These days youtubers (especially in crafting) have such advanced equipment that I always feel like I just picked up a stick and a rock in the field when I look at my tools.
I agree, this is how I install my threaded inserts especially ones with a shoulder. I also use the minimum clearance around the screw which will then pull the bottom of the threaded insert onto another shoulder in the part.
i would also like to know. nut rear pocket shear strength is actually off the charts since the failure mode is non shear related (this appears to be unique situation for thin materials with not much thickness above the nut). rear pocket insert would be interesting. it should give a slight boost from the lip of material above the insert. for thin materials the amount of boost you can get will be tiny (looking at the diagram on 10:45). if you have thicker materials i bet you can get more strength by just using a longer insert than trying to mount it from the back. also it is more troublesome to install it this way as you need a drill a pilot hole from the front and a larger blind hole from the back instead of just pushing it in from the top with a hot iron.
Possibly another point to consider is undersized rear pockets with nut pushed in with heat. Solves the issue of tolerance too. Perhaps buff the nut on the grinder for a fraction on all sides.
Wood inserts make its way in on the soft wood by pushing and ripping wood fibers, I'm guessing 3D prints would need some kind of thread before installing this types of wood inserts. there are some of those inserts are self-threaded, not sure how the plastic would handle it though.
Image linked below are the things I recently used when building a large workbench out of 70x30mm pine (for top and middle shelf frames + doubled up on legs) and 16mm MDF (for the 2 surfaces) As I've always hated reassembling furniture that just use wood screws and tear themselves apart I went with these instead, nice and secure and can easily separate major parts without wearing out joints. The internal threads are standard M6 metric, while the outer section are effectively cutting blades with a 2.5mm pitch. Drilled a 7mm hole, then with a 6mm hex key and plenty of downward pressure, drove these things in. media.bunnings.com.au/Product-800x800/dd209542-6cb2-4293-aa97-6cc71baf2bdf.jpg Good shit, can recommend 👍
At the scale that those huge wood threads are, there wouldn't be a difference in force between the insert thread moving in the print, and the raw printed threads failing. Source: tried it with ABS to fix a table
Insert research is DEFINITELY one of the things that I can use to improve my designs the most. Also, it teaches us what hardware to buy and what to throw out. Thanks!
@@alialtinel2960 Depends on the application, I assume. If the threads between two parts is the point of failure under a specific load, and you require more strength, it is worth improving that method of assembly. Like when joining wood, sometimes you use nails, sometimes screws, and sometimes glue, depending on the required strength.
I always use square nuts (Vierkantmutter) and they have never failed. A hex nut can round over its housing which is more difficult using a square one. Helicoil and any other sort of insert don't have enough meat in the much weaker plastics that are able to be printable. I would rather have a look at what is available for woodworking than for any metal repair. Regards, Etna.
@@chasenichole813 +1000. Weld nuts I always call them, but SO much hardware out there, I never understand why so much plastic stuff (molded or printed) uses what they can find at the local hardware store when they resort to metal. Don't gripe about say torque-out due to substrate: fix it!
I have used square nuts in other applications and yes it can withstand more rotational torque embedded into a part, especially if its a thicker walled square nut compared to the thread diameter size, which they do make
@@Christian-cz9bu Prusa printers use square nuts if the're being inserted in pockets. They use hex nuts only when the bolt passes through the entire part.
I recently (this week) built a Prusa MK3S. I think the way they use nut inserts makes a lot of sense. They use square nuts when the nut is inside a part.
One other insert to test is called a Keensert, preferably the heavy duty ones (they are thicker). They have larger coarser outer threads then a Helicoil and have metal side stakes you hammer in to prevent rotational spinning. I use them all the time in heavy duty applications at work, and by far are the strongest "thread inserts" you can use
Hi Stefan, I'm a chemical engineering student who just finished a course in rheology for non-newtonion fluidic flow, focusing on polymers. You said to reach out if we knew something about polymers, and I can say with confidence that it's all just magic and made-up math to torture students (I have no idea how I passed that course - I barely know linear algebra so we weren't even speaking the same language). Always happy to help! Keep up the great videos!
