I think there's 2 important points on your setup. Firstly, those cheaper straingauge devices poll the straingauge quite infrequently (a couple of hz at most). That can influence your results significantly for very stiff material, or if you crank it fast. Secondely, this doesn't tell much about how 'strong' a material is. It currently merely gives you the ultimate strength. In practice, you wouldn't want to reach this point at all, and stay below the yield strength (the point at which a material starts to permanently deform). It also doesnt say anything about how tough a material is. A material can have a relatively low yield strength and ultimate strength, but still be very tough by being very low stiffness. I'd recommend putting a distance sensor between the two jaws, and building a measuring setup to measure the deflection versus loadcell force so you can plot these together. That will allow you to make stress-strain curves which give you much more useful information to base decisions on, and to compare materials.
I was thinking the same thing. If he's using 3D printed pieces to hold the samples, it might be better to have the distance sensor between the pins (or whatever he goes to) that secure the samples so it removes any deflection of the clamping system, rocking of the vise jaws and load cell. If you're not just manually taking data points, it does turn into a larger project to do the data acquisition though. I'd use an NI CDAQ system with Matlab, but that's at least $1k from used ebay prices.
@@JohnSheerin There are cheaper data acquisition alternatives, for example the NI MyDAW which is under 300$ and has similar integration with matlab. Although honestly building a simple arduino based logger that reads out the loadcell and a distance sensor, and spits it over a serial port isn't that much more work. Just stringing together a bunch of libraries.
For thermoplastics, the ultimate strength is reasonable approximation of yield strength, because strain hardening is countered by a high degree of area reduction (e.g. necking, crazing).
I'm picturing a pocket where the end of the sample drops into. No need to secure it, gravity holds it in place until the pulling force takes over. Not sure how to make the edges of the pocket strong enough to deal with the pulling force, but I'm guessing round edges will do the trick. Gotta say, I love the overall idea. It brings this kind of testing into the attainable realm for many more people.
@@millerbenschop9666 With plastic parts, I would not rely on gravity alone. Don't need clamping force to hold the part against tension, just a top cover of some type that would serve to prevent either end of the wishbone from popping out of the corresponding pocket as the test piece is placed under tension.
I see that at least a few of us have very similar ideas. I look forward to seeing them tried (or, for James to tell us why he chose not to try them, which I'm sure would also be informative) in a future video. :)
It was fun to watch you discover why tensile testers use clamps. :) Seriously I love watching people learn the way I did many years ago! We used to use pins but the web thickness at the fixture ends was 4x thickness!
How about instead of pins, or clamps, you mill a pocket into a piece of steel or aluminium that the test sample drops into? The pulling forces on the test piece will then be concentrated through the side faces of the part and into the neck, which is where you want the part to break anyway. Imagine a bowtie join in woodworking, and the test piece that drops in place, then the two halves of the pocket pull apart Edit to add: if you added a DRO scale and recorded both the distance the jaws moved as well as the strain on the part, you could plot a stress/strain curve. Seeing how much a part stretches before it snaps is a useful insight to have
Thank you for saving me the hassle of finding words for how I was thinking of this. ;) Except I will add that I could imagine things having a sort of lip thing, kinda like (though not identical to) the keyhole mounting holes where things slip over the head of a screw, except in this case the "keyhole" stays still, and you drop the piece into it.
It dose not need to bee steel it is sufficient to do a cf nylon part that is over 3-4 times the rigidity limit of the tested part, and have large enough surface to contact to eliminate errors from surface deformation.
I watched your video on the tensile strength system, and I have a few observations and suggestions. The current setup applies force through a pin, which creates a stress concentration. This can be particularly detrimental when the test sample is printed vertically, as shown in your demonstration. The samples broke at the side of the hole rather than in the middle where they ideally should. To improve the accuracy of your results, I suggest using clamps instead of a pin. However, avoid clamps with ridges as they also create stress concentrations. Clamps with rough surfaces only would be a better choice. Additionally, you're missing an important opportunity by not measuring displacement between the holding clamps. By adding this measurement, you can evaluate the stress in the test sample, allowing you to determine the modulus of elasticity and the maximum tensile strength when the sample breaks. The modulus of elasticity will give you a means to compare the filament in the elastic domain, as the maximum stress you are currently measuring is in the plastic region, which you want to avoid in any design.
Great ideas for a first try. Low expense is how I am forced to roll. BTW, your ELS has exceeded all expectations, especially when I don't hear gear noise any more on my 50's Atlas lathe, which I had very little gears for. Thanks to you and Daze Cars for helping me sort it out.
Again, great presentation! I wanted to add to my previous comment about the test performed. While measuring the failure point is indeed very informative, it doesn't guarantee that the part will return to its original dimensions when a stress below the failure point is applied and then released. This is why evaluating the modulus of elasticity (Young's modulus) is so important. The modulus of elasticity, combined with the yield point, allows us to estimate the maximum stress that can be applied to ensure negligible deformation when the forces are released. This is essential for ensuring the full usability and reliability of the part. By measuring the elongation of the test sample, you can calculate the modulus of elasticity and provide a more comprehensive understanding of the material's behavior under stress.
James. You’re a stud and you’ve mastered the best way to mitigate trolls while also gaining visibility to the RUclips algorithm. Brilliant!AND… you do so with pure curiosity and humbleness. Keep kicking ass man. You’re making the world a better and smarter place. 💪🏾💪🏾
I use that PolyMaker PolyLite PETG, and it was just this week I read the label and learned they make 2 grades of PETG. Now you've reminded me I want to see testing between their PolyLite and PolyMax variants.
When printed vertically, the time between layers is more than half at the narrowest point of your test strips. This would help cause the layers to separate easier at the ends because the filament has cooled more between layers. Love your work. Keep it up. 😀
I'm sure you know this, but, strain rate (the rate at which the load is applied) is very important when comparing various materials. Your design doesn't allow you to control the strain rate, but to do so complicates the test station design. Will be fun to watch and see how this evolves.
