The "heat treat" you did likely did nothing, if anything softened the parts. Heat treating steel requires specific temperatures and specific times for cooling to create the proper grain structure. It's followed by an annealing process to dial back the hardness to a specific number. I would recommend redoing those tests and obtaining pieces that have been properly heat treated to a specific hardness. I was surprised on the 3d printed part strength.
It's all pointless unless multiple samples of the same materials and treatments etc. are tested, we have no idea if these parts are representative of the average expected strength or not, what the standard deviation is etc.
I'm a tool and die maker and know by accident a bit about heat treatment. By heating up the parts to about 500°C you annealed them and reduced any existing hardnes. D2 fpr example needs a temperature of 1020-1050°C, followed by a chill in oil or moved nitrogen gas and has to be annealed at about 500°C afterwards for several hours.
Yep, there's a lot of weirdness going around there in general. He manages to heat the part to a nice orange in the centre (one spot even reaching yellow) which would suggest temperatures of >1000degC. Steel at 500 degrees shouldn't even be reddish. Which would be decent temperatures for a heat treat, except that it should then be chilled quickly (oil quenching recommended for 4140). If it gets below tempering temperatures too slowly you'll just end up with an annealed steel again. He might've actually gotten lucky with the accidental heat treat as the thinness of the part causes it to cool down really quickly. But eh, that's rather unlikely.
I couldn’t include all of the data in the video because the algorithm would punish the video. All the data is available on my Patreon. The link is in the description.
@@oli5dijksma616well he's not publically funded unlike the researchers who are and paywall all their findings, also he's not using strain gauges on the thin part of the material which is the only strain you actually need to care about
@@cybyrd9615 the researchers aren’t the ones paywalling the publishers are. Literally just ask the researchers personally for a copy of their paper, they’re allowed to provide it for free.
I understand you are doing your best, but . . . . As someone who works in materials testing, testing a single sample really isn't enough to draw many conclusions. You could just have a bad sample. AND, tensile test samples are SUPER sensitive to stress risers. Grooves or pits too small to see can be a stress riser and weaken a sample A LOT. I suspect that's why your CNC samples gave weaker strengths than the printed ones. Have you looked to see how the CNC samples compared to published material strengths? Also, how are you calculating your elongation? I feel like if I can see it visually, it has to be more than 1.2%.
@@jackdeniston59 In the real world parts are affected by variable load and what matters is plastic and elastic deformation numbers. You can kind of see that in graphs but this info is useless.
@@jackdeniston59 A thousand different kinds of defects can occur in the real world, all affecting the results in wildly different ways. Doesn't mean the data represents real world performance, if you choose a sample with a random one of those defects and measure it.
Maybe he calculates the elongation after fracture by putting the two broken pieces back together and comparing it to original sample length (negating all elastic deformation).
@@smashyrashy what? Ender 3s have been available for many years now, if you're wanting something more professional grade, then a Prusa or (nowadays) a Bambu will do almost anything most people need. (Though a Bambu is way less open source and so likely won't last nearly as long as the others) What are you looking for exactly?
@@WeAreCheckingThey said "good quality" so Ender 3s don't really fit the bill. Nowadays on the lower end there's stuff like the SV06 which takes the form factor and robustness of a Prusa but makes the whole package far cheaper.
I don't know if anyone has mentioned this but it matters which way on the plate the part was machined from. There is an elongated grain structure from the rolling process. I tested this in my nail making video as I was making nails from sheared material and the strength drastically differs from 90 degrees on the original plate. Cheers J
Grain direction in some of the cnc materials plays a big part too. The titanium looks like it was perpendicular to the test load. The break seemed like it too.
@@Mernom I worked in a field making pressure rupture discs. We would cut slits in the material with a laser leaving tabs that break upon failure. Cutting parts in the wrong grain direction for the application yields vastly different results. I can only assume the same happens here.
It's nice seeing that characteristic stress-strain profile for the steel, even with the 3d printed one. It would be interesting if you could somehow constantly increase the load instead of pumping the jack.
Depending on what "tool steel" you had, you could have heat treated it to triple its performances (i don't mean annealing it as a heat treat, i mean hardening and tempering). Heat it to cherry red, and not any hotter or under that, drop in water, and then make its surface shine again with sandpaper or a stone, you just want to see the metal underneath. Then, with the torch, slowly heat it back up until the steel starts taking tempering colors. Reach an electric blue, avoid going above that too much, and below that will just be too brittle. Flame tempering is kind of a crude way of tempering usual carbon steel, but that's how it was done before. That being said, it's not pushing the material to its best capabilities, you'd need a tempering furnace for that.
And an age oven, control for temp, duration. Aluminum processing is a different though in the quench. You don't want it to severe you can get issues with grain structure. It also matters how think the aluminum.
6:24 This is definitely way higher than 500 C. Based on my limited experience melting brass and the glow, I would say the middle part is at least 900 degrees
Great tests! I don't think it was mentioned, but I assume the 3D printing process built these pieces "flat", right? I mean as opposed to "standing" like they are on the test rig. A major source of weakness in 3D prints is at the interface between layers, so I don't see how these parts could be this strong if the layers were horizontal when the piece is being stretched. By the way, it would be interesting to also compute the tensile strength you observed in Mpa (as Stefan from CNC Kitchen does), and compare these values to the ones reported by PCBWay. For example, they claim 560 Mpa for 316L stainless, 330 Mpa for aluminum, and 600 Mpa for titanium - all tensile strength numbers.
kind of like resin the properties are more homogenous, so the difference in strength due to the orientation it was printed in should be negligible. for the SLM printed metal parts
@@tony_mfg7597 Thanks for this comment, I never knew this since I've only ever owned FDM 3D printers, so I looked it up and it does seem to be the consensus. Very interesting property, I would not have expected it given that resin prints are still done layer by layer.
You can actually see the upskin in the parts. They were printed vertically. Metal does not suffer from FDM issues as such. The big impact is more in the grain structure and that is where the orienation has an impact. XY&Z will have different z-strength and elongation.