Even metals and isotropic materials are sort of hellish to predict already, going with completely anisotropic stuff like 3D printed parts, on polymers at that, definitely brings hell to the equation. I'm not so sure we can really predict the parameters through the Von Misses criteria as well for anisotropic stuff, didn't yet have classes on that (I think that's even an optional for me, the normal ones are mandatory... Mechanical engineering sure shares hell with you chemical engineering guys).
You perfectly explained why your compressive load calcs were low, to do calculations about the area that accepts stress in bolted connections, look up 'frustrum' calculations for bolted joints. Thanks for the vid!
This seems like the kind of thing where a washer would make a huge difference. Have you done a video on simply putting a washer between the nut and the printed part?
I second that. Imagine using the slot method and adding another wider slot for a washer. Basically a slot shaped like a T. That would spread the force over a bigger area. I think this could win over everything else while still being basically the cheapest method. Needs a little bit more space of course.
For ease of design simply having a washer sized depression on the end of your through hole at a depth to accommodate the nut would be easiest. But I agree, the washer would increase the surface area significantly (assuming it's larger than the nut) and would yield best results.
Detailed and very informative as usual! I couldn't help wanting wood, sheet metal, and proper plastic screws in the comparison. Yes, I know you can't test everything all the time, and yes, I get that at least some of the point was a "reusable" fastener, but I use sheet metal "tapping" screws and wood screws all the time and they can cycle quite a few times before they strip PLA.
Man, talk about timing. I was just needing this information and had ordered some of the nuts for a task when my direct drive arrives for the Ender 5 Pro. THANK YOU
Thanks for this data, Stefan. I used to design a lot of aluminum mounting plates at a former job. Helicoils were mainly useful for resistance against galling due to repeated use of a fastener in the hole. As you saw in your tests, they don't really add strength. If we wanted a bit more strength, we would use Keenserts which are kind of like a combination of the Helicoil and your threaded insert that you melted in. The Keenserts go in like a Helicoil but use a bigger threaded hole and then provide the smaller hole for the fastener that you want to use. Bigger threads on the outside equate to a marginal (but still not huge) increase in pull-out strength. They are also a fair bit pricier and availability of small sizes like M3 might be limited likely because the smaller the diameter, the less improvement there will be so there is less demand.
I prefer Timeserts over anything else when doing thread repairs on motorcycles. But plastics would fail during an attempted install because install tool spreads out the sert at the end of it.
Can you test the nuts by treating them as heated insert? make the hole tighter. Also you can try flanged nuts or square nuts (as already mentioned) Thank you for your great work 😀
@@pacomb A cheap soldering iron works too. You want the nut to warm slowly enough that it just reaches the needed temperature goes in place then cools fairly quickly.
You can also easily pull nuts into the hole using a screw. It lets you use very tight tolerances and ensures the nut is aligned perfectly with the screw hole.
Use heat set inserts on the opposite side just as you did for the hexnut. This yielded fantastic results for me. Also for the hexnuts, make slanted walls, the nut will pull itself into the material on first use, and make it not fall out after.
there is a "T" nut for wood that could add strength compared to other options. It has spurs that can be heated in from the back to add keying for twist and a large surface area for pull out strength
Just yesterday my dad gave me about 20-30 kg of Helicoil inserts and tools. Imagine, what is the first video, I've found, looking for implementation? Yes, it's Yours. Thanks a lot!
love that this video allows me to continue doing things the lazy way, never buying threaded inserts. need it to be strong? put nut in hole. strength not a priority? just screw into a slightly undersized hole. was doing it this way for convenience, and it is so nice to know im not just saving time, im saving money and not compromising my build strength. at least not a significant amount.
For embedded nuts, using half-thickness nuts (which are readily available) would give you some extra depth to load in shear, and useful extra strength if you have little space over the nut. Also, pausing the print at the point the nut can be added lets you make captive nuts at the cost of slightly more annoying printing.
Immidiately ordered threaded inserts after the last video. I have a pretty big 3d printed case for the hardware of my 3d printer. Before it was pretty wobbly because most screws were worn out. Now its rock solid. So thank you very much ;) I was not aware of these inserts before!