I mean normal non-diy manually operated tensile testers are good enough for a lot of cases right? Theyre not too far different to this setup unless there's something I'm missing
@@jackygrush yes, you are missing something. The RATE at which the strain in applied must be constant and equal for every test. This homemade system does not accomplish that. Take a sample and fail it. Now, that another sample of the same material, twist it to almost failure and walk away for several hours. Now, come back and continue it to failure. The results will not be the same. Even if you speed the test up, without each test being identical, the end result will not be the same. Now, if one wants to use a test like this, running 15, 20, 30 tests of the same material will help with the statistics, but one or two tests will not. And we haven't even got to test configuration and sample consistency/homogeneity, etc., etc. My comments are not made to discourage learning, but people should understand what they are doing so they can understand biases in their work.
@@stevemarschman3202 yeah I get your point about the effect of strain rate on the yield and tensile strength results, I was just wondering whether the rate precision that can be accomplished by a manual tensile tester is acceptable for cases like this. In our last design project at uni we did manual tensile testing (testers used a hand-wound screw like this) just trying to keep strain rate reasonably similar between tests and the results were close enough for us. I guess what I'm trying to ask is "is strain-rate control the most important upgrade for this tester right now?". Surely the first things to do would be to actually implement strain measurement so that he could graph results and identify yield strength and elastic modulus, not just the UTS. Maybe then strain-rate control would come after that, or maybe it wouldn't really be necessary for the type of work he's doing?
You could use an HX711 24-bit ADC and an Arduino to get continuos force measurements. An LVDT sensor (some have RS485 serial output) can complement the measure with elongation values.
I think the pin isn't hopeless, you just need one that's a lot bigger to spread the load more. I use 16 mm pins, which are part of D-shackles, in my rig.
The instant I saw the hole, I immediately thought stress raisers and crack propagation I'm going to assume your day job is not as a mechanical engineer 🤪 Edit : Just watched your follow up, please continue to post failures, they are an invaluable teaching tool, and they provide the opportunity for failure analysis P.S. your YT work is brilliant 🙂
I used to run similar tests on aviation related materials. Our testing setup was very similar but there was one significant difference: We utilized uniball type bearings between both ends of the sample and the test apparatus. This way the sample was not carrying any torque/twist because of imperfections in sample prep or rest of the test setup.
That looks interesting especially if you have a use for it but these days i have all I can do to make sure I took the right pill at the right time. Thanks for the video keep on keeping on.
Cross section area of the tensioning bolt is less than the cross section of the neck, so greater force/area on that area, so that tends to cause crack formation. You could make the ends as a dovetail or T nut style slot, slide the sample in from the side. Getting carried away, you've got the skills to add a stepper motor on the vise, and a digital controller/logging setup to produce stress/strain curves showing deformation.
Hey to deal with the jaw lift and any unintended forces from the moving jaw you can use a cap head screw that has a rounded shoulder and a PTFE seat that's concave to match you can make these using a ball end mill and a form tool this can be with two mating PTFE washers or just turning a round onto the shoulder of the bolt, this will self align the moving jaw side. Also for vertical print test a golf T shaped test part is normally used this allows them to be just dropped into a slot in the test machine and pulled apart I can sketch an example up and send you a link to it in onshape if you'd like
Super Lazy you say? I'm in. On another note: I got some of that ABS-GF you talked about last time and am really impressed with it. It is my new go-to for functional parts.
Generally with plastics you use a wedge grip or pneumatic grip. Spent many years as an applications engineer for a high end materials testing company. ASTM and ISO have standards that spell out the test requirements for gripping. Strain rate is pretty critical. Impact testing in conjunction will give a more complete view of the material properties.
Define “stronger”. And yes print settings and “grain” orientation makes a huge difference as well. Rigidity, ductility, impact resistance, thermal and chemical resistance etc. “stronger” may be a relative term depending on how it’s used.
As others have mentioned, I don't think you need clamps to hold the samples, a keyed holder that sits around the sample will probably do the trick. Another thing I noticed in the video is that it seems easy to compress the gauge. If you then set the reader to zero, I'm not sure that you are actually offset-free. A block between the vice jaws to have it at teh nominal dimension when loading a sample may be suitable for the current setup, or if you go for a keyed approach, then including some slack in the holders may be sufficient to have a proper point to zero out the reader.
If the goal is the lazy way - I would suggest to drive the vice with some electric wrench that is set to (or kept at) some slow constant and repeatable speed. Thanks for interesting video!
I like the ideas people have about supporting it on the sides, either with a ball shape, or even just on the shoulders. But could you also just use a smaller cross section in the area intended to break? And as others noted, and as I'm sure you also remember from schooling, stress and strain are both important in this, so measuring stretch/ length will also become important, but should be easy to do in this setup.
I built load cells for 18 years. That is an “S” type load cell. It is calibrated in one direction as the primary direction. They are also generally only calibrated in 5 points, 0%-50%-100%-50%-0%. Also there is a temperature compensation that may not have been done in the more entry level units get.
Remake your sample holders with a cut out matching your coupons end. The force will then be distributed around the outside of the coupon and you won’t need a pin. Interesting video, agree on your opinion of using the vise.
If you redesign the holder parts so that the test objects fit into a hole shaped the same as the bases of the test parts but just larger than the parts so that the force pulls against the shoulders of the test samples. No need for any clamps or holding devices. You wouldn't need to use any support either. Basically you would use an extruded cut in the shape of the test samples base down into the part holders and then offset it by .1 or .2mm.
The pins might still work but you would have to lengthen the coupon in the area between the pin and the vise jaw. That should reduce the bending and peeling action. Even with other arrangements it'll be hard to make one design that works with all anisotropic specimens.
Love this set-up! However I know(and I think you also know) that at the end you will build something overengineered with a clamp. Because..... well you know..
I ran into similar problems with vertical-print dog-bone specimens occasionally breaking across the eyelet. When printed vertically, the layer lines aren't strong enough to yield the material (as much as in the horizontal orientation). If the temperature is low enough (e.g. room temperature for PLA), this leads to a brittle failure that is highly sensitive to stress concentration. I needed to change the constant-thickness (minor dimension) to have thickened ends with radiused tapers. Otherwise, it helps to double the eyelet-to-test-section radius by lengthening the specimen, without lengthening the working/test section. Widening the material around the eyelet (in the intermediate dimension), might seem like a simple fix, but it barely helps. While this increases the cross-sectional area around the eyelet, it also serves to increase the degree of peeling-stress concentration, due to greatly increased rigidity around the eyelet. The outer "perimeter" takes little share of the load until fracture has initiated, and the eyelets fail from the hole outwards, with a peeling failure mechanism.