Please provide stress/strain curves or other useful data next time! Really appreciate the approach here but I'd really like to see more of the useful data
Two explanations for the results; One, the Alsimg10 has much higher tensile strength than 6061-T6 aluminium. And two, the CNC parts from PCB way cold be laser cut from a sheet, if they are, that could explain the results, aluminium would lose its temper and steels would precipitate carbon near the cut edge and become brittle.
@zomgthisisawesomelol usually in 3D print is used AIS10Mg (it Is written in the video) alloy wich is equivalent to something like EN 1706, 2010 aluminium grade.
Does metal, especially CNC or forged/hammered metal, have "directional grain/layers" like wood and FDM 3D-printed plastic does? If so, would that significantly affect tensile strength to the degree that plastic is affected or is it not noticeable enough, and is it not possible to ask a fabrication shop to CNC certain metal parts along a certain axis of the sheet if it is?
I second this. What's the strength to cost ratio? I been using sunlu pla+ for some things and it's surprisingly strong and their black is under $20 a roll.
Very interesting results here Curious what the cross sectional area of the test piece is? Could be more useful to have the numbers in MPa. An idea for a future video could be comparing the metals at varying infill levels to high performing plastics (peek, ultem...) For strength:weight and strength:volume comparisons
It would be very difficult that way once you start plastically deforming the test part. The necking would reduce the cross section. unless he had a way to keep track of that, maybe a cylindrical test piece?
@@MrBricks148 Mpa is always calculated with reference to the unstressed cross-sectional area. However calculating the poison ratios would be interesting regardless.
@@Zestybwoi not always. Tensile reports are in engineering stress as you’ve noted, but in most finite element model solvers (notably excepting NASTRAN) true stress vs strain is used. Tensile tests also give reduction in area and elongation values, as well. These datapoints combine to enable approximate generation of true stress vs strain curves.
I expect the CNC metals used here were low-grade billet. This means that the "grain" of the metal would be going all sorts of different directions rather than in straight lines. By printing flat, you are forcing the metal grain to lay a specific way, which is more or less making the printed result a kind of custom billet. By aligning the grain along the stressed plane, you've given the metal the greatest chance to demonstrate its tensile strength in billet form. I wonder if the carbon, kevlar, and glass linings could benefit the printed metals?
Most common alloys are not suitable for 3D printing, like 6061. The 3D printed aluminum part was most likely a much different alloy but cool to see they can be just as strong as a machined part these days.
The reason plastic 3d printed parts didn't change based on infill is because of how thin the test pieces were, not having very much empty space for infill to take up
Using Newtons as your unit for force is more helpful, instead of a static mass substitute (kg). You can still describe the static equivalent but then also show how dynamic forces can be huge, compared to the masses used. A very important lesson.
Here's a thought. The infil for the prints were placed at what I assume to be the optimal angle, criss crossing diagonally in the direction of the load. If it was printed at a 45 degree angle, it would no longer criss cross, and I imagine that might have an effect on the outcome.
Metal printing doesn't quite have the infill strategy implications you'd see in plastic printing, but yes, as-printed the material is typically slightly weaker in the build direction. After standard heat treatment though, the materials are pretty isotropic
Why people think that if you heat steel it gets hardened??? you need to heat it fast and cool it fast to harden it. You can also heat the surface with a high concentrated CO2 atmosphere too the canbon will make it harder too, but just heat ... noooooo. it is plain wrong!!
How did you measure the % elongation? I don't see any strain gauge nor was it explained in the video. Kg/time doesn't really tell us anything other than saying we can progressively load this specific part for this long before it breaks. Also, that's not ASTM D638 dogbone standard so you can't really say that the test has been "standardized". Pedantic, I know. But there's a fine difference between doing something that resembles a standardized test and doing an actual standardized test; I understand you're constrained by your test rig but I must point this out. Cool results though. I'd like to see an improvement in the experimental setup and eventually see an actual stress strain curve.
I see, it looks like elongation was measured in post. Which makes me question its credibility even more. Was the distortion from the camera lens compensated for?
The word standardised does not imply it follows a specific standard, just that it follows a standard. It just means that the data from all of his samples are comparable to each other (but not necessarily to other peoples, though I agree that this would have been better).
@helpmeimconfused Error due to lens distortion and other camera effects would not necessarily have been significant compared to other sources of error in the design. Regardless though, if he didn't move the camera much between samples, the values for each sample would have been comparable to each other at least.
Electrosync Great video! very cool to see results like this, one thing I'd suggest is looking into something like ASTM E8 for coupon designs and having larger radius from the grip sections to the gauge lengths, this focus stress into the gauge length and gives a more true uniaxial loading for measuring tensile strain, I'd also consider a bolted clamp mount onto solid grip tangs and have that mount be the point of connecting your test frame mount points to remove the KT factor of the giant hole influencing sample loading behaviour. you can then also mark the gauge length before testing and measure the distance between the marks after loading for a more accurate elongation measurement. Since you are recording the test with a camera you could also look into 2D optical displacement measurement of those marks as well.
This was incredebly useful and informative! I've not seen any one compare the propper metal prints to each other, let alone to CNC, this was really awesome. THANK YOU!
I'd imagine the higher strength is because the laser sintering is a high temp, fast cooling process that leaves a high strength heat treat. The lower density and inherit porosity would be more apparent with equal heat treats applied
Although others have criticised the heat treatment, not to matter as it's no big deal really - the most interesting element was the difference between 3D printed metals versus CNC machined metals - that was GOLD. Would love to know the cost comparison between the 3D printed versus CNC machined metal parts if it's not commercially sensitive.
A number of things come into play when doing tensile test. On the CNC they will break at corners and sharp internal corners focus stress. rounding and smoothing will help and any working of test structures can screw results. Learned this in a lost wax casting company. A cast part of same dimension will always put perform a CNC in the same material.
