Couple notes on the results: - I'm currently not tracking how quickly filament is being pulled through the sensors. While this won't affect the readings themselves, it does "squish" the graph in its X-axis. - My graph lines up with the graph from Prusa because I had already printed some filament off it (which stretches my graph back into shape). That's also why it seems too mismatch at the start of the graphs. - The way I calculate un-roundness is not the same as what Prusa uses to measure ovality. The two values are not comparable. - The sensors themselves are way exceeding my expectations, but I've identified a few spots that can still be improved. I'll rework the design and publish it as CC-0 when I know it's done. - Filament winding was a massive challenge. Thinking about putting an automated winder arm on there at some point - unless there's already better solutions out there than that? But overall, I'm not just impressed by how well the setup worked, but also by how well these filaments are made!
Have you thought about using pairs of light sources and filtered photo transistors as sensors? With a wavelength that is absorbed by all filaments this should work as a diameter sensor without moving parts. The filters are important if the filament has a fluorescent dye in it.
Please do ones for Amazon basics, polymaker, and more protopasta. I'd love to see how different and more expensive filaments vary but also the cheapest reasonable quality like amazing basics. P.s. I miss your filoween episodes sooo much! I love this return to filament testing and I hope it stays!
I was thinking about point 1. It's probably not worth the effort to fix, but you could feed through an old extruder gear and wind with a spring loaded spool and motorized winder with a stop switch when the spring is tensioned. This would give you a constant speed.
But why don't you just order stack of ones if you're jealous? I mean very basic profiles like 2020 or 4040 are really dirt cheap per meter if you do some browsing. I have about I guess 6 meters on my stock all the time and as said it didn't cost much.
@@jothain mitsumi 2020 material, 75 day ship right now. With the CNY, it could be months before materials arrived even if ordered today. I have some 2020 arriving soon, but it took some digging to find stock. Nice pile of stock you have Tom. :p
Same here ☺️ it’s not as easy as it looks. Ofc you can just follow the basics to make filament (Pellets+ colorant) but I am sure we need more additives to make higher quality. Any luck in finding the almost perfect composition?
@@rastyle11 So far, after many modifications, I have a well-functioning extruder. For several weeks I have been installing the entire further process, cooling, measuring, filament stresses and filament winding. Today I start preparing electronics and programming all the regulators at the same time.
@@ukaszczupakstudia3085 that cool. Hope you can make it 😁. I am more into searching about the composition of the raw materials to make filament. Yes the diameter is important but also the additives (Plasticizers, Impact modifiers, lubricants etc) to make a better quality. I still can’t figure that out.
@@Trent-tr2nx At the moment, I do not want to spend tens of thousands in a project that may not work, so far it is at the threshold of thousands. I want to safely invest in the machine, and if it works very well then I'll start investing in sensors etc. At the moment, I use a dial gauge with an accuracy of 0.001mm.
I had been working on a filament diameter sensor that used optical mouse parts, but this is more practical. The thing is, I have extensive experience with Hall-effect sensors, and am a bit confused why I didn't start there in my design. Doh! My initial plan was to integrate real-time filament diameter sensing with the Klipper firmware. However, the overall quality improvement across the entire 3D printing ecosystem improved faster than my project was progressing, making Klipper much less of a need. I finally shelved my project when I got better at performing extrusion tests that showed filament diameter was NOT a significant factor compared to all the others. For example, filament feed accuracy made the largest contribution, particularly when the hobbed gear teeth get filled with filament. Nozzle quality also makes a difference. It's not just nozzle diameter errors: I think the smoothness of the passage to the outlet is also a factor. The extrusion tests I performed worked backwards, allowing for and eliminating measured factors to determine how much was left to be accounted for by things not yet measured. The amount left that could be due to variations in filament diameter were too small to be worth measuring! Those nozzles, though. What a mess, given that I use cheap ones. What I'd like to try next is to get smaller nozzles then polish them to the next larger size. Thinking of using thread and jeweler's rouge, but haven't tried anything yet mainly because of measurement issues: How will I know when to stop polishing? What I may do is settle for polishing until the ID is "smooth enough", then back-calculate the actual final diameter.