Didn’t read any of the comments so don’t know if this has been suggested or something better. Switch from a rounded surface on the trailing edge of the pin hole in the specimen to an angled surface. I noticed you have an angled surface on the leading edge. You need This in back too. This will split the loading on the specimen more evenly across the structure. Might help. Might not.
Adding nuts at the end of the pins, imbedded into the holding jaws. Would allow you to pinch the sample enough to force the break in the narrow section.
Looking forwad to seeing a more developed version and lots of different material tests, especially CF and GF reinforced materials as they interest me a lot for use on my X1C
Very nice, James. By my calculations the range of results falls within the 5% accuracy you noted for the strain gauge/force meter, at least for the horizontal test pieces (~25% for the vertical, but that's to be expected as the failure is more erratic). Maybe two pins, or a slot/bar arrangement to better equalize the forces? Not as good as clamps, but could still be 3D-printed, and a T-bar easily machined (or printed ?). Stay safe, Charlie
I'm sure you've come across this in your research for this project but ASTM D638 describes a standard shape for plastic tensile test articles. I would recommend changing to that shape, prints in any orientation you want, and changing your end points to support it. You could do it clampless in a few ways I can think of if you're married to that requirement. Switching to the ASTM D638 would likely make your results more directly comparable to manufacture claims, easier to compare results with others, etc.
my first thought is to use 2 pins right after the thin part to lock the part in place without using a hole. just use two pins right where it gets wider. would be pretty simple to model and print.
For the samples printed in the vertical orientation, the »fat« ends had double the print time per layer -> had double the time to cool down between layers, as compared to the narrow parts, and therefore had reduced layer adhesion and so broke first. For the horizontally printed samples the layer print times were identical for all layers, had identical layer adhesion and so the narrow part broke first. I think a better shape for the samples would be a capital “I” with a double sided wedge on the top and bottom with a narrowing (or maybe a notch) in the middle. This way they can still be printed without supports and can still be inserted without clamps in appropriately designed, closely fitting, holders (probably made of metal).
The force meter has quite a slow update rate - it could well be "missing" values between updates. A few of the tests broke after one update but before the next and you were still turning the handle so increasing the force, but the meter didn't reflect that.
Use a larger pin and reduce the concentrated load on the end. Cad it up and run a virtual strain test using different size pins and see the change in the load paths and stress concentrators. As long as the x-section area of the dogbone ends is greater than the necked area it should give good repeatable results. Maybe 8mm would be a better size pin.
Another idea that I haven't seen in the comments: use the same setup and sample design, but make the holders act as a clamp. A captive nut on the bottom "jaw" and tighten the M6 screw from above. I think the holders would need a bit of redesign to flex, or they might not, who knows.
I think you should try a roller shaft on a ramp to clamp those samples. Think about the way a one-way bearing works. If you set the pin up so that it can ride on top of the sample, then gravity will pull it down and as the jaws separate, the rollers will be drawn into the ramp and clamp down on the part. Unfortunately, this will probably require machining a part because a 3D printed ramp will probably not withstand the spreading forces. Of course in this case, the shaft should be machined as shoddily as possible (rough finish) as it will grip better.
The real time graphing on the Instron testers was the best feature by far. Being able to see the exact profile of a stress-strain graph let you instantly know how hard a metal was. For 3d printed parts, being able to see delamination, brittle failure, and plastic deformation would really be amazing if you needed to see that much detail. IIRC, the grain direction for metal coupons was always at a 45 degree angle for tensile tests when possible. If you were to use the pins they could still shear through the layer lines. Your gauge certainly seems to be sensitive enough, might i suggest making the weak area weaker instead of making the strong area stronger?
hi!!! an idea... redesign the jaws that leva you to make the test in that way: put the specimen from the side, and put the pin from the top and only restrict the movement "radially" (the test you has supose make in axial form)... the specimens are to be not with an eye but with an wider area (similars in form to the AWS B4.0. with the pin supporting in the radius zone)
Nice video. Two things: 1. You mentioned that you might get into the control box to adjust the pots. Who do you think you're fooling? OF COURSE you're going to get into the control box to adjust those pots! 2. I'm almost certain that you'll have another two or three 5C collet chucks around your shop. You'll find your 5mm collet in one of them. :-)
Hot and slow is usually the way to go for layer adhesion. Please take the time to look into the materials and techniques used in “gun cad” / 3D printed home built firearms I think there may be some techniques to learn and implement to make your models and prints stronger.
You also want to have temperature monitoring. Add a R-Pi camera. Have it view the display of the force gauge, the display on a (digital) dial gauge monitoring the extension and the display on a thermometer. Add in some AI to read the displays and then graph the output.
Just check on the type of strain gauge to see if it wants the attachments to be free to swivel if so you can arrange a ball and socket arrangement for the mount.
Curious if you print one sample all by it self instead of batch printing would you gain anything. The temp is sure to drop moving from piece to piece so is there a loss of adhesion between layers.
Do you think it would be possible to key the samples into a printed nest matching the existing angles already in the sample design? You could print the nest with tapers in the angles so as to lock the samples in while testing. The hole could be omitted.
I think holes are fine because most people would design the part with holes anyway. I noticed you got Fireball key holder. Did you get keys too? Is it as handy as it looks? I really want one.
Great video! Personally I would cite the Pareto rule that says that 80% of all results come from 20% of the same cause. Personally I think that the difference taken in percentage shows clearly the difference between the types of filaments and then secondary, the arrangement of the layers, which helps one to draw a conclusion.
Should be easy enough to make a simple collet arrangement based on an X-Acto knife. Just have to make the slot wide enough to accommodate a 3mm thick "blade".
I like the pin idea because of the speed of making the samples and speed of installing them accurately. Could you increase the pin count to, say 4 per side, but keep the sample section the same side. Would this spread the force enough to overcome the “zipper” action?
Generally Kurt vises don't and I doubt that one does. It would increase cost for an unintended function. There are issues with the attachment of the moving jaw to the nut as well. They are not made to apply a lot of force in that direction. That said, it could be that these forces are too small to damage the vise.
Have you thought about incorporating a real-time length measurement, so you can get a stress-strain plot? I think most 3D printed parts will have very brittle failures, so the initial portion of the plot may be fairly linear, leading to a good Young's Modulus.
