Is this the right way to find out? - Ultimate HotEnd Testing - Episode #2
HTML-код
- Опубликовано: 3 окт 2024
- Can we test all HotEnds accurately? here is another preparatory video to this great series.
Test Rig Components
Rotary Encoder used: ► s.click.aliexp...
Load Cell (0-5kg): ► s.click.aliexp...
You want to know more about the HevORT advanced DIY 3D printer?
WEBSITE ► www.hevort.com
🍺 SUPPORT ME
If you like my work and would like to support my efforts in making more open source 3D printing solutions, please consider a donation or becoming a Patron.
Patreon ► / miragec
PayPal ► www.paypal.com...
You want to learn and contribute to the project?
FORUMS ► forums.hevort....
FACEBOOK ► / hevort
DISCORD ► / discord
Thank you very much!
MirageC 
Наука
This type of Engineering Level Data+Analysis is really what’s going to push FDM Printing further.
If you ever feel down, remember how huge the impacts of this will be, even if it ends up imperfect at the end!
I'd expect E3d, Prusa, Slice, and other big names to have similar info. But it's awesome to see it published on the internet. Collecting and publishing good data like this is a TON of work.
@@martylawson1638 Not neccessarily. Printing speeds of around 60mm/s has been around for a long time and focus typically would be on part cost and manufacturability. And research is non linear. Any progress and open data made here will be a big boon for the companies to produce the type of filaments and print heads suitable for fast printing. The creation of a datum like this would help all the companies and the industry. Ultimately us as consumers. Also reminds me that 3D printing has a lot of potential areas to grow forwards.
At this point you should write a research/white paper because this is top quality research!
Good point. I am not familiar with the process. Will look into it.
The most anticipated project on 3d print RUclips... Cmon bro, we need you!
I hope Episode 3 surfaces soon
me too
@@DiomedesDominguez Did he abandon this?
Sooo. this has been a year now.. is it over?
Great video!
Thanks man! glad you enjoyed!
Just want to say that I really appreciate the solid adherence to the scientific method while maintaining an open and very understandable presentation format. I'm very much looking forward to the results from all this hard work, though I'm sure nowhere near as much as you. Thank you for your incredible efforts.
This kind of scientific testing is so important, I’m really happy to see you taking it so seriously and comprehensively
A very interesting data set, Im looking forward to the Olympics, it will give us insight into which hotend design can produce the best prints and at which settings, and the data will prove this.
As someone who did material property research in college, this is so cool to see. We had similar problems in sense that all the "norms" of material testing were done in 2D planes (compression, tensile and similar) but the real world problems emerged when material started "bowing" outside of the plane. Of course, non of the standards covered that and some of the manufactures "cheated" on testing because their products were "perfect" at bearing loads in 2D (perfect shape for that is triangle) and they seemed far superior than others. Now seeing you coming up with ways to "equal" all of the variables just reminded me of the time when we found material property that wasn't direct result of 2D measurements, but was combination of material strength itself and bowing deformation and load it could take... Sorry for the long post, but this is so cool... Thank you for this, it was so cool to watch
The possible negative impact the addition of pigment to filament appears to have on printing speed could be a very useful discovery.
From my experience semi-transparent and natural go faster than colored. I run up to 23mm3\s@0.6 nozzle with mil camo but black only 20 until skips occur. on 0.4 nozzle I had 14mm3\s on gray and black while natural and semi-transparent goes well at 15-18
It really annoys me they don’t publish which pigments/plasticizers etc are added. I wish there was a list like the one on food labels by the nutrition chart.
@@ericlotze7724 I guess this changes a little. base material of different manufacturers is different, dyes are different, colors are mixed from multiple dyes so even you know all the chemicals inside. I googled one common dye name please enjoy:D disodium;2-[[4-[ethyl-[(3-sulfonatophenyl)methyl]amino]phenyl]-[4-[ethyl-[(3-sulfonatophenyl)methyl]azaniumylidene]cyclohexa-2,5-dien-1-ylidene]methyl]benzenesulfonate
Can you guess what is it? Can you guess how harmful is it? Can you guess how it affects plastics viscosity or temperature of PLA? Neither do I. That's why your best bet is to find quality material, do many tests if you make something on a scale, and stick with it :DDD That's why I use more or less average settings for different PLA colors so they a bit overheated so tougher ones get enough to be fast printing while it doesn't really harm those that are faster because of less dyes.