It depends on the process. If you're using the binder jet or something like that, they're pretty inferior. Powderbed processes though actually weld the materials, so it's like stacking a bunch of super tiny weld beads in a bunch of different directions. They can outperform billet in certain ways, but there will always be trade-offs
Something is wrong with the elongation calculation. It was very apparent in the Onyx FR + Glass test. I meassured it on the screen so it is not 100% accurate. At start the distance between the clamps was 77mm. When the first cracks started to appear the distance was 84mm. This is a 9,1% elongation. Also the aluminium alloys you worked with are very different. AlSi10Mg contains cca. 85-88% Al and 10% Si and a bit of Mg while 6061 contains 96-98% Al + 0,7% Si and 1% Mg. Tensile strength of the former is 450MPa while for the latter it is 180-280MPa. The idea is good but in order to get useful data you need to compare fairly similar things e.g. like the PLA and the titanium parts. I dont think base on this test alone we can conclude that 3D printed metal parts are stonger. Too many variables.
Metal heat treating has to happen at very specific temperatures to get the results you want. That D2 tool steel you used has an austenitization temperature of around 1030ºC, a lot higher than what the torch can give it. You didnt get a half-baked heat treat, you got no heat treat at all because you didn't get the steel hot enough to reset its microstructure. If anything, you may have actually softened it.
I would imagine that because the parts are so small, they're more susceptible to small variations. I would predict that the difference in strength would reduce as the test part size increases. Assuming the material is the same grade and all that jazz
This was good but as a student i had to do a few of these and had access to norms and the like. We had a protocol where we only took the results from samples that broke clearly in the middle section, your sample have a stress point at the neck, it should be longer if you want to prevent them from breaking at the stress point. Your results may be qualitative but not quantitative, you'd need an average over several good samples.
Thanks for the clarification. I thought that the S in SLM stood for sinter, same as with DLMS. Even so, it's a surprising result to me. I would expect a part machined from billet to be stronger than anything cast or welded - I would expect it to come second only to forged.
@@Syscrush no problem. Most of the og metal printer companies are based in Germany. From what I have been told by them, DMLS is actually somewhat the result of a poor translation. DMLS is actually fully melted as well, but the industry didn't want to refer to the process as SLM because that is also the name of one of the competitive printers brands. Because of the confusion around this, the industry now more commonly refers to the process as Laser Powder Bed Fusion (LPBF). Not every printed material is stronger than billet, and it's still a tradeoff between properties. You can heat treat to beat the ultimate tensile strength of billet but you trade away elongation, for instance
Would be nice to see this test with cast metal samples as well, specifically for aluminum. That’s what most applications are replacing with 3D printed metal parts. Also would be helpful to know the alloys of aluminum used and the stock thickness sizes, if the samples were made from thicker aluminum plate and cut down, they could very easily have a lower hardness. In aerospace this is something that we check after the parts are fully machined from a plate to ensure that there wasn’t a soft spot in the center of the block and that it was tempered properly.
Thanks for making a video that compares metal 3d printed parts to conventional CNC metal parts. Seems like the trade off is between elongation and ultimate strength. Makes sense since I would imagine grain size of SLS parts is much smaller due to quickly cooling the parts from melted to solid below glowing hot. Your video helped me realize this and that 3d printed metal parts are actually viable for normal applications
There's a clever way to measure toughness of a material, requiring a *heavy* hammer, a protractor, a set starting position, and a pointer to capture swing-through. The less height the hammer gets on the other side of the swing, the tougher the material.
It makes sense. If you’re printing your metal, you can somewhat improve the molecular geometry to allow for the highest strength facing the work, sort of like a blacksmith using billets with alternating grains to strengthen their product. I can only imagine how this tech will fair in a couple decades or even a century.
I'd love to see comparisons of more practical tests; tests that reflect the intended uses of 3D printed metals. Like how a 3D printed piston for a combustion engine, or a 3D printed heat exchanger, compares to one that's been cast or forged.
Surprised to see the SLM handles the static test pretty well! I wonder if they are also good on the dynamic loads test. In my experience of engineering, porosity in metal parts doesn’t make a huge difference in static but on on the dynamic test, it can differ 10 to 100 times!
It doesn't really matter how fast you load the sample. It is more interesting to look at the tension in the material. What are the cross sections in the samples? I would like to see a graph of stress versus strain. Even if it is absolute. The truth, if possible, is relative.
Any idea how this is even possible. I get some of it might be a fluke due to different alloys and work hardening. But how can a more porous unevenly stressed version of a Material be stronger in tensile stress
Different Aluminium alloys can behave quite differently. This looks to be one that in german we sometimes call "gooey aluminium". Horrible to machine because it's so soft, it's like cutting peanut butter with a sharp knife.
The grain orientation for 3d printed parts is probably more uniform and the cnc parts most likely arent. I think this would also explain the higher elongation on the cnc parts. I couldn't imagine 3d printed parts to be stronger than a properly forged metal
Cool setup. But most of the time I'm not interested in the ultimate strength, cause if my part has deformed permanently it might not function anymore. The behaviour of all these materials are very different but some kind of "yield strength equivalent" would probably be more useful for me at least.
I do have a theory on why all the CNC parts performed worse. Imperfections on the surface finish - it's clearly explained in every manual that the surface for a test sample should be rectified. I know that this would increase the cost of the samples, but it would give us a more reliable indication of whether they perform worse simply because the 3D samples are just better or not.
After bad experiences with pistol firing pins... this printed titanium seems interesting. Some people say titanium can be filed, firing pins can't. Interesting, I'll study more.
I like how much the steel yields before breaking. Very useful for many applications I was also really surprised how weak that cnc machined aluminium was. Also the 3D Printed titanium was very impressive. The heat treatment was pointless though. You can't just point a torch at D2 and hope it gets stronger.
So what about a part that is cnc machined from a 3D printed block? It would be interesting to see if the difference comes from the machining, or from the base material which of course has had a different heat treatment history if it was 3D printed.
im not a metallurgist but my bet for why the 3d printed parts where stronger was because of the crystal structure in 3d printing it is layer after layer but cnc can be in every orientation mainly because of the way it was made - just for clarification 3d printing is done in layers and each layer is like a different part that is stuck to the ones around it if you realy want to test 3d prints you need to print them is several orientations
had the aluminum steel and titanium been machine hardend or not? Further more i believe that the laser sintering will likely have affected the grain size and structure of the metals
I'm curious as to how forged parts would compare. I'm sure they'd do better than the CNC parts, at least, if they were forged properly, due to the grain structure, but I wonder how they would do against 3d printed ones...