Only at the 12 minute mark, but my initial thoughts are: Why not also measure/record the position of the moveable jaw? This will inform how far the material was stretched prior to failure? I'm concerned that the 3D printed clamps will also deform during testing and wondering how that would affect the tension measurements? I'm also confident that by the end of the video, I will be able to answer the concerns based on the remaining content! Which is why I so enjoy watching Clough42 videos!
1) Dogbone ends with no holes, make keyhole slots for them to drop into to prevent vertical movement. 2) Another way: Nothing says the ends have to be the same thickness as the thin section being tested, we’re 3D printing! Make the ends thicker and you can keep the hole and pin design, then taper down to the thickness for the test section. (Wouldn’t normally do this as you’d be machining the test coupons from thicker stock so it’d be wasteful and slow. But we don’t care because the 3D printer is doing the he work for us and no waste because … 3D printer.)
Few things to think about among the comments here. You could set up a precise distance measurement between the two jaws, zeroed when the test gets dropped in. Log this over time along with the force reading. Graph the newtons vs distance at each time, and that would give you stress/strain charts where you can tell the young's modulus and the areas of elastic and plastic deformation leading up to the actual full mechanical failure.
I would think that with the amount of available space you have, you could model up some drop in fixtures that get pinned into the end pieces, and use the angles of the dogbone of the test piece as it's wedge clamping structure for part holding under tension. You could try orientations that are both holding the dogbone with the pin holes vertical, or horizontal, so you can test to see if the jaw lift is having an impact on the application of force. If it is, you may want to build something like a universal joint pivot set into the fixtures so that all force being applied to the test piece is in line with the test piece rather than applying twisting forces. A variation on the clamp fixture could provide a "good fit" for the part in it's testing configuration, but while the part is going into the fixture it's 90 degrees from it's test position. Mirror the fixtures so that the part gets dropped in with pin holes horizontal, then once fully in, rotate cw or ccw (depending on which fixture is closest to you, until the pin holes are oriented vertically. Set up the fixture so that you can use the pins to validate that the part is properly inserted (don't want one or both ends to suddenly release the part in a rotational method and shoot the part up into the lights.
I'd love to see you hack into the signal after the amplifier and use something like an rpi to capture the force response curve. Even better if you also add a linear potentiometer for displacement. BTW those are called S-beam load cells. Internally they use 4 strain gages in a wheatstone bridge. The cheap display box has an amplifier, ADC, and lcd driver. Could tap into its circuit with oscilloscope to see how it works
On your missing collet: Yeah, I know exactly how that goes. You buy a replacement, and go to put it away where it goes, and lo, the missing one will be right where it was supposed to be.
I understand wanting an inexpensive strain gauge, and the argument for "relative" instead of calibrated measurements, however what happens when your inexpensive one breaks and you have to replace it with another one that may not have similar "relativity"?
how about making the Specimen parts on the plate have a bow tie shape to drop it in to, so it is pulling on the whole end. Not sure if that will work or not.
Hi James, quick question: how do you import an outside 3D model into Fusion? I've done a few quick google searches, but I only saw some very convoluted results, like opening the file, saving it as something specific in the cloud, and some other weird stuff. Is there a way to simply directly import a model into a project?
For those of us without fancy support material, heres a cool trick: set your support distance to 0, then insert pauses between the part and the support, and cover the surface that is going to be removed with sharpie. I havent tried it in nylon, but it works way better than I could have ever imagined with pla, petg, and abs. I would expect it to work in nylon as well. It does have the caveat of only really working for flat surfaces parallel to the build plate, but where it can work, its a game changer.
How about print a clamp that just has a void that is the negative of the dog bone so you just drop them in and pull. Maybe a piece that drops on top of the dog bone to constrain it vertically with a pin through the side
I like what you're doing here. I think you are pulling too quickly though. I've used a tensile test machine at work to test adhesive bonds and they advance at .010" / second. I think there is something to letting the material move and settle as it approaches failure.
@@KNfLrPn Indeed. The max reading always seems to match the displayed reading. That would indicate to me that the issue is one of a low sample polling rate and not just a low display refresh rate.
please test both normal slow speed and also shock/simulate a impact) since it might react diffrent and then you know if the material will work for the aplication that you plan for that part ( some might be real good is slow bot sucks at inpact and others might be the other way around )
I think there's 2 important points on your setup. Firstly, those cheaper straingauge devices poll the straingauge quite infrequently (a couple of hz at most). That can influence your results significantly for very stiff material, or if you crank it fast.
Secondely, this doesn't tell much about how 'strong' a material is. It currently merely gives you the ultimate strength. In practice, you wouldn't want to reach this point at all, and stay below the yield strength (the point at which a material starts to permanently deform). It also doesnt say anything about how tough a material is. A material can have a relatively low yield strength and ultimate strength, but still be very tough by being very low stiffness.
I'd recommend putting a distance sensor between the two jaws, and building a measuring setup to measure the deflection versus loadcell force so you can plot these together. That will allow you to make stress-strain curves which give you much more useful information to base decisions on, and to compare materials.
Low effort distance sensor option - clamp caliper jaws to each side, and record force at various deflections to get stress/strain curve.
I was thinking the same thing. If he's using 3D printed pieces to hold the samples, it might be better to have the distance sensor between the pins (or whatever he goes to) that secure the samples so it removes any deflection of the clamping system, rocking of the vise jaws and load cell. If you're not just manually taking data points, it does turn into a larger project to do the data acquisition though. I'd use an NI CDAQ system with Matlab, but that's at least $1k from used ebay prices.
@@JohnSheerin There are cheaper data acquisition alternatives, for example the NI MyDAW which is under 300$ and has similar integration with matlab.
Although honestly building a simple arduino based logger that reads out the loadcell and a distance sensor, and spits it over a serial port isn't that much more work. Just stringing together a bunch of libraries.
For thermoplastics, the ultimate strength is reasonable approximation of yield strength, because strain hardening is countered by a high degree of area reduction (e.g. necking, crazing).
Make your samples have a round ends ( no holes ) and new holders to suit the round ends so you can just drop them in and test !
I'm picturing a pocket where the end of the sample drops into. No need to secure it, gravity holds it in place until the pulling force takes over. Not sure how to make the edges of the pocket strong enough to deal with the pulling force, but I'm guessing round edges will do the trick.
