledge at least most slicers don't really change extrusion widths within the same layer. Maybe some slicers will just leave a gap as well if they can't fill it solid, but I'm not that well versed in all the different ones out there. I was having this problem when I way building a shaft adapter for a motor. It prioritized the outside diameter and was only one wall thick, so the inside diameter was just going to be whatever it was, not what it was designed to. It's kind of like printing inside vs outside walls first, but ultimately if you have any solid walls between parts you should do it in multiples of your line width.

after buying a new roll of filament you gotta print all tests out there to dial it in, but then you run out of filament and repeat that for the next roll.

This is fantastic and, as a note, should be wonderfully useful! Thanks!

I posted a remix of your clearance gauge on printables along with the source code (python build123d) that I used to make the object. I created an 11mm version for 0.6mm nozzle printers too!

The most important concerns when trying to fit parts together are squish and thermal expansion. Squish is related to nozzle diameter, and thermal expansion is related to the filament and extrusion temp. You must understand that a round peg will expand outward, and the receiving hole will expand inward - so you must compensate for overextrusion in both parts. I’ve found that 0.2 for a 4mm nozzle is a good starting point for overextrusion. Assuming a 4mm nozzle, a round peg of 10mm would need a hole of about 10.8mm. Remember you’re compensating for both parts in this hole and each part has two expanding directions of concern (4 faces * 0.2 = 0.8). There’s also the issue of fitting a round peg made of a non-printed material (steel, wood) into a round hole which is printed. In this case the hole only needs to compensate for its own expansion so it would be 10.4mm for a 10mm rod made of metal. Talking about percentages of increase or decrease is definitely not the way to approach this problem. A ten percent increase on a 10mm hole is 10.1mm. A ten percent increase on a 100mm hole is 110mm. Hopefully, you can see why this is not a good approach?

If you have trouble with overhangs when printing outer perimeters first set the outer line width to 0.6mm wich increases the overlap between the outer lines and therefore makes them stick better together. You can do this even with a 0.4mm nozzle, you loose a tinny bit of details but if you can live with that it’s almost like a cheat code. Perfect layerlines and even on fairly steep overhangs doable depending on your layerheight

This explained so much, and as an engineer you're words and smarts are very smarty, thank you sm

Tolerance is a specified deviation from a nominal value. Fit is a specified fitment based on it's intended purpose. Clearance is the space between two mating parts. From an Engineer, hope that helps everyone.

I looked into this video in order to improve some clearances in order to make screw style nuts and bolts for my own use, I'm surprised at how hard it seems to be for others to get an interference fit on their parts. I have an Ender Creality 3 and what I did was get 2 parts in blender, these were 2 halves of shells foe a mold, and basically booleaned one part out of the other. Then I fed this into the bog standard creality slicer I got with it and it worked perfectly. The parts need to be squeezed together with a fair bit of force but there's no sign of overstressing on the parts even after lots and lots of uses

Excellent explanation, right to the point.

Also another thing that I run into is "elephant's foot" If the circle fits but not the square, that might also be an sign of that issue.

Just got my first printer. Glad I found this already 👍

excellent explanation. very very helpful to my printing. thank you

Thanks for the tool. Often use SV06 and Prusa Mini together on the same project.

Interesting. I will have to look for that setting to see if it can improve my pegs. A tip I got from another video to reduce "ghosting" has helped with my key and hole prints, I uncheck "infill before walls." Besides helping the outside of the print I noticed the hole is cleaner and the pegs (ones I made, not from the slicer) almost always fit better because of what you described as the "armrest" example.

2:23 It also causes some upward squish. I have noticed that on machines with a single leadscrew, the nozzle will move up a little bit when stacking lines next to eachother. The main advantage of dual leadscrew IMO is the ability to overpower that upward force and keep laying down flat, consistent layers, even at 100% infill.

Great video I can't wait to print some of these and test them out! Also, for a really quick hack fix for getting square fitment without reprinting with more tolerance, giving it a quick pass with a hand de-burring Tool on the four corners can do the trick!

Extrusion multiplier calibration is very important towards this end.

I Always design with a chamfer for a square fit. Really helps those corners. Even just a tiny one. (essentially a rounded corner, but only 1 segment at 45 degrees)

What I've found helpful, in cura, is the horizontal hole offset found in wall setting.

Fantastic video, I've had so many issues with this phenomenon as a beginner and this was put simply enough for me to understand it : D

around 10-12 years ago I designed a file called "hole size test print" (on thingiverse and Printables) and it is by far the most used tool when I am designing for real world. It's a thin sheet with holes from 1mm to 25mm in 1mm increments and has the size next to each (use Prusaslicer chang color feature to make the numbers a different color) and when I need to print a part with a hole to fit something into I use it as a test. I have had 4 printers and on all 4 the outside diameter has always been very close to accurate (+/- 0.5mm) but it's inside holes that are not. I found my MK3S is about .2mm off inside diameters so that is what I use for a snug but not tight fit. Even though the test print uses round holes, you can still use it for squares and such if you have calipers, measure the square part you need to fit a hole with the caliper, then place the inside part of the caliper into the test holes and there you go you then have the square size hole to make.

Excellent info as always. One additional variable that will affect parts fit is "elephants foot", i.e., the slight bulging of a part where it contacts the build plate. I control this in Cura by setting the first layer horizontal expansion to around -0.12 mm. You'll have to adjust this number depending on the setting of your Z height setting. I get no elephants foot in my prints.

