Hi. Interesting video. Question: Ever printed with SBS? It can be smoothed with D'limonine extracted from the skin of citrus fruits. It is also used to recycle polystyrine used in food take aways 100%. Your answer would really interest me. We have a manufacturer in South Africa called Fil X that makes SBS filament. Videos on RUclips refere to imports from Russia years ago. Love your channel. Always learning something new.
Hi mate. What is the point in that? I believe 3d printing should be used to make prototypes and parts for hobbyists to use. If you want strong parts and to print multiple items, wouldn't it be better to injection mould them?
Now that I think about it you might use a wire brush of some sort to texture the layers. Using the tool head changer or doing a quick scratch manually sounds simple.
@@mike-ology22 Costs mate, costs. I have a sole proprietor company with a very limited budged. To have injection moulds made for every part, that changes every 6 months as I learn new methods and designs would not be cost effective at all. So my 6 3D printers have to cary the weight of producing parts for the products that I sell. My 6 printers are about R5000 US in total. To have just 1 part made into an injection mould would cost double that due to complicated structures.
Failures are just as important, if not more important, than successes. Props to both you and Tom for both putting the effort into publishing "stuff that didn't work". 👍
I was just having this thought yesterday. I've been looking into hydroponics lately and have so many ideas and I think "I wonder if anyone has tried that". What a time saver it is to know what doesn't work ahead of time. Of course there is the tried and true methods, but I always want bigger, better, faster, and easier.
I had an idea for trying to improve layer strength: staggered layers. Basically if you were printing a 2 perimeter hollow box print the outer layer, raise up half the layer height, print the inner, up half the layer height again, print outer and so on. I tried doing it hand editing gcode and successfully printed a 2mm high test to see if it was possible without dragging the piece off of the bed. I never went back to make a piece big enough to test the strength due to how long it took to edit the gcode for even that small test. I'm not a coder so I couldn't easily think of a way to automate it.
Have you seen the "Alternate Extra Wall" feature on the Cura slicer? It looks like it does a similar pattern. I haven't tried it but maybe it will speed up the process?
@@archercureton3738 it's not quite the same. It improves bonding to infill, but it still prints on the same plane as the rest of the print. I have never looked at the Cura code, but it may be possible to add a z shift to the extra wall more easily than writing a whole new pass for the slicer.
I tried the same stuff a few years back with epoxy, what I observed was the liquid would be vaporized by the heat instead of helping bond layers together. Meantime, the hot plastic would cool down faster because of the vaporization, which only made thing worse.
Depends on what the cause of layer non-adhesion is. For example, if the filament is just greasy and this grease film interferes with layer bonding, then removing that with acetone from at least one surface would help.
@@haphazard1342 I thintk the primary goal of doing all these experiments is to enhance the layer bonds, not to fix an existing malfunction. I woud doubt that someone will keep their fialment in a dirty environment. Even if you want to clean the filament, it is easier to just add a sponge to clean the filament before it enters the hotend or extruder. I also don't think using acetone to clean filament is a great idea, since several plastic can be dissolved by acetone.
This makes me wonder about another option. Print the piece "spongy". By which I mean either use a special material or adjust the extruder settings to leave voids by continually starting and stopping instead of one continuous extrusion. Then use a thin epoxy or resin that soaks into the part. For example, you can get superglue as thin as water. Basically, 3d print a scaffold then use something else for the actual material strength.
I just googled different glues and epoxies: The gorilla glue or the normal superglue seems to be two good candidates for this but they might dry a bit too fast... (10-45 seconds) A modified 3D printer that is applying the glue (or solvent, epoxy, etc...) automatically with a second nozzle would be a lot better and might be able to print very strong parts.
You could try leaving a hole or two through the model and force a single extrusion of filament through it. It might anchor the layers or normalize the cooling related stresses
What an interesting idea! This made me think of another idea: what if you made only 1 infill layer every 4 layers, but when you did print it you significantly over-extruded it so it filled up the extra space and it overflowed a little. Now you have a large blob of plastic that is hot and melted and may hold more sturdily ( it will look ugly, but it is the job of the outer layers to look pretty) The idea I came here to post was to force the filament through a wet pad of solvent _before_ printing. It will soak into the filament melt and make the plastic softer. It might give an effect like printing at a higher temperature.
you would need really high pressures/very fast extrusion and it still would probably have injection defects. I don't think it would be worth the effort over just printing more walls.
At the viscosity FDM usually works at you would need a hole probably 3x the extrusion diameter for it not to just squirt out the sides and solidify into a huge blob. Even then it's likely going to stick to the walls high up in the hole, solidify and plug the hole without filling it.
One possible way of improving layer strength might be to take a variant of non-planar printing and "weave" the infill lines across 2-3 layers. It wouldn't need the same level of hotend modification as true non-planar printing, but should interlock the layers much better than normal slicing.
Kind of like Prusaslicer's Combine Infill layers, but on nonplanar stereroids? I think that's an interesting avenue of investigation. Maybe printing the infill at like 2x normal layer height or something along with that could have interesting results. I've been including small through holes to put glues into.
This is what I think about as well. But not online infill but outer layers as well. Let the perimeter zigzag in Z direction so each perimeter is in contact with multiple perimeters
(I work with a plastics company in their R&D department) Put in mind they might be weaker due to annealing! Test a normal sample conditioned at 60C to compare with the bonded ones. Putting it at 60C Will probably relieve internal cooling stresses which weakens the part. Condition both at 60C, one as-is and one broken/rebonded, so that you’re comparing apples to apples ;)
Some kind of non-planar printing might be cool: each layer "weaves" up and down so that each layer is partially bonded to the layer below on the sides in addition to underneath, if that makes sense. Might be a bit difficult to program, tho, IDK.
Yeah. I think it could easily work like that for perimeters: You print the entire perimeter, but you leave out a few small sections. In the next pass you basically just print on top of that and get the "weaves".
I've also thought about a method like this often and would be interested to see it in action could possibly weave from the wall back and down one layer into the infill and have the infill lines cross over those weave points
Indeed, for a rectilinear infill (horizontal and vertical), if you imagine printing: 1/2 the vertical lines -> 1/2 the horizontal lines -> other 1/2 of vertical lines -> other 1/2 of horizontal lines -> ... you get a weave!
I would like to see a test where each flat layer of a print was converted to and printed with a sinusoidal pattern that was something like this: xahlee.info/SpecialPlaneCurves_dir/Sinusoid_dir/wave_surface.png where the units on the 3d graph are in Layer Heights. The bed would have to constantly move up and down in Z as the head moves in the X-Y plane but that is a fairly simple mathematical distortion. It shouldn't effect the external geometry any except at the top and bottom surfaces but that would be ok for a Z strength test sample. Since the Z travel would be so small it should be achievable on any printer. I expect the z-strength would increase proportional to how high you could make the peaks relative to the diameter. If you could get a 1/1 ratio (instead of the 2/7.5 ratio shown in the graph) you could probably reduce the difference between the X-Y strength and Z strength by 70%. It i might be hard to get it to that ratio though depending on the geometry of the nozzle tip. The sharper the nozzle the higher the ratio probably. Most nozzles have a small flat area around the hole where the filament comes out which would be a limiting factor probably. A sharper nozzle like how this one is depicted might help: i.ebayimg.com/00/s/ODUyWDEwMDA=/z/UY4AAOSwYzZgxx89/$_57.JPG?set_id=8800005007 (I don't think that picture is really like that, I was just looking for an image that showed what I was talking about). I think this would make a big difference and could be implemented without too much modification.
@@oriansbelt This is how I envisioned achieving the original suggestion by Bjorn-Oskar Ronning. I think it would improve the strength but I'm not sure by how much.
When the acetone evaporates it absorbs heat from the surface. You can feel this if you dip your finger in the acetone and hold it up. Blowing on it increases the evaporation rate and you feel your finger cooling down even more. So the acetone on the part will effectively cause the surface temperature to drop making the ABS have a tough time to adhere to the previous layer. At least that's my take on the whole thing. I don't have any 3D printing experience but everything I've read, seen or heard about printing ABS it's best done in a heated chamber. So the cooling from the acetone evaporating sounds like it would work against you when trying to get the layers to adhere to each other. Someone in the comments suggested using a laser to heat the previous layer just before the extruder puts down a new layer. It might be worth thinking about. But I have no idea how to implement something like that. The extruder will after all move in all directions so either the laser has to move around it so it can hit the area the nozzle is going to cover with a new layer next, or there have to be several lasers placed around the extruder. Seems impractical and expensive, and I have no idea if it would actually help,
Maybe a small heat blower nozzle instead of laser? That would create a bit wider heat island on the surface. Perhaps the printing head nozzle could be constructed in such manner that it has additional external wall (a sleeve) and air cooling the nozzle is forced downwards between the nozzle and its external wall/sleeve, so that extrusion flows down surrounded by downward stream of hot air which heats the point of contact between the print and newly added material.