Gotta say, I love the overall idea. It brings this kind of testing into the attainable realm for many more people.
@@millerbenschop9666 With plastic parts, I would not rely on gravity alone. Don't need clamping force to hold the part against tension, just a top cover of some type that would serve to prevent either end of the wishbone from popping out of the corresponding pocket as the test piece is placed under tension.
I see that at least a few of us have very similar ideas. I look forward to seeing them tried (or, for James to tell us why he chose not to try them, which I'm sure would also be informative) in a future video. :)
Four out of five dentist agree. Boomer joke. Carry on.
Or a bowtie shape
It was fun to watch you discover why tensile testers use clamps. :)
Seriously I love watching people learn the way I did many years ago! We used to use pins but the web thickness at the fixture ends was 4x thickness!
How about instead of pins, or clamps, you mill a pocket into a piece of steel or aluminium that the test sample drops into? The pulling forces on the test piece will then be concentrated through the side faces of the part and into the neck, which is where you want the part to break anyway.
Imagine a bowtie join in woodworking, and the test piece that drops in place, then the two halves of the pocket pull apart
Edit to add: if you added a DRO scale and recorded both the distance the jaws moved as well as the strain on the part, you could plot a stress/strain curve. Seeing how much a part stretches before it snaps is a useful insight to have
Thank you for saving me the hassle of finding words for how I was thinking of this. ;)
Except I will add that I could imagine things having a sort of lip thing, kinda like (though not identical to) the keyhole mounting holes where things slip over the head of a screw, except in this case the "keyhole" stays still, and you drop the piece into it.
I was thinking of something like a dovetail joint.
@@davethetaswegian Butterfly or bowtie shaped test samples (that's back-to-back wedges or double dovetail).
It dose not need to bee steel it is sufficient to do a cf nylon part that is over 3-4 times the rigidity limit of the tested part, and have large enough surface to contact to eliminate errors from surface deformation.
@@David_Best any of these should work. I am on the keyway side of the fence so you can still print the holders
square corners not dovetails
I watched your video on the tensile strength system, and I have a few observations and suggestions. The current setup applies force through a pin, which creates a stress concentration. This can be particularly detrimental when the test sample is printed vertically, as shown in your demonstration. The samples broke at the side of the hole rather than in the middle where they ideally should.
To improve the accuracy of your results, I suggest using clamps instead of a pin. However, avoid clamps with ridges as they also create stress concentrations. Clamps with rough surfaces only would be a better choice.
Additionally, you're missing an important opportunity by not measuring displacement between the holding clamps. By adding this measurement, you can evaluate the stress in the test sample, allowing you to determine the modulus of elasticity and the maximum tensile strength when the sample breaks. The modulus of elasticity will give you a means to compare the filament in the elastic domain, as the maximum stress you are currently measuring is in the plastic region, which you want to avoid in any design.
This Old Tony called to congratulate you on the expressiveness of your hand gestures, young padawan.
Great ideas for a first try. Low expense is how I am forced to roll. BTW, your ELS has exceeded all expectations, especially when I don't hear gear noise any more on my 50's Atlas lathe, which I had very little gears for. Thanks to you and Daze Cars for helping me sort it out.
Again, great presentation! I wanted to add to my previous comment about the test performed. While measuring the failure point is indeed very informative, it doesn't guarantee that the part will return to its original dimensions when a stress below the failure point is applied and then released. This is why evaluating the modulus of elasticity (Young's modulus) is so important.
The modulus of elasticity, combined with the yield point, allows us to estimate the maximum stress that can be applied to ensure negligible deformation when the forces are released. This is essential for ensuring the full usability and reliability of the part. By measuring the elongation of the test sample, you can calculate the modulus of elasticity and provide a more comprehensive understanding of the material's behavior under stress.
James. You’re a stud and you’ve mastered the best way to mitigate trolls while also gaining visibility to the RUclips algorithm. Brilliant!AND… you do so with pure curiosity and humbleness. Keep kicking ass man. You’re making the world a better and smarter place. 💪🏾💪🏾
I use that PolyMaker PolyLite PETG, and it was just this week I read the label and learned they make 2 grades of PETG. Now you've reminded me I want to see testing between their PolyLite and PolyMax variants.
When printed vertically, the time between layers is more than half at the narrowest point of your test strips. This would help cause the layers to separate easier at the ends because the filament has cooled more between layers. Love your work. Keep it up. 😀
I'm sure you know this, but, strain rate (the rate at which the load is applied) is very important when comparing various materials. Your design doesn't allow you to control the strain rate, but to do so complicates the test station design. Will be fun to watch and see how this evolves.
a stepper motor driving the vise would be pretty easy to set up. you could automate it to record the data and then advance a few steps
@@handdancinthe strain must be constantly applied at a specific rate. Stopping can allow material changes that will affect the results.
I mean normal non-diy manually operated tensile testers are good enough for a lot of cases right? Theyre not too far different to this setup unless there's something I'm missing
@@jackygrush yes, you are missing something. The RATE at which the strain in applied must be constant and equal for every test. This homemade system does not accomplish that. Take a sample and fail it. Now, that another sample of the same material, twist it to almost failure and walk away for several hours. Now, come back and continue it to failure. The results will not be the same. Even if you speed the test up, without each test being identical, the end result will not be the same. Now, if one wants to use a test like this, running 15, 20, 30 tests of the same material will help with the statistics, but one or two tests will not. And we haven't even got to test configuration and sample consistency/homogeneity, etc., etc. My comments are not made to discourage learning, but people should understand what they are doing so they can understand biases in their work.
@@stevemarschman3202 yeah I get your point about the effect of strain rate on the yield and tensile strength results, I was just wondering whether the rate precision that can be accomplished by a manual tensile tester is acceptable for cases like this. In our last design project at uni we did manual tensile testing (testers used a hand-wound screw like this) just trying to keep strain rate reasonably similar between tests and the results were close enough for us. I guess what I'm trying to ask is "is strain-rate control the most important upgrade for this tester right now?". Surely the first things to do would be to actually implement strain measurement so that he could graph results and identify yield strength and elastic modulus, not just the UTS. Maybe then strain-rate control would come after that, or maybe it wouldn't really be necessary for the type of work he's doing?