Another issue, along the many that I’ve seen about the cnc cut parts , is that grain direction is of paramount importance when testing to ultimate yield strength. As well as stress induced machining practices, stress risers, sharp edges, and lot variations. IE; where did you get the material from? Etc…. This seems like an experiment to “prove” 3d printing is the only way things should be done.
I think there was something wrong with the aluminium and stainless parts. They shouldn't be so weak (relatively). Of course they might not suit the SLM process so well without a heat treat. Oh and the heating you did for the tool steel is most likely just yielding the part. And I don't know much about titanium but the test might be bias for materials that don't suffer from work hardening since the force is pulsing.
Id love to see a repetitive stress test of median stress over a ton of reps to see if the pieces get weaker/more brittle over the tests. Because if cheap plastics/metals perform the same as expensive stuff over multiple cycles then it would make sense to just mass produce cheap parts for replacement rather than try and get top shelf stuff one time.
Also all samples must be pulled at the same and constant rate. The rate can effect the failure significantly. Try reading how an Instron test machine works
Bypassing that single item testing or the specific setup are far from ideal, were the SLM/CNC alloys the same? When we say CNC is it machined, waterjet ot laser? Were they deburred? In such small parts cutting process can have a big impact.
I'm a CNC machinist by trade, and I have definitely noticed a severe drop in the quality of metals in the past 2 years, I did not have any idea that they were so poorly made that they were weaker than 3d printed parts. I'll see if I can find some old steel, stainless and aluminium that I can make into test pieces if you would like Just send me a step file
That's really interesting with the 3D printed parts being stronger, I think it may be due to the printing process harding the material but that's a guess
Or like another comment pointed out, too small sample size. But what it did give me an idea about is that 3D printing might give more consistent results, while CNC is stuck with whatever metal they are provided. If it has any flaws CNC isn't going to fix any of that.
Very good. It would take a lot of parts,sensors, time and everything else to get enough data to satisfy everyone, but you showed how the materials and processes compare. I'm surprised that the printed parts were stronger than the machined.
How do milled/CNC plastic parts perform in comparison to printed? Probably difficult to test for considering the abundance of different plastics available. Also, may be worth looking at forged parts.
I know that 6061 aluminum is not very strong its mostly for corrosion resistance. You should try 7075-T6 of 2024-T4. also those alloy has heat treatment that make them a lot stronger. Heat treating aluminum is more difficult to perform than steel and should be done by the factory if possible.
Very interesting l had no idea 3D printed metals were so good. But being able to use extremely pure and good quality material when printing and having a consistent material cell structure and also being able to play with wall thicknesses and infills obviously makes a big difference. I remember doing gravity die casting and the end results sometimes left a lot to be desired. Interesting results with some surprises. The tolerances and surface finishes that can be achieved with a CNC are still superior but 3D printing definitely has it's place. Traditional casting is also something I would like to see tested. 😎👍
I wonder if there could have been a process bias. Some articles I've read suggest one over the other, but if 3D printed is consistent then it's likely the better of the two. What this does say is pcbway has better metal 3d printing strength with this run as opposed to CNC machined. However, it does make me wonder if there was maybe a tolerance or weight bias that occurred. What's the margin of error here?
The "heat treat" you did likely did nothing, if anything softened the parts. Heat treating steel requires specific temperatures and specific times for cooling to create the proper grain structure. It's followed by an annealing process to dial back the hardness to a specific number.
I would recommend redoing those tests and obtaining pieces that have been properly heat treated to a specific hardness.
I was surprised on the 3d printed part strength.
Yeah, basically invalidated the whole test since regular 4140, when heat treated, has a yield 200MPa+ that of Ti6Al4V
It's all pointless unless multiple samples of the same materials and treatments etc. are tested, we have no idea if these parts are representative of the average expected strength or not, what the standard deviation is etc.
I'm a tool and die maker and know by accident a bit about heat treatment. By heating up the parts to about 500°C you annealed them and reduced any existing hardnes. D2 fpr example needs a temperature of 1020-1050°C, followed by a chill in oil or moved nitrogen gas and has to be annealed at about 500°C afterwards for several hours.
Yep, there's a lot of weirdness going around there in general. He manages to heat the part to a nice orange in the centre (one spot even reaching yellow) which would suggest temperatures of >1000degC. Steel at 500 degrees shouldn't even be reddish. Which would be decent temperatures for a heat treat, except that it should then be chilled quickly (oil quenching recommended for 4140). If it gets below tempering temperatures too slowly you'll just end up with an annealed steel again. He might've actually gotten lucky with the accidental heat treat as the thinness of the part causes it to cool down really quickly. But eh, that's rather unlikely.
haha i was about to comment this myself bravo im surprised anyone else thought of it
The kg/time graph is not really useful. Kg/elongation would give way more info on the stiffness
I couldn’t include all of the data in the video because the algorithm would punish the video. All the data is available on my Patreon. The link is in the description.
@@electrosync You mean it would have been too long?
@@1fareast14 and now he can safely hide the answer behind a paywall
@@oli5dijksma616well he's not publically funded unlike the researchers who are and paywall all their findings, also he's not using strain gauges on the thin part of the material which is the only strain you actually need to care about
@@cybyrd9615 the researchers aren’t the ones paywalling the publishers are. Literally just ask the researchers personally for a copy of their paper, they’re allowed to provide it for free.
I understand you are doing your best, but . . . .
As someone who works in materials testing, testing a single sample really isn't enough to draw many conclusions. You could just have a bad sample.
AND, tensile test samples are SUPER sensitive to stress risers. Grooves or pits too small to see can be a stress riser and weaken a sample A LOT. I suspect that's why your CNC samples gave weaker strengths than the printed ones. Have you looked to see how the CNC samples compared to published material strengths?
Also, how are you calculating your elongation? I feel like if I can see it visually, it has to be more than 1.2%.
Yeah, there is definitely something going on with those elongation numbers
Although, it does appear that the whole setup moves when it is cranked
Yeah, but that is how they are used in the real world.
@@jackdeniston59 In the real world parts are affected by variable load and what matters is plastic and elastic deformation numbers. You can kind of see that in graphs but this info is useless.