Very clear and informative and nice design

Very professional, learned a lot of industry content!

As a rule of thumb for parts that fit inside another part such as with an enclosure with a battery compartment for example, I design the battery cover 0.1mm smaller (all round). This friction fit almost always removes the need for fasteners/screws, for the everything else to fit with slight friction I use 0.2mm and 0.4mm for loose fitting...

that gauge is actually super useful. my brother asked me to print a small part for his bb pistol that broke, but i didn't even bother because not only is my printer wonky with filament stopping feeding (haven't been able to print anything for a couple years now because i can't be bothered), but also because i know i would never find the correct clearance even if i could actually model the damn thing he asked me to. this would help with that. it still won't work and is more work than it's worth, but meh, at least i can try.

I printed and measured outer and inner diameter on a test print. I used that to adjust horizontal expansion so that I know that the outer geometry is near the correct dimension and then I compensate the inner diameter in the part. Then I have a printed part with almost correct geometry. I use an Ultimaker S5 at work and for 0,4 nozzle I have - 0,057 and for 0,8 nozzle I have -0,114. Inner diameter for holes I add 0,3-0,5mm. Works great for me.

Nice idea! CNC Kitchen has something similar to test the correct diameter for their thread insets. But they added all axies so they have holes on the horizontal, 45° and 90° axis so you can test every possible angle. Maybe this would imrove your model so you can actually test how a peg for a 45° printed hole has to look like.

I always make the parts with 0.2mm clearance so they fit together especially if I have to print each part in different printers. If it still doesn't work, there's always the blow torch.

Very interesting and informative video, thank you.

Excellent video and file - thanks!

thanks for the good info I am having a good time watching the video.

Informative and well explained, as always! ;)

Really appreciated the accuurate alliteration.

In general, when designing parts that need to fit to something (either another part or an existing object), I would always recommend you make test prints of every critical dimension. I also find that a 0.2 mm tolerance is a good place to start. Finally, fillet or chamfer every sharp corner. In the horizontal plane, a 0.2 mm radius rounded corner is as good as it gets anyway, and baking it in means the printer won't "color outside the lines", so to speak.

For a person who mainly wanna makes props and helmets, but may advance to this type of stuff. I appreciate it a lot 😂

Dang this video took off! Nice work 😎

To avoid bulge in sharp outside corners, try fillets with radius = shell thickness. Or shell thickness x some constant close to 1 if you like to tweak. Draw the path of a sliding disk to see how that works. Generally, when modeling for precision I've been parameterizing shell thickness & layer thickness then using multiple shell overlap math in CAD, and sometimes turning off Arachne in the slicer to make features that will slice to exactly n strings of plastic.

I add or subtract.3 to .5 mm on the dimensions one one of the 2 parts to compensate and they always fit and have accuracy

ahhhhhh this is what I needed!!! thank you!! so many of my prints wouldnt fit with cura... hahah the struggle bus is real.

Something that helps when you need square parts to fit is putting a .2mm radius on the edges. I also make the hole .25mm to .35mm larger than the pin or mating part. Great video thanks.

Bahaha, I loved the trigger disclaimer at the beginning. Epic.

i just use 2 cura sittings called hole expansion and horizantal expansion they work well , so in design you dont need to worry about claearance

I found that using just a 0.5 mm filet helps not only get rid of the bulging corners, but also helps prevent elephants foot if done on the bottom edges.

I don't have linear advance, I just design smooth or beveled corners so that the corners don't stick out. Subscribed.

In my arsenal of settings used for printing with tolerance in Cura (each depending on situation) is: Flow (with tweaked flow for top and bottom)

Great video, in my experience one of the biggest problems for tolerance is elephants foot. Mine was caused by a low mesh levelling z-fade height, but also overextrusion on the first layer which I fixed by using first layer horizontal expansion setting in cura. I have also had issues with small amounts of warping on the first few layers which ruins tolerances, so I always use brims now. For more complex models slicing tolerance is important and if supports are used the z distance between supports and the model is a problem for small tolerances. However, tolerances would be much simpler in 3d printing if people who designed models in fusion 3d released the .f3d files with user parameters for tolerance, so people can customize the tolerance for the model after doing a test like yours. When doing parameters for complex models compound tolerance becomes another big issue, but often 3d printed things are simple enough for this to work well.

Great idea with calibration gauge. One thing to consider being fixed. The dots in the pegs affect geometry of the square facets as they sit to tight to them.

thank you so much for posting the stl to test for ourselves instead of making our own

i had to inlay some letters into a piece before and it was a pain in the butt to get them pressed in. as true as it is about corners, i just got a razor blade and trimmed it down, as for round prints, ive actually had quite a problem printing them, usually sanding them down works intill you get a size/shape where you cant get sandpaper in. again like the letters i had to inlay. if i were to do it again, sure i would try to fine tune everything, but i would probably still have to brute force them in

when I design things in fusion 360 that need to go together I use about a .12mm tolerance and it seems to work well for many things so far

Awesome video as always mate. I would really like one of your videos to cover linear advance😊

Another trick I found that works well for threaded parts is to use polishing compound on the threads. The paste type used for polishing car paint works well. If the threads are too tight, lubricate them with the polishing compound and work the threads in and out until you have smooth clearance, then wash off the compound.