Also I heard that some kinds of matte\pastel etc colors can even be more brittle but never checked myself
@@ricokaboom_ I get it may not mean much to the average person, but i think the data should still be there, same reason the stuff is put on the food labels even though people will often get scared by basic ingredients with big names, rather than look them up! Also Safety Data Sheets even for things like Cleaning Solutions may contain more information than most need, but are useful for research/planning.
I’m rambling on a bit, but i feel like it would require minimal effort on their part, would aid in further refinement of Filament science (what specific additives make “good filament” etc), and FDM printing in general, as well as make planning for research (such as this) or health/safety easier.
@@ricokaboom_ but for all practical purposes, you are right!
Great work! That is what "true engeneering" is
Thank you! 🙂
I am beyond pumped about this series. This is type of content I love watching the most, and exactly this thoroughness I why I built a HevORT myself. Keep it going Olivier :)
For consistent temperature reporting, would you consider a way to embedding a probe into the nozzle? Considering the different anatomies used by different hotends, specially CHC heaters which undershoot a fair bit.
I really appreciate you are refining your testing methodology. You are the first content creator who is doing this to a high standard IMHO. Thinking of my other passion, computer hardware, there we are spoiled for choice and, generally, accurate benchmarking and testing methodologies are quite commonplace.
Hats off to you!
Do they make sensors that thin? That would be really neat though, like a “downhole sensor” lol. Perhaps use a larger nozzle like 1mm?
A flexible surface mount thermocouple glued around the tip of the nozzle maybe a good compromise.
There's a paper (In-situ monitoring of polymer flow temperature and pressure in extrusion based additive manufacturing, by Anderegg et al) that did just this. They also introduced a direct pressure transducer into the nozzle as well and were able to get some really interesting results. Worth a look.
I was wondering when the next video would come out. I didn't realize you were reworking the entire system. Love the effort you are putting into this.
Yeah... I felt we needed a more true life representation. Sorry for the delay.
Guess its never coming now eh @@MirageC
Man... I hope that you continued this series so much
man you rock!!! great video
Thanks man! still a lot to be done ahead, but I am quite thrilled with where we are right now.
Finally! Ive waited an eternity. Really hyped
This should go viral
I would go to the theater and buy popcorn to watch this.
Take your time, do it right!
I've worked as a test-engineer for over 40 years and there is nothing worse than inaccurate instrumentation or flawed test-methodology.
Until you get that all figured out, you cannot make any definitive conclusions. Many times we would be fooled by assuming say, that something was linear when it either had gaps or non-linearities when in certain ranges. Nothing is worse than having to admit that your data is junk and you have to go back and do it right. So, I always learned to take the flak instride and persevere to a proper destination.
What your doing is both ambitious and painstaking but the results will be so valuable. Hang in there!
Different layer heights would create different amounts of backpressure and amount for "pull" when printing. Testing all possible temperature, nozzle size and type (eg. CHT nozzles, or copper vs steel), layer height and even filament brand configurations would be impossible. Testing with one configuration would create "realistic" results only for that specific configuration. For people using different setups, results of "while printing" tests might be as useful as "Mid-Air" testing results. The data points would have to be moved depending on how much more/less backpressure and "pull" person expects when compared to Your setup. Therefore, I wonder if testing while printing is worth the hassle and inaccuracies introduced with it?
In any case, I very appreciate the effort and thought You put into this. I am very excited for the results.
Not impossible, just tedious.
I fee the data is worth the effort. Also basic data can be done by then, but it can also be crowd sourced (ie Have a list of parameters to try / components +nozzles etc to be tried and have people submit data if they do so, maybe even make it a competition a al “#SpeedBoatRace” / “Speed Benchies” etc.