A drop of some kind of adhesive to hold the nut in place is my favorite solution. Yes, I could fix the tolerances to where I wouldn't need glue when initially placing them, but even a little wear makes that unreliable, and I would need to do that for every kind of filament. Glue is cheap, especially when it doesn't need to be good and is only used by the drop. A temporary shim would also work. Some kind of press-fit nut holder design might work, but its long-term durability would depend on its orientation when printing.
I was going to recommend a cyanoacrylate adhesive as well. It also sounds like using a washer to spread the force over a greater surface area might also be beneficial.
An excellent video. Some notes: 1. square nuts will resist spinout better than hex nuts 2. I think you will find coarser pitch threads will better resist pullout in plastic 3. externally threaded inserts, will outperform Helicoils, due to larger and coarser external threads and increased shear diameter
One important one I feel you missed, was when insert the threaded insert from the opposite side of the bolt insertion. The pull out test as well as torque would probably be even higher as then it also takes in some compression forces.
I have been following your channel for a while now. In fact, one od your videos from 2020 is what sparked my investigation and about 1.5 years of lab work for testing the ability to sanitize and clean 3d printed parts for food and medical applications. I finally got the paper published too. So heres to you my friend:)
Hi Stefan, as always I love seeing actual engineering data and fundamentals being applied to DIY AM. I wanted to comment on the embedded nut technique and the cone shaped deformation. This case is not like the others because it is not longer a threaded joint and can be treated like a bolted joint where the nut acts as a washer, which is why you see the frustum shaped crater where the nut pulled out. Awesome video as always, keep up the great work!
I've been using PRINTED threads, not straight into material but actual printed threads for years now. No extra tolerances or anything, straight from fusion360 thread function. Those seem to hold perfectly, typically have failure elsewhere in the part, not the threads. Typically use M3, M5 and M6 and on PLA. Atleast the first time you use the thread the bolt can heat up significantly, i would assume this causes some level of annealing on the material as well. Due to the 3d printed thread being ever so slightly undersized it also works as threadlock more or less :)
Interesting thought about the annealing. I wonder if when using holes without printed threads if there would be any gain by inserting a much long bolt than is needed, heating it with something as basic as a cigarette lighter, allowing it to cool, then replacing it with the correct length bolt. Also makes me wonder if the heat used when installing threaded inserts is annealing the plastic and helping with torque out.
Just yesterday my dad gave me about 20-30 kg of Helicoil inserts and tools. Imagine, what is the first video, Iэму found, looking for implementation? Yes, it's Yours. Thanks a lot!
One thing that could be tested are the different kinds of T nuts: t Nuts for aluminum rails (like the one in 3D printers / cnc with 20*20 aluminium rails), T Nuts for wood (those we normally hammer in with the tooths) and thin T Nuts that are flat rectangulare piece of metal with a threaded hole in the middle) and also square nuts
I use hex nuts in a hex cavity that has a draft from the surface to the circular area slightly larger than the diameter of a matching hex socket head screw ('lofted' in my CAD system). When tightened the nut slightly deforms the cavity compressing the material around it. The nut will no longer "fall" out. It has the added advantage of having recessed cavities where either the screw head or the nut can be used. .... very handy for symmetrical part designs (i.e. split circular clamps).
Excellent useful video. My feelings are that you could have gone further by making a second thicker sample for the nut and the printed thread because their strength is geometry dependent -but we can calculate the improvement. My preferred fastener for prints is a wood screw with deep course threads -people refer to them as "deck screws" in the US. "Drywall screws" are very similar but lack rust protection. If you have plenty of depth these screws can achieve impressive strength. As an automotive Engineer I was pitched a bolt for plastics that outperformed machine bolts with inserts. The impressive thing was you could strip them abusively and if you let them cool they still fastened and released with minimal degradation (in nylon 6/6). The design was VERY similar to a deck screw with design rules on the hole. The only negative is that it takes lot of energy to screw (torque and turns) and it is not so obvious when to stop. if you don't have much depth, T-nuts are the way to go. I don't see the point of inserts.