@@jackdeniston59 A thousand different kinds of defects can occur in the real world, all affecting the results in wildly different ways. Doesn't mean the data represents real world performance, if you choose a sample with a random one of those defects and measure it.
Maybe he calculates the elongation after fracture by putting the two broken pieces back together and comparing it to original sample length (negating all elastic deformation).
I hope we get consumer-grade metal printers soon.
Or just cheap good quality 3d printers
That would be amazing! I think we’re a little while away from that though.
@@smashyrashy what? Ender 3s have been available for many years now, if you're wanting something more professional grade, then a Prusa or (nowadays) a Bambu will do almost anything most people need. (Though a Bambu is way less open source and so likely won't last nearly as long as the others)
What are you looking for exactly?
@@WeAreCheckingThey said "good quality" so Ender 3s don't really fit the bill. Nowadays on the lower end there's stuff like the SV06 which takes the form factor and robustness of a Prusa but makes the whole package far cheaper.
Just need a furnace-y enclosure, a tungsten nozzle, and an induction heater coil.
I don't know if anyone has mentioned this but it matters which way on the plate the part was machined from. There is an elongated grain structure from the rolling process. I tested this in my nail making video as I was making nails from sheared material and the strength drastically differs from 90 degrees on the original plate. Cheers J
Grain direction in some of the cnc materials plays a big part too. The titanium looks like it was perpendicular to the test load. The break seemed like it too.
True. This would probably require further testing to discard or confirm it as a variable.
@@Mernom I worked in a field making pressure rupture discs. We would cut slits in the material with a laser leaving tabs that break upon failure. Cutting parts in the wrong grain direction for the application yields vastly different results. I can only assume the same happens here.
Also the difference between billet and forged parts in the case of aluminum
Grain direction in metal is negligible, which is why you don't care about orientation of raw material when CNCing parts. Metal is isotropic.
From my research, albeit quick, the grain direction has little to no effect here.
It does, however, play a huge role in *bending* metal.
It's nice seeing that characteristic stress-strain profile for the steel, even with the 3d printed one. It would be interesting if you could somehow constantly increase the load instead of pumping the jack.
A stepper with gears xD
It’s not a Young’s modulus graph though so I think it’s just a coincidence
@@angrydragonslayerthe much more obvious and practical solution is an electric hydraulic pump.
@@JacobLeeson-zk1ol rude
Depending on what "tool steel" you had, you could have heat treated it to triple its performances (i don't mean annealing it as a heat treat, i mean hardening and tempering). Heat it to cherry red, and not any hotter or under that, drop in water, and then make its surface shine again with sandpaper or a stone, you just want to see the metal underneath. Then, with the torch, slowly heat it back up until the steel starts taking tempering colors. Reach an electric blue, avoid going above that too much, and below that will just be too brittle. Flame tempering is kind of a crude way of tempering usual carbon steel, but that's how it was done before. That being said, it's not pushing the material to its best capabilities, you'd need a tempering furnace for that.
And an age oven, control for temp, duration. Aluminum processing is a different though in the quench. You don't want it to severe you can get issues with grain structure. It also matters how think the aluminum.
Well I didn’t expect to see that result… thanks for showing this👍
6:24 This is definitely way higher than 500 C. Based on my limited experience melting brass and the glow, I would say the middle part is at least 900 degrees
It’d be interesting if you could test metal cast from 3D prints, as that’s a DIY way of “3D printing” metal. Maybe a collab?
or even forged, but i dont know where he could get forged parts
@@Fantastika would probably need to compare to a commercially available forged part, e.g. a moto brake lever.
no its not, thats just casting
Great tests! I don't think it was mentioned, but I assume the 3D printing process built these pieces "flat", right? I mean as opposed to "standing" like they are on the test rig. A major source of weakness in 3D prints is at the interface between layers, so I don't see how these parts could be this strong if the layers were horizontal when the piece is being stretched. By the way, it would be interesting to also compute the tensile strength you observed in Mpa (as Stefan from CNC Kitchen does), and compare these values to the ones reported by PCBWay. For example, they claim 560 Mpa for 316L stainless, 330 Mpa for aluminum, and 600 Mpa for titanium - all tensile strength numbers.
kind of like resin the properties are more homogenous, so the difference in strength due to the orientation it was printed in should be negligible. for the SLM printed metal parts
3d printer metal is kinda different... it's almost more like simultaneously welding the entire part from powder.
@@tony_mfg7597 Thanks for this comment, I never knew this since I've only ever owned FDM 3D printers, so I looked it up and it does seem to be the consensus. Very interesting property, I would not have expected it given that resin prints are still done layer by layer.
@@desmond-hawkins no problem
You can actually see the upskin in the parts. They were printed vertically. Metal does not suffer from FDM issues as such. The big impact is more in the grain structure and that is where the orienation has an impact. XY&Z will have different z-strength and elongation.
The results were a little surprising! Any ideas for a future strength test video?
Test casted parts
@@smashyrashy Great idea! I'd need to do some research into that.
Please provide stress/strain curves or other useful data next time! Really appreciate the approach here but I'd really like to see more of the useful data
Two explanations for the results; One, the Alsimg10 has much higher tensile strength than 6061-T6 aluminium.
And two, the CNC parts from PCB way cold be laser cut from a sheet, if they are, that could explain the results, aluminium would lose its temper and steels would precipitate carbon near the cut edge and become brittle.
@zomgthisisawesomelol usually in 3D print is used AIS10Mg (it Is written in the video) alloy wich is equivalent to something like EN 1706, 2010 aluminium grade.
Your „heat treatment“ was interesting. Rest of the video: awesome 👍
Does metal, especially CNC or forged/hammered metal, have "directional grain/layers" like wood and FDM 3D-printed plastic does? If so, would that significantly affect tensile strength to the degree that plastic is affected or is it not noticeable enough, and is it not possible to ask a fabrication shop to CNC certain metal parts along a certain axis of the sheet if it is?
How much were the different parts? It would be interesting to see which printed material gets you the most strength per dollar.
I second this. What's the strength to cost ratio? I been using sunlu pla+ for some things and it's surprisingly strong and their black is under $20 a roll.
Very interesting results here
Curious what the cross sectional area of the test piece is? Could be more useful to have the numbers in MPa.