You could use an HX711 24-bit ADC and an Arduino to get continuos force measurements. An LVDT sensor (some have RS485 serial output) can complement the measure with elongation values.
Cnckitchen has entered the chat
That was the first thought that came to my mind, he already checked all easily available plastics years ago
“Guten Tag everybody, I’m Stefan”
“Guten Tag everybody, I’m Stefan”
Thank you for commenting on the reverse force on the vice, it's exactly what I was wondering.
I'm wondering if the jaw lifting was contributing to the eye failures.
I think the pin isn't hopeless, you just need one that's a lot bigger to spread the load more. I use 16 mm pins, which are part of D-shackles, in my rig.
The instant I saw the hole, I immediately thought stress raisers and crack propagation
I'm going to assume your day job is not as a mechanical engineer 🤪
Edit : Just watched your follow up, please continue to post failures, they are an invaluable teaching tool, and they provide the opportunity for failure analysis
P.S. your YT work is brilliant 🙂
Excellent innovation - very useful to be able to validate the myriad of filaments. It also matches our Bambu printer and filaments 😊
Ahhh brings me back to strength and materials class. Tenies Olsen tensile testing machine lol. Great work as always love the show.
I think you are really good at experimenting and problem solving.
I used to run similar tests on aviation related materials. Our testing setup was very similar but there was one significant difference: We utilized uniball type bearings between both ends of the sample and the test apparatus. This way the sample was not carrying any torque/twist because of imperfections in sample prep or rest of the test setup.
That was fun, I love breaking stuff!
That looks interesting especially if you have a use for it but these days i have all I can do to make sure I took the right pill at the right time. Thanks for the video keep on keeping on.
Shop science continues to be awesome. Carry on, sir.
Cross section area of the tensioning bolt is less than the cross section of the neck, so greater force/area on that area, so that tends to cause crack formation. You could make the ends as a dovetail or T nut style slot, slide the sample in from the side. Getting carried away, you've got the skills to add a stepper motor on the vise, and a digital controller/logging setup to produce stress/strain curves showing deformation.
Hey to deal with the jaw lift and any unintended forces from the moving jaw you can use a cap head screw that has a rounded shoulder and a PTFE seat that's concave to match you can make these using a ball end mill and a form tool this can be with two mating PTFE washers or just turning a round onto the shoulder of the bolt, this will self align the moving jaw side. Also for vertical print test a golf T shaped test part is normally used this allows them to be just dropped into a slot in the test machine and pulled apart I can sketch an example up and send you a link to it in onshape if you'd like
Super Lazy you say? I'm in. On another note: I got some of that ABS-GF you talked about last time and am really impressed with it. It is my new go-to for functional parts.
Generally with plastics you use a wedge grip or pneumatic grip. Spent many years as an applications engineer for a high end materials testing company.
ASTM and ISO have standards that spell out the test requirements for gripping.
Strain rate is pretty critical.
Impact testing in conjunction will give a more complete view of the material properties.
Dovetails on the specimen ends that you slide in the holders would work as well I believe.
Define “stronger”.
And yes print settings and “grain” orientation makes a huge difference as well.
Rigidity, ductility, impact resistance, thermal and chemical resistance etc. “stronger” may be a relative term depending on how it’s used.
This is off topic but I love spotting the iced coffee in every video.
As others have mentioned, I don't think you need clamps to hold the samples, a keyed holder that sits around the sample will probably do the trick. Another thing I noticed in the video is that it seems easy to compress the gauge. If you then set the reader to zero, I'm not sure that you are actually offset-free. A block between the vice jaws to have it at teh nominal dimension when loading a sample may be suitable for the current setup, or if you go for a keyed approach, then including some slack in the holders may be sufficient to have a proper point to zero out the reader.
If the goal is the lazy way - I would suggest to drive the vice with some electric wrench that is set to (or kept at) some slow constant and repeatable speed.
Thanks for interesting video!
I think I need to buy that eink tablet. It looks so nice to put ideas on
I like the ideas people have about supporting it on the sides, either with a ball shape, or even just on the shoulders. But could you also just use a smaller cross section in the area intended to break? And as others noted, and as I'm sure you also remember from schooling, stress and strain are both important in this, so measuring stretch/ length will also become important, but should be easy to do in this setup.
I built load cells for 18 years. That is an “S” type load cell. It is calibrated in one direction as the primary direction. They are also generally only calibrated in 5 points, 0%-50%-100%-50%-0%. Also there is a temperature compensation that may not have been done in the more entry level units get.
Remake your sample holders with a cut out matching your coupons end. The force will then be distributed around the outside of the coupon and you won’t need a pin.
Interesting video, agree on your opinion of using the vise.
If you redesign the holder parts so that the test objects fit into a hole shaped the same as the bases of the test parts but just larger than the parts so that the force pulls against the shoulders of the test samples. No need for any clamps or holding devices. You wouldn't need to use any support either. Basically you would use an extruded cut in the shape of the test samples base down into the part holders and then offset it by .1 or .2mm.
Very nice work sir
The pins might still work but you would have to lengthen the coupon in the area between the pin and the vise jaw. That should reduce the bending and peeling action. Even with other arrangements it'll be hard to make one design that works with all anisotropic specimens.
Love this set-up! However I know(and I think you also know) that at the end you will build something overengineered with a clamp. Because..... well you know..
I ran into similar problems with vertical-print dog-bone specimens occasionally breaking across the eyelet. When printed vertically, the layer lines aren't strong enough to yield the material (as much as in the horizontal orientation). If the temperature is low enough (e.g. room temperature for PLA), this leads to a brittle failure that is highly sensitive to stress concentration. I needed to change the constant-thickness (minor dimension) to have thickened ends with radiused tapers. Otherwise, it helps to double the eyelet-to-test-section radius by lengthening the specimen, without lengthening the working/test section.
Widening the material around the eyelet (in the intermediate dimension), might seem like a simple fix, but it barely helps. While this increases the cross-sectional area around the eyelet, it also serves to increase the degree of peeling-stress concentration, due to greatly increased rigidity around the eyelet. The outer "perimeter" takes little share of the load until fracture has initiated, and the eyelets fail from the hole outwards, with a peeling failure mechanism.
Didn’t read any of the comments so don’t know if this has been suggested or something better.