You can make a frame for your testing machine to test shear strength. Take a "rigid" square frame with hinges at the corners. Rotate it so it is a diamond shape and pull the top and bottom corners apart. You need teeth on the frame and interlocking teeth on the specimen to transfer the shear to the specimen. The shear stress would be the tension force divided by (1.4142 x length of side of square x thickness of specimen). I would be interested in seeing the results. The main assumption is that the compressive force in the horizontal direction generated by the frame would is equal to the tension force. Otherwise the principle stresses would not be equal and the peak shear stress would be some combination of tension and shear. A pure tension specimen (as you tested) has a peak shear of 0.5xT/A (occurring 45 degrees to the axis of the specimen) in combination with a tensile stress of 0.5T/A. The idea is to get rid of the tensile component at peak shear stress when you use the frame. This way, you get a pure shear response.
We recently did pullout certification for 1/4-20 brass threaded brass thread inserts in Type-1 PVC. After doing the required staged static loads at 50, 100, 150 and 250 for 30 seconds each we took each of the 6 sample pieces to failure. To our surprise the average pullout failure averaged 1800 pounds across all six samples with the highest topping out at 2300 pounds. In all cases the failute was in the PVC and the thread insert and bolt were undamaged.
A method I like to use is to print captive nuts. In your design program you can make a void in the part and pause the print when it reaches the correct height, move the print head over, insert a nut and then continue the print therefor encapsulating the nut inside the print. It works surprisingly well.
You can insert the threaded inserts by pre-heating the printer nozzle and using the manual Z axis control to push the insert in place with the nozzle. This is a GREAT technique because the temperature of the nozzle is perfect for melting the plastic - and it presses on the insert perfectly straight. That turns those inserts from being a pain to being by *FAR* the easiest approach.
Even after buying heat set inserts, nut pockets are still my go-to for reusable threaded connections. They're quicker to install since I don't need to get my soldering iron out, (usually) not too hard to add in to my designs, and you can even reuse the nuts from other parts. For the problem of loose nuts (hehe) I just put a small drop of CA glue on the side of the nut. It's just enough retention force to hold them in, and might even improve the torque-out strength.
Something I did a while back was I set a nut into a square, then pressed that square into the print and printed over the top of it to completely embed the nut inside
Helicoils, oh how I hate those. Had a client who refurbished an older machine, he had to "fix" over 20 threads with those due to rust and simple wear, they constantly failed so he asked me if I could fix them somehow. I ended up drilling out all those holes, beveling their edges, and then weld in steel tubes to then drill and tap them those to the correct size. But on topic, thanks for these tests. I'm now rather glad that I started building an insert station for my Ruthex inserts, but I might try some more bottom pockets with added geometry if the parts allow for it space-wise. I wonder how it would perform if I add some geometry to add epoxy.
You inserted the heat seated nuts from the same side as the pull direction. For a lot of design that's required, if you don't have access to the back side. But I would bet it would perform equal to or better than the nut if it was seated on the far side. Heat seating would also solve the problem of nuts falling out. Design your hex recess for the back side just a touch undersized and heat seat (or just press) the nut in place. Thanks for this. Now I know how to improve my design connections.
very nice comparison! I'd recommend sticking to physical force measurement in Newtons again and adding a 2nd scale with kgf 😅 Remform II screws would be also interesting to investigate since their shape is optimized for polymers. They are often used in IM components and available for different ranges of elastic modulus (esp. unreinforced/filled and fiber reinforced).
Great work Stefan! Worth mentioning that wire inserts are available in different lengths up to at least 4.5x thread diameter. This should improve their performance on tear out and torque.
Stress distribution between turns of thread is uneven, the highest amont of load is taken by 3-4 turns, at least for steel nut and bolt, it may differ for polymer, but it still is pointless to lift the thread lengrh so much.
Thanks so much for your great efforts and information. As a person who works at a mechanical test lab and who enjoys 3D printing I really appreciate all of the work that you do.
I really appreciate your attention to detail and the stress analysis. The criteria you discussed assume very ideal loading conditions and failure modes, so while they do a good job of comparing the thread methods to each other in a relative sense (as you pointed out at 10:56), they don't work so well in the real world. A few other factors that I would enjoy seeing you discuss in the future are stress concentrations (definitely present in the internal sharp corners of the side pocket/nut combo!), failure via crack propagation (which is likely how the two embedded nut samples failed), and finite element analysis (you may be able to get fairly accurate theoretical results!). As always, great video and I look forward to your next one.