Now this is more like it! Great content Tom. It's really nice to see more making as opposed to just the same-old printing and reviewing. Now, how about a video showing us the end-to-end process and tools you used to get those PCBs designed and made?
I just want to say, in only a few minutes into your video and I decided to subscribe simply because of the organizational detail you put into your workshop! Outstanding!
I never saw cheap "no name" filament, but I do have my suppliers of inexpensive filaments in Europe: Devil Design and Azurefilm. I get PETG from Devil and A85 TPU from Azure.
Watching an assembly it is incredible how nice music can be mixed using basic two spool deck player ;). The topic just hit my problems and thinkering. Great!
MASTERPIECE - wow that’s the most impressive project around 3d-Printing I have seen so far. Not solely the project itself, also the documentation and video quality is outstanding - CONGRATULATIONS Thomas ⭐️⭐️⭐️⭐️⭐️
would be cool to compare the cheapest filament you can find and the most expensive, and the cheapest reccomended filament (imo yoyi) Edit: seems I got lucky on pricing with yoyi filament, so OWL filament seems to unanimously be cheapest
Yeah definitely. The Amazon house brand and the direct from China brands, would be interesting also. There aren't that many places in the world that have the extruder plant to make these filaments. who they are and where they are might be difficult to identify in China but in the US and Europe probably not so hard to put your finger on what plant has what level of quality.
Like most 3D print youtubers Tom refuses to acknowledge the mere existence of cheap filament. BTW: Yoyi costs 40€ a spool. I pay 17-20€ for precise filament from Sunlu or eSun and 11-14€ for the cheapest stuff from OWL/Filamentwerk.
@@MetalheadAndNerd +1, nobody ever talks about cheap filament but they're probably the most used. Yeah filament is nice but even cheap filament is expensive, but I won't spend the double, even if it can be a bit better.
i am almost positive there would not be a big enough difference to even talk about. if one could magically go back in time and get a roll of filament from the 2016 ish lol then you would get your socks knocked off.
@@patprop74 There are visible differences. But the question is: Do I care? I do lots of my prints with 0.3mm layer height, 80mm/s outer wall speed and 120mm/s infill and inner wall speed. I don't care if the walls are not perfectly smooth since at these speeds there are other artifacts as well.
@@renevile There are cheap filament brands with very visible diameter tolerances. I would definitely buy a sensor that could mostly eradicate the differences in print quality between cheap filament and not so cheap filament.
It'll eventually be incorporated into newer machines i guarantee it. Much like auto bed level, filament diameter sensors will be able to adjust on the fly. No reason why something like this couldn't be achieved and achieved soon.
I was talking about that maybe 6 years ago. I even suggested it to Joseph Prusa when I met at Bay Area Maker Faire, but he wasn't convinced. My idea was to use linear image sensors to scan the filament in two axes.
The testing is great but seeing a bit of your workshop was even better, lol. I'm super jealous that you have all that raw stock of AL extrusions and nicely sorted fasteners.
I once used filament that a colleague of mine created out of pure pla from a chemistry supply store. As he did it by hand, the diameter was all over the place. I ended up cutting the filament into pieces with similar diameter, measured it with calipers and scaled the extrusion multiplier in my slicer to compensate for the diameter. Then I fed the short pieces (like 20cm each) one after the other into the printer. Fortunately it was only for a few test prints, and it only cost me half an afternoon. Good times ;)
From my experience on modding the hangprinter. Spool winding improves if theres less than 1º-2º between top and bottom of winding. Just make sure the first layer is correct, the rest just spools up nicely.