Switch from a rounded surface on the trailing edge of the pin hole in the specimen to an angled surface. I noticed you have an angled surface on the leading edge. You need This in back too.
This will split the loading on the specimen more evenly across the structure. Might help. Might not.
Adding nuts at the end of the pins, imbedded into the holding jaws. Would allow you to pinch the sample enough to force the break in the narrow section.
Looking forwad to seeing a more developed version and lots of different material tests, especially CF and GF reinforced materials as they interest me a lot for use on my X1C
Very nice, James. By my calculations the range of results falls within the 5% accuracy you noted for the strain gauge/force meter, at least for the horizontal test pieces (~25% for the vertical, but that's to be expected as the failure is more erratic).
Maybe two pins, or a slot/bar arrangement to better equalize the forces? Not as good as clamps, but could still be 3D-printed, and a T-bar easily machined (or printed ?).
Stay safe, Charlie
I'm sure you've come across this in your research for this project but ASTM D638 describes a standard shape for plastic tensile test articles. I would recommend changing to that shape, prints in any orientation you want, and changing your end points to support it. You could do it clampless in a few ways I can think of if you're married to that requirement.
Switching to the ASTM D638 would likely make your results more directly comparable to manufacture claims, easier to compare results with others, etc.
ISO 527 would also work if you prefer the ISO standards over ASME.
my first thought is to use 2 pins right after the thin part to lock the part in place without using a hole. just use two pins right where it gets wider. would be pretty simple to model and print.
For the samples printed in the vertical orientation, the »fat« ends had double the print time per layer -> had double the time to cool down between layers, as compared to the narrow parts, and therefore had reduced layer adhesion and so broke first. For the horizontally printed samples the layer print times were identical for all layers, had identical layer adhesion and so the narrow part broke first. I think a better shape for the samples would be a capital “I” with a double sided wedge on the top and bottom with a narrowing (or maybe a notch) in the middle. This way they can still be printed without supports and can still be inserted without clamps in appropriately designed, closely fitting, holders (probably made of metal).
The force meter has quite a slow update rate - it could well be "missing" values between updates. A few of the tests broke after one update but before the next and you were still turning the handle so increasing the force, but the meter didn't reflect that.
Try making a notched receiver piece for the ends of your test samples… the notch/ wedge will transfer the load better to the desired beam.
Use a larger pin and reduce the concentrated load on the end. Cad it up and run a virtual strain test using different size pins and see the change in the load paths and stress concentrators. As long as the x-section area of the dogbone ends is greater than the necked area it should give good repeatable results. Maybe 8mm would be a better size pin.
Another idea that I haven't seen in the comments: use the same setup and sample design, but make the holders act as a clamp. A captive nut on the bottom "jaw" and tighten the M6 screw from above. I think the holders would need a bit of redesign to flex, or they might not, who knows.
Consider adding an enlarged "Tee-Nut" to the bottom of the movable soft jaw to eliminate lifting of the vise jaw.
I've never actually seen anyone use a cbore endmill anymore, fascinating
I think you should try a roller shaft on a ramp to clamp those samples. Think about the way a one-way bearing works. If you set the pin up so that it can ride on top of the sample, then gravity will pull it down and as the jaws separate, the rollers will be drawn into the ramp and clamp down on the part. Unfortunately, this will probably require machining a part because a 3D printed ramp will probably not withstand the spreading forces. Of course in this case, the shaft should be machined as shoddily as possible (rough finish) as it will grip better.
The real time graphing on the Instron testers was the best feature by far. Being able to see the exact profile of a stress-strain graph let you instantly know how hard a metal was. For 3d printed parts, being able to see delamination, brittle failure, and plastic deformation would really be amazing if you needed to see that much detail.
IIRC, the grain direction for metal coupons was always at a 45 degree angle for tensile tests when possible. If you were to use the pins they could still shear through the layer lines.
Your gauge certainly seems to be sensitive enough, might i suggest making the weak area weaker instead of making the strong area stronger?
hi!!! an idea... redesign the jaws that leva you to make the test in that way: put the specimen from the side, and put the pin from the top and only restrict the movement "radially" (the test you has supose make in axial form)... the specimens are to be not with an eye but with an wider area (similars in form to the AWS B4.0. with the pin supporting in the radius zone)
Nice video. Two things:
1. You mentioned that you might get into the control box to adjust the pots. Who do you think you're fooling? OF COURSE you're going to get into the control box to adjust those pots!
2. I'm almost certain that you'll have another two or three 5C collet chucks around your shop. You'll find your 5mm collet in one of them. :-)
Hot and slow is usually the way to go for layer adhesion. Please take the time to look into the materials and techniques used in “gun cad” / 3D printed home built firearms I think there may be some techniques to learn and implement to make your models and prints stronger.
You need to use cnc to control the speed you apply the tension. Otherwise you may go to fast and skew the results with shock loads
That's what a drill is for?
You also want to have temperature monitoring.
Add a R-Pi camera. Have it view the display of the force gauge, the display on a (digital) dial gauge monitoring the extension and the display on a thermometer. Add in some AI to read the displays and then graph the output.
Just check on the type of strain gauge to see if it wants the attachments to be free to swivel if so you can arrange a ball and socket arrangement for the mount.
Use two larger diameter pins to reduce the stress concentration with a gap for the tag to pass through, one pin can be removable.
Curious if you print one sample all by it self instead of batch printing would you gain anything. The temp is sure to drop moving from piece to piece so is there a loss of adhesion between layers.
It would be great if you were able to add a lvdt and use both of their outputs to make stress-strain charts for the materials .
Do you think it would be possible to key the samples into a printed nest matching the existing angles already in the sample design? You could print the nest with tapers in the angles so as to lock the samples in while testing. The hole could be omitted.
a larger diameter pin would be my first go to fix.
Coming next week "Adding a clearpath servo to tenaile tester."
I'd be interested to know in the vertically printed version which side of the test broke, was it the top with overhangs or the bottom without?
I like the way you channeled lockpickinglawyer.
You have clamps. Turn the Mount ends 90 degrees weld your threads back on (j/k get new bolts and nuts) and clamp the samples in.
I think holes are fine because most people would design the part with holes anyway. I noticed you got Fireball key holder. Did you get keys too? Is it as handy as it looks? I really want one.