Great video continues like this! For a future video, it would be interesting to test the screws for plastics (screws with a larger pitch). To give a reference, for example, "Brossard PT K30x20". Personally since I tried, I use more than screws for plastic.
I take apart old laser and inkjet printers for parts and have a nice collection of various screws for plastic now. In many cases, I prefer them to machine screws - regardless if the machine screws are anchored bare, with nuts, or with threaded inserts. They don't chew up hole perimeters as much as bare machine screws do. I'd really like to see how their holding strength compares via Stefan's meticulous testing though.
Great work and video. Seems like square nuts would have fared slightly better than hex, and square nuts are very popular. The lower cost melt-in nuts are very popular around here due to the cost factor. One technique I often use is to melt the nut into the far side rather than the near side which may help in some cases.
I just got my first 3d printer. Started learning fusion 360 and I’m working on simple projects to get into the game! You’re videos are very informative for a newbie!
Another interesting idea is that with embedded nuts, it can still be possible to embed a washer underneath as well. I would check that out since the nut often failed in compression a washer to spread the force would probably increase it's strength significantly.
There are inserts with large lips that provide more pulling force that a nut. Sometime you can’t use them depending of your design but in the case of your plate, you could have use it and it’s the best option because it combines both types of resistance.
Greight video, thanks. Much useful information there, but there's one test you missed - threaded insert fitted underside, I use this method frequently and find it to be the best way to use threaded inserts.
Another great scientific test Stephan thank you. Threaded nut and bolt joints are great for parts that you want to be able to take apart again but, for permanent fixing, self tapping or deep threaded screws are the way to go. There's a reason they are used in plastic products. The deep thread cuts into the plastic really well and doesn't pull out like fine metric threads. I would love to see a test on these wood/chipboard/self tapping screws. I'd wager that they would beat all of these at least in the pull test.
This is great content! In my mind’s eye I see a combination of a bottom pocket for the nut and a side pocket for a washer being very strong. I imagine for accurate testing the test block would need to be larger.
I have a few parts I've made with fully captive nuts that require pausing printing at the correct layer height, placing the nut, and then resuming printing to trap them in the part. These I'd expect to have a similar strength to press-in-from-the-bottom parts, proportional to the material thickness "above" them. If you need more pullout strength in those the strength is proportional to the fastener's flat surface area. "Flanged nuts" are stronger than "square nuts" are stronger than "hex nuts".
I'm a 3D printing enthusiast, and Stefan's videos present clear and direct ways for me to make my prints far better. I'm also an engineer, so I'm most fascinated by how Stefan's experiments are designed to obtain the desired data using what are often very simple or rough tools, with the initial data analysis checking the experimental technique itself before being trusted as a basis for future engineering decisions in the problem domain. That's part of the field called "Design of Experiments", something I'd love to see covered as it's own topic in future videos. I'm a Systems Engineer taking a new integrated system through FCC and EU "Class B" RF emissions certification testing on Tuesday, with susceptibility and safety testing in a month or so. A key aspect is to minimize the time spent at expensive test facilities (large RF anechoic chambers) while still guaranteeing the needed data will be gathered with high quality and full coverage. This is done by first "reducing the test surface", ensuring nothing is tested other than what must be tested. Yet also while simultaneously ensuring a level of testing that extends just beyond "full coverage" to ensure edge-effects and corner-cases are covered. Every good experiment must have places where, in advance, it is known to fail, so we know what a failure looks like before focusing on success within critical areas. Far too often an experiment may be lauded as a success for returning "no negative results", when in reality it is a failure because it was so poorly designed (physically, procedurally or analytically) it was unable to capture any negative results! Our prior RF emission certification efforts consumed up to 12 hours of facility time, with a team of 3-4 engineers and consultants. I've trimmed that to just one hour with a team of one (me), largely by optimizing the physical aspects of getting equipment in and out of the test chamber, and more rapidly getting it into test configuration. Still, I reserved a 4 hour block. Mainly to reduce management worries that I may not have anticipated all difficulties. But also partly because these facilities are huge fun to visit, and I have a few "special" experiments I'd like to run. I'll be using every minute of that 4 hour test window! Seeing how Stefan constructs his "attack strategy" for physical experiments helps me take a fresh look at my own processes for testing and experimentation in very different areas.