@@tardonator As I think I said earlier, I don't have any 3D printing experience, but as I understand it we have two conflicting interests, layer adhesion and keeping the shape without warping. Layer adhesion is best when both the previous layer and the extruded plastic is at melting temperatures, but that also means the print is soft and prone to warping. Any overhang will be very likely to sag and it's unlikely the print will be able to finish. The cooling fan is intended to combat this by removing heat from the fresh layer before it can sag. Using a warm airflow to create a "puddle" is therefore not likely to work. Using a very concentrated heat source, such as a laser, also creates a "puddle", but a much smaller one. It would be similar to soldering or welding the plastic and would require a lot less energy and therefore total amount of heat to accomplish. So less heat for the fan to remove to avoid sag. Having written this I kind of wonder if it would be possible to avoid the hot-end and weld the filament directly using a laser... Though If I can dream it up I'm sure others have also thought about it and either tried it or dismissed it for practical reasons that I haven't thought of. I think that when 3D printing metal powder a laser is used to fuse the powder, so there is some merit to the idea. Note that I said "I think"...
@@blahorgaslisk7763 > Having written this I kind of wonder if it would be possible to avoid the hot-end and weld the filament directly using a laser... That's what Selective Laser Sintering (SLS) does. "Print" a layer by melting metal powder. Move the bed down a bit, put more powder on top. Repeat.
@@arthurmoore9488 Thank you for confirming that. Still I don't know if it would be a good idea to try it with plastic. On one hand it's probably possible to get good adhesion, but it also makes it impossible to use infill as the unfused powder would stay trapped in the infill structure. So either you make things solid or you have to leave holes where the powder can evacuated once the print is finished. Edit: Yes I realize that metal prints have the same problems, but when printing metal you are probably looking for maximum strength so you'd print things solid anyway.
Hello Stephan, I think that's the best way to improve adhesion is to use non planar 3D printing, it will largely increase the contact surfaces between 2 layers, but there is a lot of work to be done in the slicers side in order to to this ! An other way would be to over-extrude and machining after each printed slice, with your tool-changer it would be possible. By the way, we have made an aluminum milling head for ASMBL, feel free to contact me if you want a free one for your Tool Changer. Regards,
What about changing the nozzle tip geometry? Something like a cake decorating nozzle that lays star shaped beads instead of round ones. It should allow more surface area between layers and the layers would no longer be planar.
I think you could improve strength if you dual extruded a part with a PETG or ABS outer shell and 100% infill of PLA, then heated the part above the glass transition temperature of PLA but below that of the outer shell. This would allow the PLA core to fuse but the detail of the model to not be disrupted. (Please like this comment so that Stefan sees it! 😊)
iirc even below glass transition many polymers still deform under any stress
2 года назад+47
Ideas: Printing in CO₂ atmosphere, Argon atmosphere, under vacuum, all to avoid surface oxidation. (Vacuum would require huge interlayer times for radiative cooling). Or with laser preheating the area next to nozzle to print liquid ABS onto a thin liquidized surface layer.
Printing under CO2 creates further oxidation at that temperature since there's an oxygen source. Printing under a vacuum causes VOC release. I've managed to get the result you're looking for under dry, high purity nitrogen at positive pressure. The impact on mechanical strength for PLA, ABS, and PETG is minimal, but there are other impacts from the reduction of oxidation, namely the material is more ductile.
2 года назад+2
@@louisvaught2495 I don't see how CO2 could serve as an oxygen source for organic polymers, given that it'd be one carbon reducing another carbon. But perhaps it could attack the polymers as an acid.
@@louisvaught2495 CO2 cannot oxidize things under ~1000°C! But otherwise I agree with you that impact of inert atmosphere would be minimal. The polymer doesn't oxidize at printing temperatures, main mechanism of degradation is depolymerization.
@@Bizzon666 Problem with your claims is they directly contradict both published literature, and the XPS+NMR results hanging out on my laptop. In the future, I'd recommend testing rather than speculating, otherwise you might wind up telling people outright misinformation. Polymer behavior gets very complicated very quickly. At the direction of some of the best in the industry, I've stopped trying to speculate versus just testing it.
Thanks Stefan, this is something I tinkered with also, with quite similar results. Your methodical approach makes the results useful, valid data even when they aren't 'successful', helps us either come up with a more refined version or reject and put energy into other ideas - and it was an enjoyable video, and you got some Japanese learned!
I would love to see a laser (or even a hot air gun with an incredibly small nozzle) that would melt the previous layer just ahead of the print nozzle. That way, both the new layer and the layer it is being laid on are molten, thus allowing an actual weld to occur. It is like if you were trying to "weld" two pieces of plastic together. If only one surface is molten, the weld is incredibly weak. Proper welds, no matter the material, require both surfaces to be molten. I feel like a laser is the most accurate method to ensure only the plastic just ahead of the hotend will be molten (which will minimize deformation).
Maybe using something like a heated needle to stipple the surface of the layer could improve adhesion similar to how you rough up contact surfaces before applying glue? Great video by the way
I read about non-plane printing a year or two ago where a long narrow nozzle is used and the g-code tells the printer to move through X, Y, and Z while printing a layer so that the outside of the part is smoother and much less stepped. I thought that could be used during the infill to make infill that overlaps multiple adjacent layers. However I'm not too sure on the specific geometry that would be needed so that each layer can have some overlap with the layers next to it. Something that would be much easier to execute would be to pause the nozzle over an opening in the infill and continuously extrude to fill multiple layers of infill at once. Each layer only a fraction of the openings would be filled so that it would be staggered and there would be a lot of overlap. This can also be combined with the non-plane printing where you have a long narrow nozzle that lowers itself to the bottom of each opening, starts extruding, and slowly spirals back up. For example if your infill makes a grid of 10 squares, only two squares would be filled each layer, so the extrusion would fill 5 layers. Then the next layer two different squares would be filled and so on until that pattern repeats every 5 layers. That way there's a lot of overlap and no individual layer would be 100% layer adhesion. If you were also doing non-plane printing for this example, the nozzle would position itself over a square, lower down 5 layers, then extrude continuously as the nozzle slowly either spirals or goes straight up 5 layers.
Your dedication to research is so valuable, perhaps especially when the results aren’t positive. Your videos are always a joy to watch from a content as well as production perspective. Thanks, Stefan!
Have you considered printing a part with an outer shell made from a plastic with a high melting point and then 100% infill of another plastic with a much lower melting point, then annealing it at a temperature sufficient to melt the inner plastic but not the outer shell. I have had this idea for some time but unfortunately I don't have a dual extruding printer so I could never try it. Maybe if its a new idea to you you could test it, or maybe you've already tried it and I just missed that video. One last thing to consider is that it might not give significant improvement to your Z axis tensile strength test because the cross section of the part is such a small area.
Thank you for your dedication to these time consuming tests of new ideas. I think that evaporative cooling ruined the layer adhesion. Would it be feasible to laser heat bottom layer in front of a nozzle printing new layer? Or induction heat iron filled filaments?
As someone who prints in almost exclusively ABS I feel the part which caused your welded parts to fall short in strength stems from the accelerated curing in your dehydrator. ABS which has soaked up acetone is very sensitive to heat which can cause the material to puff up weakening the layers. Doing tests myself with trying to speed up the curing of welded parts has lead me to the conclusion of allowing the acetone to evaporate on its own will lead to a stronger bond.
You just saved me of lot of time, thanks for sharing your findings! This was one of the things I wanted to automate with my pump extruder. Maybe I still will ;)
One way to increase inter layer adhesion along the slice axis is to increase your layer width to 0.425 and decrease your layer height as low as 0.05 for a 0.4 mm nozzle. This causes under extrusion to be represented in the horizontal direction instead of the vertical which is the primary cause of layer adhesion failure due to filament diameter tolerances and accuracies along the axis of slicing. Thank you for your work Stephan these videos are very important.
I think you could leave small holes in the print and extrude a bit of (rough?) heated wire through them in the Z direction, then cut it just above the last layer. This would stitch the layers together at a few places. The extruding and cutting could be done with a single toolhead and the wires could either be short, or longer and straightened by pulleys before their extrusion. There are printers that can automagically lay glass or carbon fiber in the XY direction, so this could be quite achievable. More complex than brushing the parts with acetone, but certainly possible. I think you could modify filament extruders and nozzles to make the system easier to make.
Print the parts with a hole in the center and thread a long bolt into them. It might seem as a cheat. But many people oversee this solution despite it being the most easy one to achieve. A hole of 3.1mm in the bottom and 2.7 in the top part is perfect for a 3mm bolt.
The more videos I watch about 3D printing, it seems that the majority of the community is not interested in actually designing things, just messing around with the printers.
@@xSiliconKnightx yep, I am guilty of that too. But its fun to get a cheap ender and upgrade the shit out of it. I am at the final stage tho, just need a dual Z axis and I can sleep again.
2 года назад
That's why I bought a roll hanger strap for, putting it in slots in prints. Now I honestly haven't used it yet though, it would only be needed for huge parts with huge need for strength.