An idea for a future video could be comparing the metals at varying infill levels to high performing plastics (peek, ultem...) For strength:weight and strength:volume comparisons
It would be very difficult that way once you start plastically deforming the test part. The necking would reduce the cross section. unless he had a way to keep track of that, maybe a cylindrical test piece?
@@MrBricks148 Mpa is always calculated with reference to the unstressed cross-sectional area. However calculating the poison ratios would be interesting regardless.
@@Zestybwoi not always. Tensile reports are in engineering stress as you’ve noted, but in most finite element model solvers (notably excepting NASTRAN) true stress vs strain is used. Tensile tests also give reduction in area and elongation values, as well. These datapoints combine to enable approximate generation of true stress vs strain curves.
I prefer to do it in psi. 😅 but yes, cross-sectional is key. Especially when looking at your offset and your proportional limit.
@@MrBricks148you need to measure it when you do the calculations for your offset. It's how you get compliance to astm😢 tensile testing standard.
I expect the CNC metals used here were low-grade billet. This means that the "grain" of the metal would be going all sorts of different directions rather than in straight lines. By printing flat, you are forcing the metal grain to lay a specific way, which is more or less making the printed result a kind of custom billet. By aligning the grain along the stressed plane, you've given the metal the greatest chance to demonstrate its tensile strength in billet form. I wonder if the carbon, kevlar, and glass linings could benefit the printed metals?
The sharper edges on the cnc cut parts are the likely culprit of lower fail strength. Those edges will tear first and propagate
Very, very, very interesting, and with some surprises for me. Thanks also for the strength to weight ratio chart at the end.
Most common alloys are not suitable for 3D printing, like 6061. The 3D printed aluminum part was most likely a much different alloy but cool to see they can be just as strong as a machined part these days.
Most likely AlSi10Mg or F357. There are 6061 substitutes being developed pretty heavily right now too though
A chart of strength to cost would be nice.
The reason plastic 3d printed parts didn't change based on infill is because of how thin the test pieces were, not having very much empty space for infill to take up
Using Newtons as your unit for force is more helpful, instead of a static mass substitute (kg). You can still describe the static equivalent but then also show how dynamic forces can be huge, compared to the masses used. A very important lesson.
There I am on the wall! 🩷
Check out the stretch going up and down on the aluminium before it breaks!
Here's a thought. The infil for the prints were placed at what I assume to be the optimal angle, criss crossing diagonally in the direction of the load. If it was printed at a 45 degree angle, it would no longer criss cross, and I imagine that might have an effect on the outcome.
Metal printing doesn't quite have the infill strategy implications you'd see in plastic printing, but yes, as-printed the material is typically slightly weaker in the build direction. After standard heat treatment though, the materials are pretty isotropic
Why people think that if you heat steel it gets hardened??? you need to heat it fast and cool it fast to harden it. You can also heat the surface with a high concentrated CO2 atmosphere too the canbon will make it harder too, but just heat ... noooooo. it is plain wrong!!
Is there info on less expensive fdm materials with chopped fibers? These continuous fibers are insanely expensive
How did you measure the % elongation? I don't see any strain gauge nor was it explained in the video. Kg/time doesn't really tell us anything other than saying we can progressively load this specific part for this long before it breaks. Also, that's not ASTM D638 dogbone standard so you can't really say that the test has been "standardized". Pedantic, I know. But there's a fine difference between doing something that resembles a standardized test and doing an actual standardized test; I understand you're constrained by your test rig but I must point this out.
Cool results though. I'd like to see an improvement in the experimental setup and eventually see an actual stress strain curve.
I see, it looks like elongation was measured in post. Which makes me question its credibility even more. Was the distortion from the camera lens compensated for?
The word standardised does not imply it follows a specific standard, just that it follows a standard. It just means that the data from all of his samples are comparable to each other (but not necessarily to other peoples, though I agree that this would have been better).
@helpmeimconfused Error due to lens distortion and other camera effects would not necessarily have been significant compared to other sources of error in the design. Regardless though, if he didn't move the camera much between samples, the values for each sample would have been comparable to each other at least.
Thank you so much for sharing. I’m an engineer and found your tests very valuable.
Love your videos! Can't believe I only now discovered your channel
Electrosync Great video! very cool to see results like this, one thing I'd suggest is looking into something like ASTM E8 for coupon designs and having larger radius from the grip sections to the gauge lengths, this focus stress into the gauge length and gives a more true uniaxial loading for measuring tensile strain, I'd also consider a bolted clamp mount onto solid grip tangs and have that mount be the point of connecting your test frame mount points to remove the KT factor of the giant hole influencing sample loading behaviour.
you can then also mark the gauge length before testing and measure the distance between the marks after loading for a more accurate elongation measurement.
Since you are recording the test with a camera you could also look into 2D optical displacement measurement of those marks as well.
This was incredebly useful and informative! I've not seen any one compare the propper metal prints to each other, let alone to CNC, this was really awesome. THANK YOU!
That was so useful and informative. Thanks!!
I'd suggest fatigue testing, though without some automation that will be a very tedious set of tests!
Were the aluminum cnc and printed the same qlloy?
It would be interesting to compare forged vs cast vs additive vs cnc metal
I'd imagine the higher strength is because the laser sintering is a high temp, fast cooling process that leaves a high strength heat treat. The lower density and inherit porosity would be more apparent with equal heat treats applied
Good video, a lot of hate but you did this for free and kept this concise. Thank you.
Although others have criticised the heat treatment, not to matter as it's no big deal really - the most interesting element was the difference between 3D printed metals versus CNC machined metals - that was GOLD.
Would love to know the cost comparison between the 3D printed versus CNC machined metal parts if it's not commercially sensitive.
A number of things come into play when doing tensile test. On the CNC they will break at corners and sharp internal corners focus stress. rounding and smoothing will help and any working of test structures can screw results. Learned this in a lost wax casting company. A cast part of same dimension will always put perform a CNC in the same material.
I was not expecting that. I thought 3D printed metals were way less durable.