Before the video even starts I'll say that helicoils are relatively weak and rely on the surrounding metal to support them. Better would be to use flanged threaded plugs. So the flange is bigger than the formed bore, almost like having a nut on the side opposite to the bolt. But you might also used double threaded plugs, with a bigger thread on the outside and your required thread through the centre.
@steffan - great video as always ! Ty - I’d love to see plastite screws tested - Little known - and relatively recent technology, but amazing engineering, “thread forming screws” are designed for quality assembly of plastic components - without having to use nuts, inserts, or taps, but have many clever features to ensure high pullout force and the ability to be used In the same hole again and again - unlike self tapping screws which are awful in plastic. The threads are tri lobular in plan - which reduces hoop stresses when inserted into plastic - as they move material into the lobes - rather than compressing it outwards as in a circular thread form, particularly useful for brittle plastics which otherwise are likely to crack during a “ self tapping” screw insertion due to hoop stresses. Would love to see these compared to self tappers and nuts and bolts. I am using them for a production parts in fdm with the holes printed - so it eliminates drilling / reaming/ needing nuts nor “nut assembly” operations. Soooo much easier .
Source on below.... I work for a fastener distribution company. The 3-lobe body feature you describe allows displaced material to cold flow back into relief areas which minimises radial stress and reduces the possibility of boss failure. Plastite is a trademark like Helicoil is. Many manufacturers sell them under generic descriptions. They are collectively call thread forming screws for plastics. There are many, many types of screws specifically designed for plastics, all with different and specific features. They key points in selecting the best one for your application are the materials you are screwing them into and the boss design (if the hole is not just in a block body). These are not really little known or new technologies, I've been selling these for over 15 years business to business.
Craig Struthers thank you for your thoughtful reply - I take your point that thread forming technology isn’t new in the sense of “this year” I think I was meaning in the sweep of “since we started using screws for mass production” sense, and as you say there are many thread forms competing in that market sector, I feel I should have worded my comment more carefully, given clearly as an expert you read it and correctly corrected me. To specialists of course they are known - but I think not to many hobbyist makers and I feel should be more commonly used / known about by diy / hobbyists.
Bottom inserted nuts can be heat-seated too. But it's usually not necessary. I usually print the hex socket a tad smaller than the nut and press it in with pliers or pull the nut in with a bolt. Holds tight and doesn't fall out.
on smaller parts, i like to model a cavity for a metal nut. then i use the pause at layer feature in cura to momentarily pause the print so i can drop in my nut and resume printing over top of it.
I just simply pause the print and drop a nut into a slightly over sized pocket, this allows me to place a nut at any depth, this strengthens the model greatly
Very interesting. I am looking at using the press in inserts on a project at work, but using a flanged insert and pressed in from the back side of the part in a hole sized for the length of the insert. That way I am combining the press in insert with the nut on the bottom version. I tested this with some 4-40 inserts in CF nylon with excellent results. Parts printed on a Markforged Onyx One printer.
I would highly recommend people have a traditional tap and die kit on hand and use that to repair the damaged threads. You don’t need to drill out a new hole with them, you just thread the cutter through.
I have investigated that the best way is using big washers on both sides of the 3D-print. This can withstand the most force, as the area is maximised. The torque is not resisted by the plastic but by a counter tool. This works perfectly.
if you use SQUARE NUTS instead of terrible hex nuts, you should get a lot better results, and you can even slightly undersize the pocket for them so they stay where they are during assembly. this is a problem specifically for M4 or smaller nuts, because there's hardly any flat on the hex nuts. Essentially the nuts are wedges that like to bury into the material. There's another problem that is the nut being able to spin, since there isn't very much plastic to support the corners of the hex and prevent it from spinning. Square nuts are cheap, easy fixes to both of these issues.
I do have 2 things that you could test: Universal layer height. Where the Layer height changes based on the design {Beta/experimental feature of CURA} Gradient infill. {The infill changes depending on the stress the part will experience}
If mounting a nut from the bottom is "fair", then what about using the cheap "3 cent" threaded inserts from the last video installed from the bottom and leaving a similar amount of plastic between the insert and the bolt head. They were as good as the more expensive inserts in torque out loads, but were significantly worse at pull out loads. Installing them so that torque out is physically prevented by the plastic, as is the case with bottom pocket nut, would in theory dramatically increase their torque out resistance compared to top mount inserts, and make them the best all around insert (at least in situations where a bottom mount is practical).