After trying quite a few filaments, I have found taulman 910 nylon to have the best layer adhesion of any filament i have used. A few things I have noticed though: 1) being that it’s nylon it’s very important to have it thoroughly dry (as with all nylons) to achieve optimal adhesion with prints. 2) I anneal most of my prints with it, but notice that the layer adhesion is lacking a little bit after it’s been dried 3) once it is rehydrated, weather by the environment or dying it, the layer adhesion is outstanding, probably as strong as the x/y axis. I know this is a little off track but my point is that I’ve noticed with this particular filament that very dry printing and allowing it to rehydrate fully tends to greatly affect layer adhesion properties in my experience. I would imagine this is the same for most filaments. I noticed that you tested z-axis strength after drying, perhaps if you let it re absorb some moisture, the results may be somewhat different
Either should work. I’ve just noticed with taulman nylon that it gets better when it’s rehydrated. I dye it with RIT dye, so I keep water with the dye at about 160f for around 15 mins while I dye the part
For some parts like functional TPU parts I use a welding nozzle and regulator with argon to give an inert gas flow over the part - stops the TPU oxidising at the higher temps used for high speed extrusion and gives layer bonding so strong parts that fail often fail through mulitiple layers rather than at the joint - might be one to try, you don't need much gas flow, 3-4L/min. Just beware of pumping CO2 or argon into a sealed room, of course.
Could be an easy solution to gas wastage on very long prints if you just flooded and purged an enclosed chamber with inert gas, like when welding up a gas tank.
Maybe braided layers might help adhesion! Have the nozzle lay continuous lines that span consecutive layers by raising Z mid line. It doesn't sound easy to code but it could be a neat proof of concept.
This sounds nearly impossible but yes braiding between layers would add a bit of resistance to breakage (although I suspect not much because you're still limited by the strength of 2 very thin lines of extruded filament). If you had an elastic thermoplastic like TPU which could transfer the stress throughout the model then braiding could be a solution to dramatically increase strength in Z axis
Im sure this has been tried, but what if you had a wave shaped layer instead of a flat plane for a layer? The direction of adhesion would vary throughout the width of the print and de-localize the weaknesses made by the individual layers I know this would be difficult to create a path but if possible seems useful.
If each wave was ‘out of phase’ from their neighbor laterally, there would be contact between many layers. If anyone wants to mess with it they could see about using G2 arcs in loops while using XZ and YZ plane selection modes, or perhaps the latter but using code from generated gyroid fill. Would take some automation or perseverance to make it, especially considering fading from flat build plate.
Bless you for doing the testing for us. Sitting there for 3 hours repeating the same motion would drive me insane but it's definitely an interesting question that I'm glad you answered. I always am looking foward to your videos!
"a 3d printer that is a tool and not the project" is exactly why I bought a Prusa recently. I absolutely love tinkering and building things, but want my tools to actually work well. It hasn't disappointed!
perhaps a modification to the print head to include an acetone vapor to the layer after the deposition would be more automated and optimal. A small bottle of acetone, a small air pump, and a tube attached to the print head to follow the layer deposition.
i believe that to but there is a safety issue here for fire because aeton and other liquids like that start to burn from the fumes around them and with that method you will have a lot of fumes and oxygen
Or we could take the opposite route, instead of trying to melt the plastic with a solvent we could use some kind of substance to help with adesion like some kind of glue or epoxy.
I've always wondered if you could increase layer adhesion using a laser to heat the area that the extrusion is supposed to bond to in advance, and maybe keep it at a specific temperature during / after extrusion for a bit. Probably no better than a heated enclosure, but the pinpoint heat might mean you can get the surface to just slightly melt without risking the whole part.
my sophomore year of college, i brought my 3d printer with me when i moved into my dorm. it was printing primarily ABS, and in order to combat bed warping, i had to seal all of my vents and keep it almost sauna-hot in that room. fortunately i had a private bedroom, joined with 3 others by a community living room and kitchen. With the hot room and no drafts, i was able to get a bunch of very nice prints :D
Idea: Like ironing but the other way: after the layer is done, the nozzle should poke little holes/intents in the surface, so that it roughens up the surface so the next layer is potential able to get a mechanical interlock between the layers.
I would personally attempt running the test again, but with the acetone through a sponge (that is resistant to acetone), as used to be done with oiling prints
As plastic tends to stick best when it's liquid, it might be a good idea to heat up (re-melt) the previous layer before laying the new molten plastic bead on it. Heating could be done by a laser.
If I want a structural plastic part, I 3D print the shell and fill it with epoxy resin. If I want a really strong part, I also put chopped fibre glass in before I fill with epoxy.
I am aware that the solvent breaks the polymer chains which results in smoothing but reduces strength of the surface, sometimes going quite deep during smoothing. Painting on the solvent is probably the same effect. Thanks for the video!
Although it’s quite barbaric you could use a soldering iron to melt a few vertical lines for a bit more strength. I sometimes do this when I know a part is likely to split at a specific point and it seems to help.
In my opinion failures are even more important than successes, because with each failure you're closer to the path to success. Solvents may not necessarily be the way to better layer adhesion, but other untested methods might. Loved the video, great stuff as usual! 🙌
Make the part hollow and fill the cavity with epoxy, using either a funnel or a syringe. If you have a vacuum chamber, vacuum purge unwanted bubbles from the epoxy before putting it into the printed part, and also vacuum purge the part after the epoxy goes in. If you are especially handy, then filling the part with epoxy while it's inside the vacuum chamber is an even better option.
this doesn't improve the strength of the printed part it just uses the shape creation machine and replaces the material under load you could print a mold and make carbon parts out of those
@@marc_frank The original FDM print would still be there, surrounding the epoxy. Also, if you were JUST printing a mold, then you can't get multi-color results.
No, Stefan, I was not actually wondering if you putting solvent on every layer would make it strong. Because that would be insane and who would do that. I mean, I'm wondering that NOW, but like, I never thought that before. You are truly a mad lad.
Here's an idea that would take a ton of programming, basically would require a custom slicer to be built: converse concentric print patterns. For a print with three walls: 1. Print the inner wall with a counter clockwise grade. 2. Print the outer wall with a clockwise grade. 3. Print the middle wall as an over extrusion between the two others. The idea here would be that by introducing opposite-direction slope on the inner and other walls, we would create structural reinforcement. However, because layer heights discrepancies would create problems for the print nuzzle "bumping" parts of the print on the converse pass. Hence the need for an inner buffer wall that must be filled in afterwards via a slight overextrusion between the two other walls. A similar approach would not use concentric opposing directionality, but would break the wall area into sloped curved rectangular cross-layers. Like a cylinder where each layer is like a circle of collapsed dominoes layering on top of each other diagonally.
My idea for better layer adhesion is to make the layers nonplanar. E.g. leave gaps between the lines of the first layer and then fill them with the second layer adding some Z movement or maybe even add height to parts of a layer by intentionally overextruding. Finding patterns that work and interlock the layers well will probably take some effort, but I bet there is some improvement to be made. I wish I had the time and the measuring tools to test that.
you mentioned that layers with texture from the nozzle digging seemed stronger, fixing a textured roller on the print head next to the nozzle and at the same height could add texture and possibly solvent as well at each layer without manual interaction
Excellent work, I wonder if you submerged a finished print in it's respective solvent, if the outside bonding would make a notable difference in layer adhesion. Taking that notion a step further, I wonder if you drilled a hole through the model (or planned for this in the structure of the model) to allow for ingress and draining of a solvent on the inside and outside simultaneously if that would help. Great content, I love watching your analysis of the experiments!
I've been doing very large prints (using a 1200x600x600 printer) in ABS and I've found the best layer bonding strength has come from using a 0.6mm nozzle with 0.8mm line width at 0.2mm height, and cooking the ABS at 250 degrees with a print speed of 80mm/sec. I also use 3 walls and a 5% overlap between walls to minimise porosity in the actual wall structure. That said, this hasn't helped as much as I'd like with fine structures. The stuff I print is functional car parts, so needs to be strong. Currently I'm experimenting with alternative materials to try and find a better solution, so interested to see people's ideas here. I've used your videos to help guide a lot of the decisions I've made around these prints too so thank you for being so thorough! You can see some of the prints I've done over on my channel - I recently upgraded the printer from standard E3D volcano extruders to bondtech BMG and am getting even better prints now, but I think the next movement will be to the X series to really improve the speed.
For the vertical layer samples which are like cones, you need to disregard samples that failed outside of the sampling region (aka any point except the thinnest point) because it means there is a some kind of flaw/anisotropy in the material at that point and is not representative of the material itself
So as you noticed, the layers where the acetone seemed to expand the layer below such that the nozzle has to push around the plastic out to the sides when printing the new layer. That reminded me of how separate pieces of clay are conjoined using the 'score and slip' method. Where you scratch up the two pieces of clay you are joining, and attach them with dissolved clay (clay dust and water solution). What if in the code, for a nearly 100% infill print, after you printed a layer you told the nozzle to go down a tiny bit and rake a crosshatch pattern into the recently printed layer, then return back up to print the next layer. You'd be greatly increasing the surface area those two layers are conjoined by and might increase the strength. This combined with acetone might be even more effective. You'd almost certainly loose dimensional accuracy but still.