It depends on the process. If you're using the binder jet or something like that, they're pretty inferior. Powderbed processes though actually weld the materials, so it's like stacking a bunch of super tiny weld beads in a bunch of different directions. They can outperform billet in certain ways, but there will always be trade-offs
Something is wrong with the elongation calculation. It was very apparent in the Onyx FR + Glass test. I meassured it on the screen so it is not 100% accurate.
At start the distance between the clamps was 77mm. When the first cracks started to appear the distance was 84mm. This is a 9,1% elongation.
Also the aluminium alloys you worked with are very different. AlSi10Mg contains cca. 85-88% Al and 10% Si and a bit of Mg while 6061 contains 96-98% Al + 0,7% Si and 1% Mg.
Tensile strength of the former is 450MPa while for the latter it is 180-280MPa.
The idea is good but in order to get useful data you need to compare fairly similar things e.g. like the PLA and the titanium parts. I dont think base on this test alone we can conclude that 3D printed metal parts are stonger. Too many variables.
Metal heat treating has to happen at very specific temperatures to get the results you want. That D2 tool steel you used has an austenitization temperature of around 1030ºC, a lot higher than what the torch can give it. You didnt get a half-baked heat treat, you got no heat treat at all because you didn't get the steel hot enough to reset its microstructure. If anything, you may have actually softened it.
I would imagine that because the parts are so small, they're more susceptible to small variations.
I would predict that the difference in strength would reduce as the test part size increases.
Assuming the material is the same grade and all that jazz
This was good but as a student i had to do a few of these and had access to norms and the like. We had a protocol where we only took the results from samples that broke clearly in the middle section, your sample have a stress point at the neck, it should be longer if you want to prevent them from breaking at the stress point. Your results may be qualitative but not quantitative, you'd need an average over several good samples.
I admit to being shocked at these results - it's hard to understand how sintered metal could outperform machines pieces.
Thanks for this!
It's not sintered. It's fully melted/welded
Thanks for the clarification. I thought that the S in SLM stood for sinter, same as with DLMS.
Even so, it's a surprising result to me. I would expect a part machined from billet to be stronger than anything cast or welded - I would expect it to come second only to forged.
@@Syscrush no problem. Most of the og metal printer companies are based in Germany. From what I have been told by them, DMLS is actually somewhat the result of a poor translation. DMLS is actually fully melted as well, but the industry didn't want to refer to the process as SLM because that is also the name of one of the competitive printers brands. Because of the confusion around this, the industry now more commonly refers to the process as Laser Powder Bed Fusion (LPBF). Not every printed material is stronger than billet, and it's still a tradeoff between properties. You can heat treat to beat the ultimate tensile strength of billet but you trade away elongation, for instance
Would be nice to see this test with cast metal samples as well, specifically for aluminum. That’s what most applications are replacing with 3D printed metal parts.
Also would be helpful to know the alloys of aluminum used and the stock thickness sizes, if the samples were made from thicker aluminum plate and cut down, they could very easily have a lower hardness. In aerospace this is something that we check after the parts are fully machined from a plate to ensure that there wasn’t a soft spot in the center of the block and that it was tempered properly.
Thanks for making a video that compares metal 3d printed parts to conventional CNC metal parts. Seems like the trade off is between elongation and ultimate strength. Makes sense since I would imagine grain size of SLS parts is much smaller due to quickly cooling the parts from melted to solid below glowing hot. Your video helped me realize this and that 3d printed metal parts are actually viable for normal applications
you should make a rig that would test how brittle 3d printed metals are in comparison to CNC machined
its all about grain structure. A properly heat treated hammer forged machined part will always be stronger than printed, cast, or billet.
There's a clever way to measure toughness of a material, requiring a *heavy* hammer, a protractor, a set starting position, and a pointer to capture swing-through. The less height the hammer gets on the other side of the swing, the tougher the material.
It makes sense. If you’re printing your metal, you can somewhat improve the molecular geometry to allow for the highest strength facing the work, sort of like a blacksmith using billets with alternating grains to strengthen their product. I can only imagine how this tech will fair in a couple decades or even a century.
Nice to see points, where elastic deformation changed to plastic.
You should compute ratio in similar videos to follow.
Great job.
very interesting test, I think it would be also interesting if you would add a price on how much did each pice cost you to produce/buy it.
I'd love to see comparisons of more practical tests; tests that reflect the intended uses of 3D printed metals. Like how a 3D printed piston for a combustion engine, or a 3D printed heat exchanger, compares to one that's been cast or forged.
3D printing is a form of CNC.
Surprised to see the SLM handles the static test pretty well!
I wonder if they are also good on the dynamic loads test.
In my experience of engineering, porosity in metal parts doesn’t make a huge difference in static but on on the dynamic test, it can differ 10 to 100 times!
Definitely true, same with fatigue, but porosity levels in as-printed metal parts are typically
This was like watching the Hydraulic Press channel but in reverse.
It doesn't really matter how fast you load the sample. It is more interesting to look at the tension in the material. What are the cross sections in the samples? I would like to see a graph of stress versus strain. Even if it is absolute. The truth, if possible, is relative.
I'd be happy to heat treat the steel pieces for you if you want to redo this
I would love to see Fracture Toughness and/or Charpy testing on SLM vs CNC parts.
Any idea how this is even possible. I get some of it might be a fluke due to different alloys and work hardening. But how can a more porous unevenly stressed version of a Material be stronger in tensile stress
I can only image that it's a different alloy
Different Aluminium alloys can behave quite differently. This looks to be one that in german we sometimes call "gooey aluminium". Horrible to machine because it's so soft, it's like cutting peanut butter with a sharp knife.
The grain orientation for 3d printed parts is probably more uniform and the cnc parts most likely arent. I think this would also explain the higher elongation on the cnc parts. I couldn't imagine 3d printed parts to be stronger than a properly forged metal
Not too surprising typically these 3d printed metals aren’t just pure metal alot of them have other additives to increase strength and print ability
Kg/time graph is interesting for the eye but the tensile load [Mpa][N/mm^2]/Elongation graph would give us more information about material.
My favourite part was the music fill during the 316 SLM description 😅
Great video, consider adding a summary of your tests.
Cool setup.
But most of the time I'm not interested in the ultimate strength, cause if my part has deformed permanently it might not function anymore.