Suggested test for bottom fit nut, design a space inside the plastic for a washer, pause print to insert washer and finish print with hex shape for nut..... I hope my explanation was clear enough
Helicoils can be a life saver. Also many may think if using a .375 dia bolt anything deeper than .375 would be a waste but they forget that drill size for taps is 75% smaller than tap dia so you would need 25% deeper hole for full strength theoretically.
Another brilliant video from CNC Kitchen - I have a part I just designed with a side-pocket for the nut - but there is no reason why it cannot be a nut inserted from the back - so I am going to make a minor redesign to this part and make this change - which should reduce the size of the part as well as strengthening it. :-)
As other commenters have already mentioned you should also try square nuts, I have switched to using them for pocket nuts years ago and not only do they seem to support higher loads, they are also easier to insert and to design pockets for. Another option you should test are threaded inserts for wood, which have much coarser outer threads and might work better in plastic, which really is closer to wood than to metal in material properties. For comparison you should also try putting a nut with a washer under it on the back of the part. Another test I would like to see is a comparison of different sized threads, from M2 to M6, with threaded inserts as well as nuts and directly in plastic.
You can also embed nuts by printing around them. In cura you can set pause at height or layer. This way you can put the nut in when the print stops and then resume the print.
Large washer as a side insert, as far back as possible? Ideally just in front of a bottom inserted nut. The washer will spread the load much wider, and then a standard sized hex nut can still be used. Both the washer and nut could be heat inserted to stop them falling out, as others have noted. (Start by applying heat to the washer from the side, and then move to heating it through the hole the nut will be inserted into, if the heat applied from the side isn't enough.)
In response to the reply about adding the insert to the bottom and putting the screw to the top if you add the screw to the bottom of the insert then heat and push in it pushes the plastic up and around the tiny notches on the insert making it a little stronger, I have noticed
Next, you could begin to do some joint analysis. It looked like you did not apply any preload to the bolts when testing them in your fixture. The ability of the heat inserts to carry more tightening torque would let them establish a higher preload which would help with preventing joint separation. Also, I like the idea that some others had about using the heat inserts from the back side. Great job!
Oh and one more thing to try are the Böllhoff Amtec inserts, especially the type that expands slightly when screwing in a screw, similar to a wall dowel.
Try a fanged nut with an exposed top face, the flange bearing the pullout force, the hex in the top surface bearing the torque-out. Also a shouldered bolt with a traditional nut. The face of the nut bearing the pullout, and the shoulder flush with the nut preventing it being pulled through the plastic when testing torque-out.
I like to use T-nuts that are designed for wood. You can get them almost any length and the barbs can be set into the plastic with heat so they don't fall out easily.
If you can reach the back of the part, T Nut inserts are tough to beat, also. They don't spin as easily as a nut, are easy to install (if you can handle the small flange on the surface), and you can readily pick up a 50 pack of m3x6 zinc T nut inserts on Ebay for about $4. Might be worth a shot.
The embedded nut also benefits from confinement at the bearing plane. Due to the plastics being very flowable I would assume this would result in a strange poissons ratio... or perhaps a variable poissons ratio based on strain.
Sometimes low cost, ease of assembly, and just using what you have on hand are more important than mechanical strength. Some additional methods and tips - enough for several more videos - are: 1. Use M4 or larger screws and printed threads (e.g. created with Fusion 360). Larger threaded holes print with more detail and grip screws better. 1/4-20 threads work well if 1/4-20 screws aren't ridiculously large for the application. 2. Use square nuts (if you can find them) in side pockets. If they fall out, use a small piece of tape over the slot. 3. If a hex nut keeps falling out of a hex-shaped hole, add a small drop of your favorite glue. 4. Use standard self-tapping screws. 5. Use self-tapping screws designed for plastic. Examples include EJOT ECO PT and DELTA PT and various Screwerk screws (all made in Germany, of course) or Plastfast screws (as sold by E3D).