Hi Stephan, working on another 3D printing process we found that using a rotary planer was necessary to make an absolutely flat layer interface with a know Z height. The rotary planer was the width of the Y stage. If you look at the cutter of a wide wood planer it is a spiral cutter with a vacuum shroud to remove the dwarf generated. The cuts need to be very thin or you will be breaking off pieces of the part in process. In practice it may require a sacrificial support and infill material to make the planing robust for all possible parts, but for well behaved geometry’s like a cube it is not necessary. A straight forward implementation for a bed slinger is to affix the planer to the Z carriage with a high speed motor for the planer and a stepper driving a torque tube with matching cam lobes at each end to move the planer up and down relative to the z carriage. Just some thing to think about. Cheers.
Here's an idea: 1. slice the hook lengthwise with a lock-and-grove feature so that a single hook is printed in two parts. Bonus point if you can stagger the ridges for extra slotting. 2. Paint the groves with acetone and squeeze them together. A simple slice should add a stronger spine throughout your hook, but by using the lock-and-grove method, you increase the bonding surface to create a wider stronger spine.
For stitching, a hot wire in front of the nozzle heating the previous layer to near melting before the nozzle spooges the next layer immediately out, putting both layers at melting point when printed.
in pottery the method is score and slip, in order to score each layer you could have a star of needles at the mouth of the nozzle, the added roughness alone may improve layer strength.
I reckon it's worth playing with annealing. IIRC ABS doesn't anneal much, but other materials do - and it's basically standard procedure for the fancy materials like PEKK and PPS.
I suppose to improve layer adhesion you would in essence either have to increase bonding strength or increase layer area. Strength would require material or chemical change like you did here but area could be improved by software or a slight hardware mod. One idea I had was to have a slightly serrated/jagged nozzle (really small bumps) so that the plastic being extruded leave a bumpy surface rather than a completely flat one that would allow for a greater surface area and perhaps better shear resistance (The ridges interlocking rather than just sliding). The hard part is you would likely need to print at a higher temp and closer extrusion to ensure that the new layer of plastic fills the gap left by the ridges of the previous layer (eg. each layer squishes into the ridges of the layer bellow whilst leaving a ridged top surface) and to remove the ridges on the very top layer ironing could be used to smooth out the layer.
I've sucessfully used this method with ABS. Model a 2mm hole vertically through the object. After printing, soak a piece of bare filament in acetone for a few seconds and then insert it into the hole so it welds creating an ABS pin through the model
I just thought about this idea, what if you design the parts in such a way that after 3D printed, you can fill the object with resin, or any other hardening kind of glue, and then the part would basically have a strong skeleton! Or you can print the inside of a print (maybe only the infill) with a different kind of material and then the perimeter a different material, to make that part stronger.
The solvent may inhibit layer strength. Ultimately the name of the game is increasing the motility of the molecules in the lower layer so they can form bonds with the new top layer. The solvent on its own does this at room temperature, but soaking into the ABS strands may also mean that once hot filament is extruded on top, the evaporative cooling of the solvent below reduces the temperature it reaches. That could mean you get less layer bonding at high temperatures, despite the presence of solvent.
I know this would be a bit tricky to pull of but I think getting the iso or acetone into a fogger (that's safe) and using a hose to gently hit each layer with fog would work better. In theory it should greatly reduce evaporation time and has the added bonus of being doable while the part continues to print. Still a bit of a stretch but hey it's an idea. Also somehow making the layers print smoother might help too, more details = more places for the liquid to pool and soften the layers above and below. Great video, thank you for the work!
Sugegstion for boding: An ultrasonic beam focused at the bond line between layers. So do one pass depositing a layer, then a second with ultrasonic transducer to remelt the previous bond at the bond line
Theres this whole host of research in non planar 3d printing, where a fourth or 5th rotational axis is used. That allows for orienting the layer lines in whatever direction is needed for strength or overhangs. I dont even think 5 axis DIY builds are that complicated and there are some open source projects out there. But I have no idea what the software side is like. I would love a video on exploring that space and its practical applicability.
One idea for increasing layer adhesion is finding a way for the hotend to run into the already printed layers while printing like what happened on accident with your ABS test. This would force the new layers to mix with old layers by injecting them right into the old layers at times. I suspect this will greatly reduce part accuracy, but it might increase strength so it could be worth testing.
Idea: injection molded infill-ish. Print solid areas by making the perimiters, then fill the part up by extruding a solid blob in the middle. For bigger holes I would try dividing bigger holes into/areas into smaller holes that can be filles by pressing the nozzle in it and do a sort of "injection molding" in your own part as infill.
A small tube or two through the print that you can quickly spray acetone through. (Or use one of those wires with fur on it used in crafts, soaked in acetone to stick into) Basically support pillars
Just a suggestion, there are isopropyl alcohol in pens for cleaning. It's possible to model and mount for those that can be mounted on the hotend. If it's directly in line with the nozzle on the x axis, all you would need to do is print 2 of the same models on the same gcode, and the printer would automatically draw with the pen on top of the first model while it prints the second model. That would give you a control with each print and automate the whole process.
If the z axis were to move within a layer and created a sort of wavy pattern which was continued through the entire height of the print you would get a higher degree of friction between the layers and thus stronger parts. Apart from G-code changes you might need a nozzle with a very fine tip that extends past the print head so that the angle of attack on these “waves” doesn’t clash with anything.
There is definitely under utilized potential in slicer software to improve layer adhesion. Besides all of the settings you already mentioned at the beginning of the video: higher temperature, slower printing speed and less cooling. To increase bonding between layers the mixing of the hot molten plastic needs to be improved. This can be accomplished by: • Printing layers while moving the nozzle up and down should improve mixing and put more heat into the plastic (basically stabbing the hot nozzle down into the previous layer and pulling it back out). • Having a nozzle that isn't flat but spikey will also improve mixing (stiring up the previous layer while putting on the next). • Printing layers in spiral patterns, so the nozzle touches the same area more often and for longer. I really hope you can test these techniques, unfortunately slicer software seems to be more oriented towards good looking prints rather than getting strong isotropic mechanical properties.
Visit printables.com at prusa.io/CNCKitchen/printables and learn more about the Prusa MK3S+ at prusa.io/CNCKitchen/mk3s
Hi. Interesting video.
Question: Ever printed with SBS? It can be smoothed with D'limonine extracted from the skin of citrus fruits. It is also used to recycle polystyrine used in food take aways 100%. Your answer would really interest me. We have a manufacturer in South Africa called Fil X that makes SBS filament. Videos on RUclips refere to imports from Russia years ago.
Love your channel. Always learning something new.
Hi mate. What is the point in that?
I believe 3d printing should be used to make prototypes and parts for hobbyists to use.
If you want strong parts and to print multiple items, wouldn't it be better to injection mould them?
"Can be used by anyone" might be false advertising, now that some models aka "NSFW" are behind an account wall.
Now that I think about it you might use a wire brush of some sort to texture the layers. Using the tool head changer or doing a quick scratch manually sounds simple.
@@mike-ology22 Costs mate, costs. I have a sole proprietor company with a very limited budged. To have injection moulds made for every part, that changes every 6 months as I learn new methods and designs would not be cost effective at all. So my 6 3D printers have to cary the weight of producing parts for the products that I sell. My 6 printers are about R5000 US in total. To have just 1 part made into an injection mould would cost double that due to complicated structures.
Failures are just as important, if not more important, than successes. Props to both you and Tom for both putting the effort into publishing "stuff that didn't work". 👍
See dad! zksHPuPngGs61kNJ says I'm important!
The only failure is when you fail to learn from the failure.
You mean Thomas The Slanderer ?
I was just having this thought yesterday.
I've been looking into hydroponics lately and have so many ideas and I think "I wonder if anyone has tried that". What a time saver it is to know what doesn't work ahead of time. Of course there is the tried and true methods, but I always want bigger, better, faster, and easier.
I agree, the deposition of material perpendicular to the plane is the villain in this case
I had an idea for trying to improve layer strength: staggered layers. Basically if you were printing a 2 perimeter hollow box print the outer layer, raise up half the layer height, print the inner, up half the layer height again, print outer and so on. I tried doing it hand editing gcode and successfully printed a 2mm high test to see if it was possible without dragging the piece off of the bed. I never went back to make a piece big enough to test the strength due to how long it took to edit the gcode for even that small test. I'm not a coder so I couldn't easily think of a way to automate it.
Have you seen the "Alternate Extra Wall" feature on the Cura slicer? It looks like it does a similar pattern. I haven't tried it but maybe it will speed up the process?
@@archercureton3738 it's not quite the same. It improves bonding to infill, but it still prints on the same plane as the rest of the print. I have never looked at the Cura code, but it may be possible to add a z shift to the extra wall more easily than writing a whole new pass for the slicer.
That’s a fantastic idea! Would love to try this idea if I had time to. Maybe there is a better way to automate this.
You could try this with super slicer's "Between extrusion role change G-code" section. Watch out for the first layer :D
A while ago I actually made a Python script that generated a test hook sample with that technique and send it here. Sadly it was never tested.
I tried the same stuff a few years back with epoxy, what I observed was the liquid would be vaporized by the heat instead of helping bond layers together. Meantime, the hot plastic would cool down faster because of the vaporization, which only made thing worse.