The behaviour of all these materials are very different but some kind of "yield strength equivalent" would probably be more useful for me at least.
I do have a theory on why all the CNC parts performed worse. Imperfections on the surface finish - it's clearly explained in every manual that the surface for a test sample should be rectified.
I know that this would increase the cost of the samples, but it would give us a more reliable indication of whether they perform worse simply because the 3D samples are just better or not.
I'd love to see the difference between a steadily increasing force vs this jerky application.
Great video either way!
After bad experiences with pistol firing pins... this printed titanium seems interesting. Some people say titanium can be filed, firing pins can't. Interesting, I'll study more.
I like how much the steel yields before breaking. Very useful for many applications
I was also really surprised how weak that cnc machined aluminium was.
Also the 3D Printed titanium was very impressive.
The heat treatment was pointless though. You can't just point a torch at D2 and hope it gets stronger.
in a sample size of one its dangerous to draw such conclusions.
It depends on many factors, you don't get around testing your final part anyway.@@krusher74
So what about a part that is cnc machined from a 3D printed block? It would be interesting to see if the difference comes from the machining, or from the base material which of course has had a different heat treatment history if it was 3D printed.
im not a metallurgist but my bet for why the 3d printed parts where stronger was because of the crystal structure
in 3d printing it is layer after layer but cnc can be in every orientation mainly because of the way it was made
- just for clarification 3d printing is done in layers and each layer is like a different part that is stuck to the ones around it
if you realy want to test 3d prints you need to print them is several orientations
Hey, could you test resins? both 3d printed and standard epoxy resins? maybe even ones with additives
had the aluminum steel and titanium been machine hardend or not? Further more i believe that the laser sintering will likely have affected the grain size and structure of the metals
I'm curious as to how forged parts would compare. I'm sure they'd do better than the CNC parts, at least, if they were forged properly, due to the grain structure, but I wonder how they would do against 3d printed ones...
Another issue, along the many that I’ve seen about the cnc cut parts , is that grain direction is of paramount importance when testing to ultimate yield strength. As well as stress induced machining practices, stress risers, sharp edges, and lot variations. IE; where did you get the material from? Etc…. This seems like an experiment to “prove” 3d printing is the only way things should be done.
This video totally earned my New Subscription.
Well done ha!
Theres a lot to be said regarding nano structures!
Not to mention laser fusion welding.
Really cool but I wish you had included pricing
Which would be best printed materials for a printed gun
I think there was something wrong with the aluminium and stainless parts. They shouldn't be so weak (relatively). Of course they might not suit the SLM process so well without a heat treat. Oh and the heating you did for the tool steel is most likely just yielding the part. And I don't know much about titanium but the test might be bias for materials that don't suffer from work hardening since the force is pulsing.
Id love to see a repetitive stress test of median stress over a ton of reps to see if the pieces get weaker/more brittle over the tests. Because if cheap plastics/metals perform the same as expensive stuff over multiple cycles then it would make sense to just mass produce cheap parts for replacement rather than try and get top shelf stuff one time.
Great Work! 3D is susceptible to fatigue failure. Would be good to compare the fatigue life of 3d metals and CNC metal parts.
Also all samples must be pulled at the same and constant rate. The rate can effect the failure significantly. Try reading how an Instron test machine works
Where is the weight of the 3d printed ones? I saw the weight of the metal but couldn’t see or missed the 3d printed ones
Did you measure the test samples to verify they were equally thick and wide? If so, the printed parts being stronger is amazing.
Amazing video, especially since most of the information I've seen suggest 3D printed metals are weaker than molded ones. 👍👍👍
Bypassing that single item testing or the specific setup are far from ideal, were the SLM/CNC alloys the same? When we say CNC is it machined, waterjet ot laser? Were they deburred? In such small parts cutting process can have a big impact.
I'm a CNC machinist by trade, and I have definitely noticed a severe drop in the quality of metals in the past 2 years, I did not have any idea that they were so poorly made that they were weaker than 3d printed parts.
I'll see if I can find some old steel, stainless and aluminium that I can make into test pieces if you would like
Just send me a step file
could hand pan instruments be 3d printed ? would the notes come out right ?
What is a stronger manufacturing method than CNCd? Forged?
That's really interesting with the 3D printed parts being stronger, I think it may be due to the printing process harding the material but that's a guess
Or like another comment pointed out, too small sample size. But what it did give me an idea about is that 3D printing might give more consistent results, while CNC is stuck with whatever metal they are provided. If it has any flaws CNC isn't going to fix any of that.
Very good.
It would take a lot of parts,sensors, time and everything else to get enough data to satisfy everyone, but you showed how the materials and processes compare.
I'm surprised that the printed parts were stronger than the machined.
How do milled/CNC plastic parts perform in comparison to printed? Probably difficult to test for considering the abundance of different plastics available. Also, may be worth looking at forged parts.
Plastic parts usually aren’t milled. They’re usually injection molded.
I know that 6061 aluminum is not very strong its mostly for corrosion resistance. You should try 7075-T6 of 2024-T4. also those alloy has heat treatment that make them a lot stronger. Heat treating aluminum is more difficult to perform than steel and should be done by the factory if possible.
That was unexpected. I would not have bet money on that.
Very interesting l had no idea 3D printed metals were so good.
But being able to use extremely pure and good quality material when printing and having a consistent material cell structure and also being able to play with wall thicknesses and infills obviously makes a big difference.
I remember doing gravity die casting and the end results sometimes left a lot to be desired.
Interesting results with some surprises.
The tolerances and surface finishes that can be achieved with a CNC are still superior but 3D printing definitely has it's place.
Traditional casting is also something I would like to see tested.
😎👍
Try 7000 series aluminium, it’s got a much higher tensile strength, which makes me wonder what grade the 3D printed ally part was?
I’d like to see Radiation Shield Tungsten PLA tested.
I wonder if there could have been a process bias. Some articles I've read suggest one over the other, but if 3D printed is consistent then it's likely the better of the two. What this does say is pcbway has better metal 3d printing strength with this run as opposed to CNC machined. However, it does make me wonder if there was maybe a tolerance or weight bias that occurred. What's the margin of error here?