Also the semi-automated nature of 3D Printers, and potentially jigs/automated scanning+photogrammetry etc could make the data collection easier, even for a single person.
(Edit: Readability Improvement)
Agreed. What I think we need to do is to define a standard for test definition. Then having a common understanding of how layer height, width, speed and other variables can affect the results obtained when the standard. A
Thus would at least get us going. Then we can complete by filling the blanks with real data when available.
Still so excited to see your new progress in this work
can't wait for the other hotends to run through the tests, especially the super volcano. thanks for sharing your results!
I have been waiting for this!
Amazing video, loving the engineering approach and thoroughness!
Thank you 😊
I'd given up hope on seeing the results.
Great work man
It's been two years - is this still ongoing? This is such an important project, not just to learn more about existing hot-ends, but by analyzing the final data hot end manufacturers could learn a lot about how to improve their products further. I really hope this project is still ongoing.
Great engineering work man! By the way the pulsating values at lower temperatures are probably do to non homogenous viscosity in the melted material flow.
Haven't used "R" myself, but have used Matlab/Octave a lot. Octave is designed for big data-sets so should cope well if you can find the statistical libraries. Also seen a lot of scientific projects move to "Sci-Pi" recently. More work up-front to work with Python, but higher potential speed if you use the multi-thread Python libraries. (which look very arcane from watching a programming colleague fight them)
Thanks for the tips.
Just subbed. You earned it my dude I'm pretty excited to see your results and the fact that your methodology is logical and consistent while being applied with factual data as much as possible. You're really setting yourself apart here. Keep up the good work
Thank you very much!
Upvoted, then watched.
Fyi, there is a NASA high temperature printer project that is a X-Y bed slinger. That could allow your direct-drive strain-gauge setup to print as well.
Yeeeesss! I've been waiting for this one!
Enjoy
I think some additional test method improvements would be:
1.) Do the test 3 to 5 times each for repeatability and averages
2.) Characterize all the various defects/failure modes. Then after each test print, document how many of each, and when/where they occur.
2.5)Some sort of “scanning jig” may make this easier, something that automatically moves a usb microscope over the surface/around it with a gantry and/or turntable? Not needed, but that could save you some time maybe!
3.) Do a few types of test prints for each variable, bridging, overhang, etc (Granted this may not be needed; I don’t know if there would be any differences!)
Just some ideas i had throughout the video, what do you all think of them?
At this point the level if effort to create one test set is still huge. Perhaps once I have improved the computing process it will be possible. But one area I feel should be improved before doing that would be to have a better control of the environmental conditions of the test room.
You're truly the hero we need, if not the one we deserve.
We need more episodes!
I bet your the only one to do this type of analysis. Keep it up this is awesome!
Thanks for your efforts, mate! 👍
Start the Olympics already. I wanna buy a good hotend. This serie looks like top quality content
I'd love to see a new episode. This is a great series
How about asking CNC Kitchen to do some physical testing on purposely under extruded test samples?
Like you said, under extruding causes the melted bead to "draw" out of the nozzle. This could make the plastic more brittle by adding continuous stress.
Why not having the bed move, and the hot end be stationary? A Voron 2.4 type gantry assembly with a heated bed instead of a print head should be doable. Then a crossbar at the top of the frame to rigidly fix the hotend/extruder bring tested.
Look up the quantum delta. It's perfect for this.
@sprkng had a comment along these lines. They mentioned using a rotating disk (akin to a “Polar Printer”) and just the Z axis to produce vasemode prints with minimal movement/vibration.
@@ericlotze7724 Yes, I see. That would work as well, but it would reduce the type of test prints you could do to different diameter cylinders, which perhaps would be good enough. If mounted on a simple linear actuator, this would improve - off course. You would also need a high VA slip ring coupling, to transfer power to the bed heater and to read thermistor values, which could also be done off course.
@@daliasprints9798 Had not seen that before, but yes, that would be excellent as well.
Excellent episode number two!
Keep on researching, keep on collecting data. The results are worth a lot for both 3D printing manufacturers as the community using and optimizing them.
Lots of compliments!