That's what I would expect, ie, solvents flash off so fast I wouldn't expect it to do much.
Depends on what the cause of layer non-adhesion is. For example, if the filament is just greasy and this grease film interferes with layer bonding, then removing that with acetone from at least one surface would help.
@@haphazard1342 I thintk the primary goal of doing all these experiments is to enhance the layer bonds, not to fix an existing malfunction. I woud doubt that someone will keep their fialment in a dirty environment. Even if you want to clean the filament, it is easier to just add a sponge to clean the filament before it enters the hotend or extruder. I also don't think using acetone to clean filament is a great idea, since several plastic can be dissolved by acetone.
This makes me wonder about another option. Print the piece "spongy". By which I mean either use a special material or adjust the extruder settings to leave voids by continually starting and stopping instead of one continuous extrusion. Then use a thin epoxy or resin that soaks into the part. For example, you can get superglue as thin as water.
Basically, 3d print a scaffold then use something else for the actual material strength.
I just googled different glues and epoxies:
The gorilla glue or the normal superglue seems to be two good candidates for this but they might dry a bit too fast... (10-45 seconds)
A modified 3D printer that is applying the glue (or solvent, epoxy, etc...) automatically with a second nozzle would be a lot better and might be able to print very strong parts.
You could try leaving a hole or two through the model and force a single extrusion of filament through it. It might anchor the layers or normalize the cooling related stresses
Diy injection molding! How cool is that! Now that is something one of those super-high melt volume hot ends would be good for.
Oooh, injecting a spine after the first printing phase sounds neat.
What an interesting idea! This made me think of another idea: what if you made only 1 infill layer every 4 layers, but when you did print it you significantly over-extruded it so it filled up the extra space and it overflowed a little. Now you have a large blob of plastic that is hot and melted and may hold more sturdily ( it will look ugly, but it is the job of the outer layers to look pretty)
The idea I came here to post was to force the filament through a wet pad of solvent _before_ printing. It will soak into the filament melt and make the plastic softer. It might give an effect like printing at a higher temperature.
you would need really high pressures/very fast extrusion and it still would probably have injection defects. I don't think it would be worth the effort over just printing more walls.
At the viscosity FDM usually works at you would need a hole probably 3x the extrusion diameter for it not to just squirt out the sides and solidify into a huge blob. Even then it's likely going to stick to the walls high up in the hole, solidify and plug the hole without filling it.
One possible way of improving layer strength might be to take a variant of non-planar printing and "weave" the infill lines across 2-3 layers. It wouldn't need the same level of hotend modification as true non-planar printing, but should interlock the layers much better than normal slicing.
Kind of like Prusaslicer's Combine Infill layers, but on nonplanar stereroids? I think that's an interesting avenue of investigation. Maybe printing the infill at like 2x normal layer height or something along with that could have interesting results.
I've been including small through holes to put glues into.
What if my parts almost don't have infill?
This! Another thought that is similar is to extrude large blobs into the infill area. Kinda like filling a print with hot glue.
This is what I think about as well. But not online infill but outer layers as well.
Let the perimeter zigzag in Z direction so each perimeter is in contact with multiple perimeters
What about printing cavities and then later inject material into them with the nozzle? Would interlink layers without the need to weave.
(I work with a plastics company in their R&D department) Put in mind they might be weaker due to annealing! Test a normal sample conditioned at 60C to compare with the bonded ones. Putting it at 60C Will probably relieve internal cooling stresses which weakens the part. Condition both at 60C, one as-is and one broken/rebonded, so that you’re comparing apples to apples ;)
Wasn't the food dehydrator already at 60°C?
The glued parts went through the dehydrator and were still stronger
Some kind of non-planar printing might be cool: each layer "weaves" up and down so that each layer is partially bonded to the layer below on the sides in addition to underneath, if that makes sense. Might be a bit difficult to program, tho, IDK.
Yeah. I think it could easily work like that for perimeters: You print the entire perimeter, but you leave out a few small sections. In the next pass you basically just print on top of that and get the "weaves".
I've also thought about a method like this often and would be interested to see it in action could possibly weave from the wall back and down one layer into the infill and have the infill lines cross over those weave points
Indeed, for a rectilinear infill (horizontal and vertical), if you imagine printing: 1/2 the vertical lines -> 1/2 the horizontal lines -> other 1/2 of vertical lines -> other 1/2 of horizontal lines -> ... you get a weave!
I would like to see a test where each flat layer of a print was converted to and printed with a sinusoidal pattern that was something like this: xahlee.info/SpecialPlaneCurves_dir/Sinusoid_dir/wave_surface.png where the units on the 3d graph are in Layer Heights. The bed would have to constantly move up and down in Z as the head moves in the X-Y plane but that is a fairly simple mathematical distortion. It shouldn't effect the external geometry any except at the top and bottom surfaces but that would be ok for a Z strength test sample. Since the Z travel would be so small it should be achievable on any printer. I expect the z-strength would increase proportional to how high you could make the peaks relative to the diameter. If you could get a 1/1 ratio (instead of the 2/7.5 ratio shown in the graph) you could probably reduce the difference between the X-Y strength and Z strength by 70%. It i might be hard to get it to that ratio though depending on the geometry of the nozzle tip. The sharper the nozzle the higher the ratio probably. Most nozzles have a small flat area around the hole where the filament comes out which would be a limiting factor probably. A sharper nozzle like how this one is depicted might help: i.ebayimg.com/00/s/ODUyWDEwMDA=/z/UY4AAOSwYzZgxx89/$_57.JPG?set_id=8800005007 (I don't think that picture is really like that, I was just looking for an image that showed what I was talking about). I think this would make a big difference and could be implemented without too much modification.
@@oriansbelt This is how I envisioned achieving the original suggestion by Bjorn-Oskar Ronning. I think it would improve the strength but I'm not sure by how much.
When the acetone evaporates it absorbs heat from the surface. You can feel this if you dip your finger in the acetone and hold it up. Blowing on it increases the evaporation rate and you feel your finger cooling down even more. So the acetone on the part will effectively cause the surface temperature to drop making the ABS have a tough time to adhere to the previous layer.
At least that's my take on the whole thing. I don't have any 3D printing experience but everything I've read, seen or heard about printing ABS it's best done in a heated chamber. So the cooling from the acetone evaporating sounds like it would work against you when trying to get the layers to adhere to each other.
Someone in the comments suggested using a laser to heat the previous layer just before the extruder puts down a new layer. It might be worth thinking about. But I have no idea how to implement something like that. The extruder will after all move in all directions so either the laser has to move around it so it can hit the area the nozzle is going to cover with a new layer next, or there have to be several lasers placed around the extruder. Seems impractical and expensive, and I have no idea if it would actually help,
Maybe a small heat blower nozzle instead of laser? That would create a bit wider heat island on the surface. Perhaps the printing head nozzle could be constructed in such manner that it has additional external wall (a sleeve) and air cooling the nozzle is forced downwards between the nozzle and its external wall/sleeve, so that extrusion flows down surrounded by downward stream of hot air which heats the point of contact between the print and newly added material.
@salec That method will result i significant warping, I think. The laser approach likely mitigates that much better by localizing heat.
@@tardonator As I think I said earlier, I don't have any 3D printing experience, but as I understand it we have two conflicting interests, layer adhesion and keeping the shape without warping.
Layer adhesion is best when both the previous layer and the extruded plastic is at melting temperatures, but that also means the print is soft and prone to warping. Any overhang will be very likely to sag and it's unlikely the print will be able to finish.
The cooling fan is intended to combat this by removing heat from the fresh layer before it can sag.
Using a warm airflow to create a "puddle" is therefore not likely to work.
Using a very concentrated heat source, such as a laser, also creates a "puddle", but a much smaller one. It would be similar to soldering or welding the plastic and would require a lot less energy and therefore total amount of heat to accomplish. So less heat for the fan to remove to avoid sag.
Having written this I kind of wonder if it would be possible to avoid the hot-end and weld the filament directly using a laser...
Though If I can dream it up I'm sure others have also thought about it and either tried it or dismissed it for practical reasons that I haven't thought of.
I think that when 3D printing metal powder a laser is used to fuse the powder, so there is some merit to the idea. Note that I said "I think"...
@@blahorgaslisk7763 > Having written this I kind of wonder if it would be possible to avoid the hot-end and weld the filament directly using a laser...
That's what Selective Laser Sintering (SLS) does. "Print" a layer by melting metal powder. Move the bed down a bit, put more powder on top. Repeat.
@@arthurmoore9488 Thank you for confirming that. Still I don't know if it would be a good idea to try it with plastic. On one hand it's probably possible to get good adhesion, but it also makes it impossible to use infill as the unfused powder would stay trapped in the infill structure. So either you make things solid or you have to leave holes where the powder can evacuated once the print is finished.
Edit: Yes I realize that metal prints have the same problems, but when printing metal you are probably looking for maximum strength so you'd print things solid anyway.
Hello Stephan,
I think that's the best way to improve adhesion is to use non planar 3D printing, it will largely increase the contact surfaces between 2 layers, but there is a lot of work to be done in the slicers side in order to to this !