Great video! Personally I would cite the Pareto rule that says that 80% of all results come from 20% of the same cause. Personally I think that the difference taken in percentage shows clearly the difference between the types of filaments and then secondary, the arrangement of the layers, which helps one to draw a conclusion.
Verry nice!
Should be easy enough to make a simple collet arrangement based on an X-Acto knife. Just have to make the slot wide enough to accommodate a 3mm thick "blade".
Nice job, Thicker area at the retention area.
I like the pin idea because of the speed of making the samples and speed of installing them accurately. Could you increase the pin count to, say 4 per side, but keep the sample section the same side. Would this spread the force enough to overcome the “zipper” action?
There is a piece of paper behind the soft foam on the top lid of the strain gauge, secrets to be discovered!
What brand and model is the tablet device with the grid paper template? It seems like it would be very handy for sketching ideas.
I wonder if a pair of large washers and a bolt and nut would transfer enough force to eliminate shear on the pin for a minimum effort fix?
11:00 I'm wondering if that vise will have thrust bearing in the opening direction 🤔
Generally Kurt vises don't and I doubt that one does. It would increase cost for an unintended function. There are issues with the attachment of the moving jaw to the nut as well. They are not made to apply a lot of force in that direction.
That said, it could be that these forces are too small to damage the vise.
Have you thought about incorporating a real-time length measurement, so you can get a stress-strain plot? I think most 3D printed parts will have very brittle failures, so the initial portion of the plot may be fairly linear, leading to a good Young's Modulus.
Only at the 12 minute mark, but my initial thoughts are:
Why not also measure/record the position of the moveable jaw? This will inform how far the material was stretched prior to failure?
I'm concerned that the 3D printed clamps will also deform during testing and wondering how that would affect the tension measurements?
I'm also confident that by the end of the video, I will be able to answer the concerns based on the remaining content! Which is why I so enjoy watching Clough42 videos!
Not only should you measure the force to break. But you also need to measure the stretch before break too.
And the elongation...
@@seapy2398I get it!
YEAH! James is going to break stuff!🤪
1) Dogbone ends with no holes, make keyhole slots for them to drop into to prevent vertical movement.
2) Another way: Nothing says the ends have to be the same thickness as the thin section being tested, we’re 3D printing! Make the ends thicker and you can keep the hole and pin design, then taper down to the thickness for the test section. (Wouldn’t normally do this as you’d be machining the test coupons from thicker stock so it’d be wasteful and slow. But we don’t care because the 3D printer is doing the he work for us and no waste because … 3D printer.)
Few things to think about among the comments here. You could set up a precise distance measurement between the two jaws, zeroed when the test gets dropped in. Log this over time along with the force reading. Graph the newtons vs distance at each time, and that would give you stress/strain charts where you can tell the young's modulus and the areas of elastic and plastic deformation leading up to the actual full mechanical failure.
Instead of pins, you could print a "saddle" that matches the radius at the ends the specimens and spreads the force over a larger area.
This is the same idea I was going to suggest. This way you need no pin at all and eliminate the concentration of stress around that hole.
I would think that with the amount of available space you have, you could model up some drop in fixtures that get pinned into the end pieces, and use the angles of the dogbone of the test piece as it's wedge clamping structure for part holding under tension. You could try orientations that are both holding the dogbone with the pin holes vertical, or horizontal, so you can test to see if the jaw lift is having an impact on the application of force. If it is, you may want to build something like a universal joint pivot set into the fixtures so that all force being applied to the test piece is in line with the test piece rather than applying twisting forces.
A variation on the clamp fixture could provide a "good fit" for the part in it's testing configuration, but while the part is going into the fixture it's 90 degrees from it's test position. Mirror the fixtures so that the part gets dropped in with pin holes horizontal, then once fully in, rotate cw or ccw (depending on which fixture is closest to you, until the pin holes are oriented vertically. Set up the fixture so that you can use the pins to validate that the part is properly inserted (don't want one or both ends to suddenly release the part in a rotational method and shoot the part up into the lights.
Babe wake up James posted another video
I'd love to see you hack into the signal after the amplifier and use something like an rpi to capture the force response curve. Even better if you also add a linear potentiometer for displacement.
BTW those are called S-beam load cells. Internally they use 4 strain gages in a wheatstone bridge. The cheap display box has an amplifier, ADC, and lcd driver. Could tap into its circuit with oscilloscope to see how it works
On your missing collet: Yeah, I know exactly how that goes. You buy a replacement, and go to put it away where it goes, and lo, the missing one will be right where it was supposed to be.
I understand wanting an inexpensive strain gauge, and the argument for "relative" instead of calibrated measurements, however what happens when your inexpensive one breaks and you have to replace it with another one that may not have similar "relativity"?
how about making the Specimen parts on the plate have a bow tie shape to drop it in to, so it is pulling on the whole end.
Not sure if that will work or not.
Hi James, quick question: how do you import an outside 3D model into Fusion? I've done a few quick google searches, but I only saw some very convoluted results, like opening the file, saving it as something specific in the cloud, and some other weird stuff. Is there a way to simply directly import a model into a project?
For those of us without fancy support material, heres a cool trick: set your support distance to 0, then insert pauses between the part and the support, and cover the surface that is going to be removed with sharpie. I havent tried it in nylon, but it works way better than I could have ever imagined with pla, petg, and abs. I would expect it to work in nylon as well. It does have the caveat of only really working for flat surfaces parallel to the build plate, but where it can work, its a game changer.
How about print a clamp that just has a void that is the negative of the dog bone so you just drop them in and pull. Maybe a piece that drops on top of the dog bone to constrain it vertically with a pin through the side
I like what you're doing here. I think you are pulling too quickly though. I've used a tensile test machine at work to test adhesive bonds and they advance at .010" / second. I think there is something to letting the material move and settle as it approaches failure.
Not to mention the sample rate of the force gauge.
@@KNfLrPn Indeed. The max reading always seems to match the displayed reading. That would indicate to me that the issue is one of a low sample polling rate and not just a low display refresh rate.
please test both normal slow speed and also shock/simulate a impact) since it might react diffrent and then you know if the material will work for the aplication that you plan for that part ( some might be real good is slow bot sucks at inpact and others might be the other way around )
That's a *_Smooth Shaft™_*
What sort of edge finder is that?