Ps: if you fancy someone else doing your voiceovers, maybe i can help.
if you want count out dynamic extra friction of bowden what variates by filament pressure you can take sample points with different lenght bowden lenghts and that way get lenght/pressure multiplier.
I think this is great! I was thinking the same thing looking at the data in the first video, I don't want to push things until pressure gets unstable. Once the pressure goes non-linear it's probably going to make inconsistent prints.
That being said, it's been a year, any update? I'd really like to see this data on different hotends.
Wow nice investigation. Great work, keep it coming. Admiring that!!!
Thank you!
Love your content. Perhaps FDA approved medical grade filament would provide better consistency, it is held to higher standards. But, $$$.
I just received 20kg of the same batch directly from Polymaker's factory. I also have the datasheet with flow index. I will use this for all olympics. Now I need to set room control and thermistor calibration.
Looking forward to the data. Subscribed for more, and commenting for the algorithms benefit 👍
Thanks man!
Very good work Mirage! Keep it up
Thank you!
Amazing video! I know how much work goes into doing a proper experiment and documenting it with a well put together video. The story and editing is all on point!
Thank you for your comment! yes indeed, a lot of work. At some point I feel I only see the glitches and the mistakes. I am glad to hear that it still makes some sense 🙂
that is great! you аre helping to develope 3d printing all over the world!
from Russia with love and PEACE,
Ivan
Thank you Ivan for your support 😀
Excellent video! Can't wait to see the Olympics results.
Very nice work. I have been able to push my hot end to higher rates by increasing the flow by 10% to compensate for the fall off in extrusion rate.
Is this project dead?
😢
Great work as always. Somewhat related, you may be interested in Bond 3D and how they control their printing via extruded pressure AFTER the nozzle for near voidless printing. Very interesting technology.
Will look into it! Thank you!
Well, I know for convienience sake I only really start adjusting flow for a material if it is really off, and it seems above ~4% is where that line is. If that 5.75 number is really that consistent, I wonder if it is true for just altering flow +- that value will leave unusable prints. It is totally out of the realm of testing for hotend olympics, but it would be a great diagnostic fact and could help curb people recommending more than +-6% flowrate increments on r/fixmyprint. Good work, and good luck with R, it isn't a great introduction to programming, but it is relatively easy if you know what you want to do.
Did you consider modifying your test rig to simulate printing, as an option to measuring pressure with a real printer? E.g. if you put a rotating glass disc below the rig's nozzle, and lower it 0.2mm per revolution, you could print vase mode tubes with very little added complexity and without shaking from movement.
Great Idea!
In order to have this glass rotating, heated for adhesion, going down at layers, being leveled etc.... This would require a lot of effort. The similarity between mid-air extrusion and printed test shown at 4:04 confirms that the bowden method is viable in my opinion.
Excellent video!
seems like the problem with using a pressure gauge on a direct extruder is that the play in it causes printing artifacts. if instead you mount the direct extruder in a solid piece of aluminum and use a strain gauge on that aluminum, you can determine the strain on that cross section and solve for the force.
Math! Lots of math! 😃 .
Very nice, thank you.
You are welcome!
Cool project, looking forward to the next episode. 👍🏻😎
Currently the flow rate is measured from the input side. Consider the dimension precision of prints and cold core extrusion, the diameters and densities of extruded filaments might vary. Would it be better to determine good or bad by comparing the average thicknesses or weights of walls (stacked lines) to the expected, calculated values under different temps & speeds? (Print quality affected by overhang angle is not considered here.)
Hope this helps. I do appreciate your great efforts of making the jigs and measurements!
I dont think the minimal difference in measuring the extrusion deviation on the input side vs output side is significant enough at this point. Perhaps in the future when we understand all other parameters and diel them to the perfection, this could be looked at. But with the very constant flow request, the fact we have no stop and go and that this test has a very smooth flow change acceleration, the delta between IN and OUT is very limited if not negligeable. Everything coming IN will come OUT.
@@MirageC Sorry for the late reply due to my own tasks. My assumption was when the speed is high enough over a certain point, the the cold core effect and the higher and higher stress in the melted plastic leaving the nozzle start to lower the print quality until the result is eye noticeable. In a macroscope view the delta of mass between in and out can be ignored, but maybe the dimensions are different?