An other way would be to over-extrude and machining after each printed slice, with your tool-changer it would be possible. By the way, we have made an aluminum milling head for ASMBL, feel free to contact me if you want a free one for your Tool Changer.
Regards,
Digging the hot Nozzle or a second hot nozzel/spear into the part in a pattern after each layer could also improve the bonding.
Totally expected you’d build a robot to do the coating 😜
😅 I was a bit lazy TBH.
@@CNCKitchen you could say the same if you did build the robot 😝
@@CNCKitchen on the contrary 🙏
@@BrazenRain Putting a lot of effort into not putting effort is a constant state of my life
This is not stuff made here dude hahha
What about changing the nozzle tip geometry? Something like a cake decorating nozzle that lays star shaped beads instead of round ones. It should allow more surface area between layers and the layers would no longer be planar.
I think that would likely create stringing issues from the sharp corners dragging.
Cool idea though. Maybe an ovoid shaped nozzle would work.
We gotta respect this man's follow through. Really sat there brushing acetone in between each layer.
I like the idea of laser sintering the layers together as it is printed. A fiber laser head at the print head to continuously stitch layers together.
I think you could improve strength if you dual extruded a part with a PETG or ABS outer shell and 100% infill of PLA, then heated the part above the glass transition temperature of PLA but below that of the outer shell. This would allow the PLA core to fuse but the detail of the model to not be disrupted.
(Please like this comment so that Stefan sees it! 😊)
Interesting idea, seems possible.
Someone needs to try this, it seems like a simple idea but I have never seen it done.
This is such a genius idea, i have to see it in action since i don’t have the supplies to test it myself!!
iirc even below glass transition many polymers still deform under any stress
Ideas: Printing in CO₂ atmosphere, Argon atmosphere, under vacuum, all to avoid surface oxidation. (Vacuum would require huge interlayer times for radiative cooling). Or with laser preheating the area next to nozzle to print liquid ABS onto a thin liquidized surface layer.
Printing under CO2 creates further oxidation at that temperature since there's an oxygen source. Printing under a vacuum causes VOC release.
I've managed to get the result you're looking for under dry, high purity nitrogen at positive pressure. The impact on mechanical strength for PLA, ABS, and PETG is minimal, but there are other impacts from the reduction of oxidation, namely the material is more ductile.
@@louisvaught2495 I don't see how CO2 could serve as an oxygen source for organic polymers, given that it'd be one carbon reducing another carbon. But perhaps it could attack the polymers as an acid.
@@louisvaught2495 CO2 cannot oxidize things under ~1000°C! But otherwise I agree with you that impact of inert atmosphere would be minimal. The polymer doesn't oxidize at printing temperatures, main mechanism of degradation is depolymerization.
@@Bizzon666 Problem with your claims is they directly contradict both published literature, and the XPS+NMR results hanging out on my laptop.
In the future, I'd recommend testing rather than speculating, otherwise you might wind up telling people outright misinformation.
Polymer behavior gets very complicated very quickly. At the direction of some of the best in the industry, I've stopped trying to speculate versus just testing it.
@@louisvaught2495 Ok, I stand corrected, I didn't know the depolymerization is of oxidative mechanism 😬 I apologize
Thanks Stefan, this is something I tinkered with also, with quite similar results. Your methodical approach makes the results useful, valid data even when they aren't 'successful', helps us either come up with a more refined version or reject and put energy into other ideas - and it was an enjoyable video, and you got some Japanese learned!
I would love to see a laser (or even a hot air gun with an incredibly small nozzle) that would melt the previous layer just ahead of the print nozzle. That way, both the new layer and the layer it is being laid on are molten, thus allowing an actual weld to occur.
It is like if you were trying to "weld" two pieces of plastic together. If only one surface is molten, the weld is incredibly weak. Proper welds, no matter the material, require both surfaces to be molten. I feel like a laser is the most accurate method to ensure only the plastic just ahead of the hotend will be molten (which will minimize deformation).
Maybe using something like a heated needle to stipple the surface of the layer could improve adhesion similar to how you rough up contact surfaces before applying glue?
Great video by the way
maybe sand each layer lightly, creating small grooves that increase the surface area
I was thinking to machine the nozzle end with very small protrusions that would groove the surface as it printed.
I would go even further: poke the needle _through_ the last layer, to re-melt its interface with the one before.
I read about non-plane printing a year or two ago where a long narrow nozzle is used and the g-code tells the printer to move through X, Y, and Z while printing a layer so that the outside of the part is smoother and much less stepped. I thought that could be used during the infill to make infill that overlaps multiple adjacent layers. However I'm not too sure on the specific geometry that would be needed so that each layer can have some overlap with the layers next to it.
Something that would be much easier to execute would be to pause the nozzle over an opening in the infill and continuously extrude to fill multiple layers of infill at once. Each layer only a fraction of the openings would be filled so that it would be staggered and there would be a lot of overlap. This can also be combined with the non-plane printing where you have a long narrow nozzle that lowers itself to the bottom of each opening, starts extruding, and slowly spirals back up. For example if your infill makes a grid of 10 squares, only two squares would be filled each layer, so the extrusion would fill 5 layers. Then the next layer two different squares would be filled and so on until that pattern repeats every 5 layers. That way there's a lot of overlap and no individual layer would be 100% layer adhesion. If you were also doing non-plane printing for this example, the nozzle would position itself over a square, lower down 5 layers, then extrude continuously as the nozzle slowly either spirals or goes straight up 5 layers.
Your dedication to research is so valuable, perhaps especially when the results aren’t positive. Your videos are always a joy to watch from a content as well as production perspective. Thanks, Stefan!
Have you considered printing a part with an outer shell made from a plastic with a high melting point and then 100% infill of another plastic with a much lower melting point, then annealing it at a temperature sufficient to melt the inner plastic but not the outer shell. I have had this idea for some time but unfortunately I don't have a dual extruding printer so I could never try it. Maybe if its a new idea to you you could test it, or maybe you've already tried it and I just missed that video. One last thing to consider is that it might not give significant improvement to your Z axis tensile strength test because the cross section of the part is such a small area.
Fascinating, that could work.
This sounds lilke a solid idea
This seems like it is worth a test. Stefan you are the guy to do it given your awesome testing setup!
I'd like to see this done
I would die to see this, i think it would work great.
Thank you for your dedication to these time consuming tests of new ideas. I think that evaporative cooling ruined the layer adhesion. Would it be feasible to laser heat bottom layer in front of a nozzle printing new layer? Or induction heat iron filled filaments?
Might be a good option, though I do think there is a patent on a similar idea.
As someone who prints in almost exclusively ABS I feel the part which caused your welded parts to fall short in strength stems from the accelerated curing in your dehydrator. ABS which has soaked up acetone is very sensitive to heat which can cause the material to puff up weakening the layers. Doing tests myself with trying to speed up the curing of welded parts has lead me to the conclusion of allowing the acetone to evaporate on its own will lead to a stronger bond.
Data, no matter how it comes out, is incredibly important. I appreciate the dive into this Stefan and also CONGRATS ON 400k SUBS!!! Keep it up!
Thanks man, much appreciated!
It would be nice to know how much of an effect a heated chamber has on layer adhesion when it is at the material's glass transition temperature.
Might try that at some point though I think the high-temp materials benefit significantly more in this regard.
You just saved me of lot of time, thanks for sharing your findings! This was one of the things I wanted to automate with my pump extruder. Maybe I still will ;)
One way to increase inter layer adhesion along the slice axis is to increase your layer width to 0.425 and decrease your layer height as low as 0.05 for a 0.4 mm nozzle. This causes under extrusion to be represented in the horizontal direction instead of the vertical which is the primary cause of layer adhesion failure due to filament diameter tolerances and accuracies along the axis of slicing. Thank you for your work Stephan these videos are very important.
I think you could leave small holes in the print and extrude a bit of (rough?) heated wire through them in the Z direction, then cut it just above the last layer. This would stitch the layers together at a few places. The extruding and cutting could be done with a single toolhead and the wires could either be short, or longer and straightened by pulleys before their extrusion.
There are printers that can automagically lay glass or carbon fiber in the XY direction, so this could be quite achievable. More complex than brushing the parts with acetone, but certainly possible. I think you could modify filament extruders and nozzles to make the system easier to make.
Post-tensioned polymer? That sounds like a fun experiment, and it’s funny how composite material strategies cross over from concrete for example
Print the parts with a hole in the center and thread a long bolt into them. It might seem as a cheat. But many people oversee this solution despite it being the most easy one to achieve. A hole of 3.1mm in the bottom and 2.7 in the top part is perfect for a 3mm bolt.
The more videos I watch about 3D printing, it seems that the majority of the community is not interested in actually designing things, just messing around with the printers.
@@xSiliconKnightx yep, I am guilty of that too. But its fun to get a cheap ender and upgrade the shit out of it. I am at the final stage tho, just need a dual Z axis and I can sleep again.
That's why I bought a roll hanger strap for, putting it in slots in prints. Now I honestly haven't used it yet though, it would only be needed for huge parts with huge need for strength.