I recently add a pocket, a bit higher than the nut, in my design and point prusa slicer to the end of the pocket. Now I add a color change where the print paused to set in ordinary nuts. After reload the filament the printer continues where it stops before. The nut gets embedded and fixed strong inside. Also prusa slicer (newest version) shows you the expected time to the "color change" an go back to the shop.
One way to keep nuts from dropping out is to use a dab of super glue. Ive also used silicon gasket maker as a makeshift glue when I needed to make things water tight. These are all mostly metal and wood working. Personally if I'm looking to clamp 2 pieces together I usually put a wide washer under the bolt head and under the nut. It spreads out weight distribution over a wider surface and can hold under more load. Most of the time the parts Crack break or shatter before the bolts will.
It's probably worth noting that infill % and style as well as wall thickness will probably have a sizeable impact on some of the applications, the one advantage I can see to using the threaded inserts that are primarily designed for use in wood is that with their more coarse outer threads they would bite into the infill in much the same way as they do with wood fibers when installed. Additionally it's worth noting that with the side inserted nuts the distance between the surface of the 3d print that another piece is being mounted on has a fairly decent impact on the point at which it fails and pulls through the part, the further away from that surface it's placed the more it behaves like a bottom inserted nut. Additionally you may want to consider testing square nuts in both of those insert methods they shouldn't dig into the plastic as much when turning due to the larger flat surfaces on the sides (of course in some hardware stores/home improvement centers you can probably find something similar alread produced for use in various home projects like electrical wiring install/upgrade or plumbing install/upgrade). A step above square nuts might be to take a length of flat stock like say 4 to 6mm thick and drill and tap a whole in it then insert it into a hole or a cutout in the bottom or side to see how well that performs compared to a flat square nut, I'd think that it would perform better in almost all aspects because of the increased surface area on the top of the piece of flat stock as compared to even a square nut when it comes to pulling through the part, and as well as rotating inside the pocket it is placed in.
You can always use the insert like you would use an embedded nut from the bottem side of your part(if you can use a through hole in your part). So you would have some extra material like you would have with the embedded nut from the bottom.
My favorite way by far is to fully embed a nut into the print by having the printer stop, raise the printhead and wait for me to insert the nuts. Once inserted I tell the printer to continue and it will just print right over them, resulting in completely embedded nuts. This way you can place them anywhere and they will never fall out (something that drives me nuts lol). I use hex nuts if they lay in the horizontal plane and square nuts if standing upright. I need to edit the gcode manually to do that but it's not that hard. When slicing with slic3r you can have it add comments that tell you in the gcode when the next layer starts at what height so I know where to put the stop and wait gcode. I make the hex hole a tad undersized so I have to put the nut at the end of on a long bolt and heat it a bit on the stove to push them in. It's not neccesary maybe but it makes them nice and tight and keeps them from rattling around. Lets say at Z=6.320 the layer will start that will print over the top of the nut. Then this is what I insert into the gcode: ; INSERTED CODE G1 E-5.0 Z80 F2000 ; lift Z to 80mm and retract 5mm M18 S900 ; Steppers inactivity timeout 15 min M0 Click to continue G1 E5.0 Z6.320 F2000 ; bring Z back and and unretract. M18 S60 ; Steppers inactivity timeout 1 min ; END INSERTED CODE
Vibrations can matter a lot too in these tests! I found out they can really change the results. I always also get my cheap brass inserts pulled out by screwing two parts together. The pulling created by the screwing can easily take the insert out. I wish you did a test where you used a screwdriver to pull the insert out! What I mean is basically a twist and pull at the same time
Unless I missed something - one thing that I felt was missing here is using a regular nut the same way as a heat-insert. This would probably be fairly weak as the sides are smooth, but you could just over-bury it a bit to compensate some of that - and it would probably be an "ok" solution anyway for structures that don't really need to be very strong anyway - which is probably the majority of 3D printed item anyway. You would probably need to print an undersized nut-shaped hole for this to not make a mess that you have to scrape off after, but that's easy enough. It has the benefit of not needing side or back access - and while not every workshop has threaded heat-inserts lying around, offs are if you own a 3D printer you probably have some nuts of a usable size lying around somewhere :)