Were the tests actually run???
Great work!!!!
Thank you!
Thank you for sharing and a beautiful video.
What is more important to measure, the pressure in the nozzle or the force the motor has to exert on the filament?
Both are part of the same equation. Nozzle pressure + system drag = motor pushing force.
beau travail de réflexion ! j'ai hâte de voir le résultat et savoir comment se comporte la hotend DYZE !
Merci! Oui moi aussi j'ai hâte d'enfin pourvoir "décoller" avec ce projet!
@@MirageC Salut l'ami, des nouvelles du test de la hotend Dyze ?
Man you roole , wat you are doing i belive it Will be very useful in future and for the future of 3d printing. 👊💪
Thanks man!
We waiting for the Next episode:DDDD
Just amazing!!!! keep doing it. hope you enjoyed my free expensive coffee ;p
Thank you for the support! it is really appreciated! I might invest it into a good microbrewry beer though ;)
Subscribed! Very well done!
Great 👍👍
mirage where are you 😢 these videos are so cool
this guy is a 3d printing genius..
Just a normal guy... very suborn though! lol
I hope you have an intern or an apprentice helping you… this is a massive amount of work.
Yeah, Hence why I stopped during the winter. I now have more computing power thanks to a partner firm who chipped in.
Thank you! You just can’t find data like this anywhere else.
Hey great video! Imo, since the performance limit is defined (in practical terms) by when the extrusion becomes "inconsistent" I'm pretty sure the metric you want use is standard deviation of flow, rather than underextrusion. This will not only tell you when the hotend maxes out its melt capability, but will also allow you to rank extruders on consistency of flow. Mihai Designs posted a video about how dual drive extruders introduce printing artifacts due to slight backlash in the gears that cause extrusions to pulse high/low once per tooth. Extruder force standard deviation would capture that. It wouldn't surprise me if single drive extruder's are "best" for print quality optimization.
damn, not often i feel the need to like a video and then notice i already have :O
This data by itself already shows one more firmware change needed // Compensation: Requested Flow Rate vs. Hotend capability.
Just like input shaping etc. this is essentially input shaping needed on the hotend/extruder combo. Increase request based on the underextrusion %.
Home gamer can measure this by printing multiple flow test turds (ie. how others do it), and inputting the average value per each temperature (few tested at few different flow rates) to establish calculatable hotend capability.
Thus keeping print quality up high further into the hotends ultiamte capability.
Outstanding work!
Nice R&D!
Thanks!
Ate you continuing this project? Iv been waiting for months to base my new designs off your research
sadly using bowden is effect badly the surface quality at highspeed printing.
the tube is pulled and pushed when hotend goes fast in X and Y coordinates 🤔
but not bad idea 😁
This config will not be used for printing, only for testing HotEnds ;)
I have been using compression fittings on my Bowden machines for nearly three years now and the pulling and pushing you mention is completely absent, I got the same pulling and pushing of the Bowden with the Micro Swiss hot ends and it was because of wear in the Bowden collet wich had to be replaced on average about every 4 months
@@AndrewAHayes Im talking bout highspeed (100mm/s+) specially with high acceleration!!!
@@zola5584 Direct drive can print multiple times faster...Bowden isn't needed.
@@REDxFROG Bowden came about as a fix to direct drive back in the days when the mass of the stepper motor and the heavy hotends caused artefacts on the prints, I still run a couple of Bowden equipped machines just so I can compare print quality on parts, some parts I have printed have been better on Bowden at lower speeds, but Im not a speed freak anyway, print quality is my main concern.
I have an idea on how to use a fixed hotend to do print testing. Use a rig where the hotend moves on Z axis, while the hotbed spins. You get a cylinder of fixed radius, and the hotend does not have to move around.
However the hotbed may have to be unpowered, so perhaps using glass plus some glue stick is suitable for this test. Or perhaps use a slip ring to power the hotbed, and allow the print head to move in the radial direction for the base layer to get better grip, then for the layers on top the hotend gets held in place while the part spins.