After trying quite a few filaments, I have found taulman 910 nylon to have the best layer adhesion of any filament i have used. A few things I have noticed though:
1) being that it’s nylon it’s very important to have it thoroughly dry (as with all nylons) to achieve optimal adhesion with prints.
2) I anneal most of my prints with it, but notice that the layer adhesion is lacking a little bit after it’s been dried
3) once it is rehydrated, weather by the environment or dying it, the layer adhesion is outstanding, probably as strong as the x/y axis.
I know this is a little off track but my point is that I’ve noticed with this particular filament that very dry printing and allowing it to rehydrate fully tends to greatly affect layer adhesion properties in my experience. I would imagine this is the same for most filaments. I noticed that you tested z-axis strength after drying, perhaps if you let it re absorb some moisture, the results may be somewhat different
So dunk in warm water for a bit? Wait a day or two?
Either should work. I’ve just noticed with taulman nylon that it gets better when it’s rehydrated. I dye it with RIT dye, so I keep water with the dye at about 160f for around 15 mins while I dye the part
Need to check that out at some point!
Thank you for making the effort to post the video even with it not having worked.
For some parts like functional TPU parts I use a welding nozzle and regulator with argon to give an inert gas flow over the part - stops the TPU oxidising at the higher temps used for high speed extrusion and gives layer bonding so strong parts that fail often fail through mulitiple layers rather than at the joint - might be one to try, you don't need much gas flow, 3-4L/min.
Just beware of pumping CO2 or argon into a sealed room, of course.
Could be an easy solution to gas wastage on very long prints if you just flooded and purged an enclosed chamber with inert gas, like when welding up a gas tank.
Maybe braided layers might help adhesion! Have the nozzle lay continuous lines that span consecutive layers by raising Z mid line. It doesn't sound easy to code but it could be a neat proof of concept.
This sounds nearly impossible but yes braiding between layers would add a bit of resistance to breakage (although I suspect not much because you're still limited by the strength of 2 very thin lines of extruded filament). If you had an elastic thermoplastic like TPU which could transfer the stress throughout the model then braiding could be a solution to dramatically increase strength in Z axis
Im sure this has been tried, but what if you had a wave shaped layer instead of a flat plane for a layer? The direction of adhesion would vary throughout the width of the print and de-localize the weaknesses made by the individual layers
I know this would be difficult to create a path but if possible seems useful.
Possibly a 3d version with pyramids for even more interface surface, kind of like that egg crate foam
If each wave was ‘out of phase’ from their neighbor laterally, there would be contact between many layers. If anyone wants to mess with it they could see about using G2 arcs in loops while using XZ and YZ plane selection modes, or perhaps the latter but using code from generated gyroid fill. Would take some automation or perseverance to make it, especially considering fading from flat build plate.
Bless you for doing the testing for us. Sitting there for 3 hours repeating the same motion would drive me insane but it's definitely an interesting question that I'm glad you answered. I always am looking foward to your videos!
You're welcome! I'm usually also not the patient guy but somehow I was able to make it through the process.
The patience required for this should be studied by neurological psychology departments worldwide.
"a 3d printer that is a tool and not the project" is exactly why I bought a Prusa recently. I absolutely love tinkering and building things, but want my tools to actually work well. It hasn't disappointed!
I have idea where you might able to make your own very long nozzle and give another test on the non planer 3D printing.
the layers attentions might be improve if you put some hole in the middle to inject some astone inside might be interesting for muti holes too.
This is a level of dedication I did not think was possible. Props to you good sir
Cool test. Love how you're still using that test rig. Keep up the good work !!
perhaps a modification to the print head to include an acetone vapor to the layer after the deposition would be more automated and optimal. A small bottle of acetone, a small air pump, and a tube attached to the print head to follow the layer deposition.
i believe that to but there is a safety issue here for fire because aeton and other liquids like that start to burn from the fumes around them and with that method you will have a lot of fumes and oxygen
Or we could take the opposite route, instead of trying to melt the plastic with a solvent we could use some kind of substance to help with adesion like some kind of glue or epoxy.
I've always wondered if you could increase layer adhesion using a laser to heat the area that the extrusion is supposed to bond to in advance, and maybe keep it at a specific temperature during / after extrusion for a bit. Probably no better than a heated enclosure, but the pinpoint heat might mean you can get the surface to just slightly melt without risking the whole part.
Just printing slow can do this, the nozzle will heat up what's around it. The part cooling fan reduces this effect though
Yep, that or a strongly directed hot air nozzle
my sophomore year of college, i brought my 3d printer with me when i moved into my dorm. it was printing primarily ABS, and in order to combat bed warping, i had to seal all of my vents and keep it almost sauna-hot in that room. fortunately i had a private bedroom, joined with 3 others by a community living room and kitchen. With the hot room and no drafts, i was able to get a bunch of very nice prints :D
Idea: Like ironing but the other way: after the layer is done, the nozzle should poke little holes/intents in the surface, so that it roughens up the surface so the next layer is potential able to get a mechanical interlock between the layers.
I would personally attempt running the test again, but with the acetone through a sponge (that is resistant to acetone), as used to be done with oiling prints
🔥⚠
Good idea but I think that would just gunk everything up^^
As plastic tends to stick best when it's liquid, it might be a good idea to heat up (re-melt) the previous layer before laying the new molten plastic bead on it. Heating could be done by a laser.
And blinding the consumer will be done by that laser. This is only viable in a full enclosure printer. Preferably with 100% opague sides.
i just thought of this idea while printing, and went on the web to see if anyone had done it and ofcourse cnc kitchen had lol! good video!
If I want a structural plastic part, I 3D print the shell and fill it with epoxy resin. If I want a really strong part, I also put chopped fibre glass in before I fill with epoxy.
Thx for the idea!
I am aware that the solvent breaks the polymer chains which results in smoothing but reduces strength of the surface, sometimes going quite deep during smoothing. Painting on the solvent is probably the same effect.
Thanks for the video!
The mention of the head digging into the previous layer reminded me of Friction Stir Welding.
He Is a Man of Focus, Commitment and Sheer F***ing Will!
Next up: Super Glue per layer.
Although it’s quite barbaric you could use a soldering iron to melt a few vertical lines for a bit more strength.
I sometimes do this when I know a part is likely to split at a specific point and it seems to help.
Bless your soul for testing this
I love how you are using a proper respirator mask when doing this!
In my opinion failures are even more important than successes, because with each failure you're closer to the path to success. Solvents may not necessarily be the way to better layer adhesion, but other untested methods might. Loved the video, great stuff as usual! 🙌
Make the part hollow and fill the cavity with epoxy, using either a funnel or a syringe.
If you have a vacuum chamber, vacuum purge unwanted bubbles from the epoxy before putting it into the printed part, and also vacuum purge the part after the epoxy goes in.
If you are especially handy, then filling the part with epoxy while it's inside the vacuum chamber is an even better option.
this doesn't improve the strength of the printed part
it just uses the shape creation machine and replaces the material under load
you could print a mold and make carbon parts out of those
@@marc_frank The original FDM print would still be there, surrounding the epoxy. Also, if you were JUST printing a mold, then you can't get multi-color results.
No, Stefan, I was not actually wondering if you putting solvent on every layer would make it strong. Because that would be insane and who would do that. I mean, I'm wondering that NOW, but like, I never thought that before. You are truly a mad lad.
Usually just leave holes for screws and nuts. Works incredibly well and widens the range of use for 3d printers quite a bit.
The best 3d printer content maker ever
Cheers, thank you!
@@CNCKitchen Hi Stefan I want to inform you, my project has finally finished! (check mailbox I sent you)
Here's an idea that would take a ton of programming, basically would require a custom slicer to be built: converse concentric print patterns. For a print with three walls:
1. Print the inner wall with a counter clockwise grade.
2. Print the outer wall with a clockwise grade.
3. Print the middle wall as an over extrusion between the two others.
The idea here would be that by introducing opposite-direction slope on the inner and other walls, we would create structural reinforcement.
However, because layer heights discrepancies would create problems for the print nuzzle "bumping" parts of the print on the converse pass. Hence the need for an inner buffer wall that must be filled in afterwards via a slight overextrusion between the two other walls.
A similar approach would not use concentric opposing directionality, but would break the wall area into sloped curved rectangular cross-layers. Like a cylinder where each layer is like a circle of collapsed dominoes layering on top of each other diagonally.
Was ein Aufwand für "bringt leider nix"! Danke dass wir das nicht alle durchmachen müssen und du einen fürs Team auf dich genommen hast😉
My idea for better layer adhesion is to make the layers nonplanar. E.g. leave gaps between the lines of the first layer and then fill them with the second layer adding some Z movement or maybe even add height to parts of a layer by intentionally overextruding. Finding patterns that work and interlock the layers well will probably take some effort, but I bet there is some improvement to be made. I wish I had the time and the measuring tools to test that.
Ok, now that takes some serious dedication. Bravo.
Publishing negative results is extremely useful. This should be the case also in scientific research.
you could ad a part to your g-code that slowly pushes the nozzle in the print at diferent points. it could "stich" the layers together
you mentioned that layers with texture from the nozzle digging seemed stronger, fixing a textured roller on the print head next to the nozzle and at the same height could add texture and possibly solvent as well at each layer without manual interaction
Experience is king, theory can only take us so far. Thanks again for sharing.
Excellent work, I wonder if you submerged a finished print in it's respective solvent, if the outside bonding would make a notable difference in layer adhesion. Taking that notion a step further, I wonder if you drilled a hole through the model (or planned for this in the structure of the model) to allow for ingress and draining of a solvent on the inside and outside simultaneously if that would help. Great content, I love watching your analysis of the experiments!
OK you got a thumbs up just for the insane dedication needed to make this video. TY 👍👍😻😻
Thanks for the great experiments as always! For a side note, also stoked to see you're learning my language... がんばって!
You should create a soldering iron tool head that will push jagged staples into the print every few layers.
I've been doing very large prints (using a 1200x600x600 printer) in ABS and I've found the best layer bonding strength has come from using a 0.6mm nozzle with 0.8mm line width at 0.2mm height, and cooking the ABS at 250 degrees with a print speed of 80mm/sec. I also use 3 walls and a 5% overlap between walls to minimise porosity in the actual wall structure.
That said, this hasn't helped as much as I'd like with fine structures. The stuff I print is functional car parts, so needs to be strong. Currently I'm experimenting with alternative materials to try and find a better solution, so interested to see people's ideas here. I've used your videos to help guide a lot of the decisions I've made around these prints too so thank you for being so thorough!
You can see some of the prints I've done over on my channel - I recently upgraded the printer from standard E3D volcano extruders to bondtech BMG and am getting even better prints now, but I think the next movement will be to the X series to really improve the speed.
saving this video as a reference for my future custom gunpla model part projects
For the vertical layer samples which are like cones, you need to disregard samples that failed outside of the sampling region (aka any point except the thinnest point) because it means there is a some kind of flaw/anisotropy in the material at that point and is not representative of the material itself
I'm also learning Japanese! Don't give up! Keep up the grind! It'll be worth it when we get to where we want to be!
I very much enjoyed this video. Good science even though the concept didn't quite work out yet. Nice work!
So as you noticed, the layers where the acetone seemed to expand the layer below such that the nozzle has to push around the plastic out to the sides when printing the new layer. That reminded me of how separate pieces of clay are conjoined using the 'score and slip' method. Where you scratch up the two pieces of clay you are joining, and attach them with dissolved clay (clay dust and water solution).
What if in the code, for a nearly 100% infill print, after you printed a layer you told the nozzle to go down a tiny bit and rake a crosshatch pattern into the recently printed layer, then return back up to print the next layer. You'd be greatly increasing the surface area those two layers are conjoined by and might increase the strength. This combined with acetone might be even more effective. You'd almost certainly loose dimensional accuracy but still.
Hi Stephan, working on another 3D printing process we found that using a rotary planer was necessary to make an absolutely flat layer interface with a know Z height. The rotary planer was the width of the Y stage. If you look at the cutter of a wide wood planer it is a spiral cutter with a vacuum shroud to remove the dwarf generated. The cuts need to be very thin or you will be breaking off pieces of the part in process. In practice it may require a sacrificial support and infill material to make the planing robust for all possible parts, but for well behaved geometry’s like a cube it is not necessary. A straight forward implementation for a bed slinger is to affix the planer to the Z carriage with a high speed motor for the planer and a stepper driving a torque tube with matching cam lobes at each end to move the planer up and down relative to the z carriage. Just some thing to think about. Cheers.
Here's an idea:
1. slice the hook lengthwise with a lock-and-grove feature so that a single hook is printed in two parts. Bonus point if you can stagger the ridges for extra slotting.
2. Paint the groves with acetone and squeeze them together.
A simple slice should add a stronger spine throughout your hook, but by using the lock-and-grove method, you increase the bonding surface to create a wider stronger spine.
For stitching, a hot wire in front of the nozzle heating the previous layer to near melting before the nozzle spooges the next layer immediately out, putting both layers at melting point when printed.
in pottery the method is score and slip, in order to score each layer you could have a star of needles at the mouth of the nozzle, the added roughness alone may improve layer strength.
This man is out there thinking how to make printing times even longer.
I reckon it's worth playing with annealing. IIRC ABS doesn't anneal much, but other materials do - and it's basically standard procedure for the fancy materials like PEKK and PPS.
I suppose to improve layer adhesion you would in essence either have to increase bonding strength or increase layer area. Strength would require material or chemical change like you did here but area could be improved by software or a slight hardware mod. One idea I had was to have a slightly serrated/jagged nozzle (really small bumps) so that the plastic being extruded leave a bumpy surface rather than a completely flat one that would allow for a greater surface area and perhaps better shear resistance (The ridges interlocking rather than just sliding). The hard part is you would likely need to print at a higher temp and closer extrusion to ensure that the new layer of plastic fills the gap left by the ridges of the previous layer (eg. each layer squishes into the ridges of the layer bellow whilst leaving a ridged top surface) and to remove the ridges on the very top layer ironing could be used to smooth out the layer.
Thank you! I would never have the patience, but the result is good to know
I've sucessfully used this method with ABS. Model a 2mm hole vertically through the object. After printing, soak a piece of bare filament in acetone for a few seconds and then insert it into the hole so it welds creating an ABS pin through the model
I just thought about this idea, what if you design the parts in such a way that after 3D printed, you can fill the object with resin, or any other hardening kind of glue, and then the part would basically have a strong skeleton!
Or you can print the inside of a print (maybe only the infill) with a different kind of material and then the perimeter a different material, to make that part stronger.
The solvent may inhibit layer strength. Ultimately the name of the game is increasing the motility of the molecules in the lower layer so they can form bonds with the new top layer.
The solvent on its own does this at room temperature, but soaking into the ABS strands may also mean that once hot filament is extruded on top, the evaporative cooling of the solvent below reduces the temperature it reaches. That could mean you get less layer bonding at high temperatures, despite the presence of solvent.
I know this would be a bit tricky to pull of but I think getting the iso or acetone into a fogger (that's safe) and using a hose to gently hit each layer with fog would work better. In theory it should greatly reduce evaporation time and has the added bonus of being doable while the part continues to print. Still a bit of a stretch but hey it's an idea. Also somehow making the layers print smoother might help too, more details = more places for the liquid to pool and soften the layers above and below. Great video, thank you for the work!
Sugegstion for boding: An ultrasonic beam focused at the bond line between layers. So do one pass depositing a layer, then a second with ultrasonic transducer to remelt the previous bond at the bond line
Theres this whole host of research in non planar 3d printing, where a fourth or 5th rotational axis is used. That allows for orienting the layer lines in whatever direction is needed for strength or overhangs. I dont even think 5 axis DIY builds are that complicated and there are some open source projects out there. But I have no idea what the software side is like. I would love a video on exploring that space and its practical applicability.
One idea for increasing layer adhesion is finding a way for the hotend to run into the already printed layers while printing like what happened on accident with your ABS test. This would force the new layers to mix with old layers by injecting them right into the old layers at times.
I suspect this will greatly reduce part accuracy, but it might increase strength so it could be worth testing.
For the question at 10:00 - for texturing layers, maybe a noise or fuzzy texture along the +z axis could lock or stitch several layers together.
Idea: injection molded infill-ish.
Print solid areas by making the perimiters, then fill the part up by extruding a solid blob in the middle.
For bigger holes I would try dividing bigger holes into/areas into smaller holes that can be filles by pressing the nozzle in it and do a sort of "injection molding" in your own part as infill.
A small tube or two through the print that you can quickly spray acetone through. (Or use one of those wires with fur on it used in crafts, soaked in acetone to stick into)
Basically support pillars
Just a suggestion, there are isopropyl alcohol in pens for cleaning. It's possible to model and mount for those that can be mounted on the hotend. If it's directly in line with the nozzle on the x axis, all you would need to do is print 2 of the same models on the same gcode, and the printer would automatically draw with the pen on top of the first model while it prints the second model. That would give you a control with each print and automate the whole process.
If the z axis were to move within a layer and created a sort of wavy pattern which was continued through the entire height of the print you would get a higher degree of friction between the layers and thus stronger parts. Apart from G-code changes you might need a nozzle with a very fine tip that extends past the print head so that the angle of attack on these “waves” doesn’t clash with anything.
There is definitely under utilized potential in slicer software to improve layer adhesion. Besides all of the settings you already mentioned at the beginning of the video: higher temperature, slower printing speed and less cooling. To increase bonding between layers the mixing of the hot molten plastic needs to be improved. This can be accomplished by:
• Printing layers while moving the nozzle up and down should improve mixing and put more heat into the plastic (basically stabbing the hot nozzle down into the previous layer and pulling it back out).
• Having a nozzle that isn't flat but spikey will also improve mixing (stiring up the previous layer while putting on the next).
• Printing layers in spiral patterns, so the nozzle touches the same area more often and for longer.
I really hope you can test these techniques, unfortunately slicer software seems to be more oriented towards good looking prints rather than getting strong isotropic mechanical properties.