I thought of something similar but the level of effort would be quite high. I wanted to test bowden first to see if it was a viable solution. Given the similarity with the first rig, i think bowden is just fine.
@@MirageC Wow thanks for replying.
I have been wondering if the limiting factor to extrusion speeds is how well the filament gets heated (if you go fast, the head might not have enough contact time to melt it even though the heater is really hot)
or is the molten filament just limited by the diameter of the nozzle and physically will not extrude at higher rates.
I have been wanting to have a really fast printer for years now, but all sources say it's relatively easy to make it move fast, but hard to get good extrusion.
oh my, I never thought you would be using excel for this D: Matlab, R, Mathematica... but not Excel xD awesome video!
Yeah.... I need to learn on of these better tool 😉
Love this!
will this project be finished before the heat death of the universe?
Can we do an endurance test with abbresives? A lot of heatbreaks fail very fast. I hope you'll read this
Although this would be super useful, I am not planning to do this as part of this series due to time.
damn I can't wait for the actual hotend comparison. I can't decide which one to get... Nova? Rapido? Mosquito Magnum? Or cheap Volcano?
I know, I am holding off upgrading as well until he completes the series haha.
cheap and good is the chc pro.
Depending what you need of course
good work so far. I can´t find the result of that hot end test. Did I miss something?
This is incredible data gathering and analysis here. Honestly you don't get enough credit for the work you do. Once thing you may want to be aware of for future video production is that for me the legend on some of your graphs is so low in the recording, that the video time bar on RUclips is covering it up and hiding the legend. It made is almost impossible for me to see the legend. Kind of annoying that RUclips doesnt have a way (as far as I know) of making that bar disappear while the video is paused, but you may want to try to make sure all important info is high enough in the video frame that it won't be covered by any parts of RUclipss UI.
Awesome! Thank you to the info! Will ensure they are higher, and perhaps bigger. Note that I made a mistake in the last part. I have translated the pressure reading on the right axis directly. I should have taken the corresponding flow reading from that timestamp. In other words, we have now proof that pressure creeping point is pretty much where we start having bad quality print.
Why did you abandon this series? This could of been the thing that elevated you to 100k subs
Hey, great work you are doing there! I love the scientific approach. Keep it up.
I was wondering, regarding the real world measurement, if it would be possible to just add a belt (like on the CR-30) to the free air extrusion rig, to simulate printing? That way you could extrude a continuous line of filament, without the inertia of the moving print head skewing the load cell measurement. You could also meassure the width of the printed line to define when the extrusion failed (maybe?! :D).
I know it's a bit late, but I am curious if this would work or if I am missing something here ;)
Love the puns! 😂
A year on, were the hotend comparisons ever completed?
It's probably going to be hard to be herd.
But I'd like to see the ceramic heater cartridges that are being supplied by triangle labs.
They have severe for v6 , which is good for us low end users to experiment.
Any possibilities
I have one in the line up. :)
@@MirageC oh fantastic.
I've just purchased the hemera heat break with V6 and the chc standard and the volcano.
Yay I'm excited.
Whats the current state of the hotend olympics? :)
Is this data available, I dont see any links to see how fast my current extruder, or future upgrade, can print
couldn't your extrusion amount just be off for every speed, the extrusion deviation is linear, like a Vernier scale, as the speed is higher the deviation per second will be higher even if there is no extrusion error. I really think you need to measure the length of filament right above the hot end because it could be buckling a few percent in the Bowden and you wouldn't know, meaning you dont really know what you are measuring when measuring the amount pulled in to the gears, a gauge at the hot end and on the spool side of the extruder would be a way to actually measure the deviation. consider did the theoretical vs the measured on the spool side of the extruder value really show much, the point that the extruder slips isnt even close to the actual print failure points. The filament will experience a "slip stick" like behavior, it will stall buckling in the Bowden but because it slowed it will get more heat into it and become extrudable again, the error will not show on the spool side.
I invite you to look at the episode one methodology and the comparision i make between the two in this video. I did measure just before the gears in ep.1.