I am so jealous of all the materials (Al extrusions and hardware) you have in your shop. I have so many ideas for designs and would never leave the shop.
Would be interesting to see eSun filament tested with this machine, it's the cheapest I can get around here, but from what I can tell it's fine. Would be interesting to see how good/bad it actually is.
I’ve been interested in something like this, but wasn’t sure how to go about it. Looks like I’ve got a great starting point now. Excellent!! Thank you!!!
Just have to say that I love your hoodie! Lol. I just bought my first 3d printer and got it set up thanks to yours and a few other's videos. Thanks for the content. I have learned so much about 3d printing but still need to learn soooo much more!
Great test and ingenious testing rig. Kudos. Yes, I had problems with one spool of PLA. One 1kg spool of Anycubic I got last year gave me trouble. It managed to get stuck(!) in my Ender 3 Pro's regular bowden tubing. Most of it was fine but it exceeded the inner diameter of my tubing in spots. I even had to use pliers in order to get it out. TBF, that was one spool in five, but still...
The (re-)winding machine alone is already super interesting. I talked with some twitter friends recently about "group buying" interesting filaments and then sending each other "samples" or 250g spools.
Hitting it out of the park as usual. Between you and Stefan @ CNC Kitchen you guys are doing great technical work for the 3d printing community, go Germany.
Very impressive. Having Prusa's data to compare as an independent verification adds a level of validity that makes this worth some serious study. You are now positioned to do some interesting filament analysis. Testing the effects of humidity and drying systems on filament? To answer your inquiry, the only problem I've ever had was after replacing the bowden tube on my Ender 3 Pro. I think the tubing was a little too tight. During a really humid month, I had a spool of ASA that kept jamming. Switching out to some Capricorn tubing fixed the issue, as well we adding a heater to my cabinet, to keep the humidity down.
This is really super interesting, I'd really love to see more designs of open source 'stuff', its great to see the design process. Maybe you could build an open source CnC machine or something else and then show improvements. You could even do a design sprint as a livestream and collaborate with viewers
This is a very cool testing rig. I love the use of hall-effect sensors here, very clever! I haven't really had any issue personally with filament inconsistencies, but it would be interesting to compare two spools of the same filament, but from different batches. Thanks for the video tom, I really like your experiment videos! 🧡💙
This is the best video I have seen from you in over a year, great engineering, it was very enjoyable watching you put this all together and have it work.
Great video!!!! Would love to see a follow-up video measuring the diameter of the the same spools after storing them on ambient air for a few days, to see how absorbed moisture is going to affect the quality of the filament.
If you fancy experimenting with the arrangement of hall sensor and magnet, there's a more sensitive configuration where either a thin magnet is placed edge-on to the sensor, or a pair of magnets used side by side. Either way, there is a very sharp transition from north to south. What's more, the response is nearly perfectly linear at the crossover point. This sort of configuration has been used in seismometers, and I once used it in a dilatometer which managed about 10nm resolution.
Those sensors are really cool. I can see farms maybe using them to monitor filament without costing a pretty penny. Also, it can be really good for future reviews of new companies producing filaments if any. My only problem with Diameter has been the absolute Cheapest Black PLA on Amazon a while back, and I wasn't too miffed about it. PLA+ feels so nice to use.
Trolleys for reeling garden hoses have a simple mechanism for evenly rolling the hose - a rod with a bi-directional thread and a hose guide wheel on it (the wheel changes direction when it reaches the end, the same at the other end). Similar mechanism could be used for filament winding.
13:22 Just WOW!!!!! Thomas Sanladerer... I'm very very proud of yourself. The filament size sensor you design is truly (I've no words to say how impress I am). Don't know if this is a German thing, but I would be so excited/demonstrative if I had got that level of precision at that price point! And... You put your design free of right And... We can build it for under 5$ All heroes don't wear cape.... some wear hall-effect sensor!
I've had big problems with Eryone Sparkly Silver PLA, Which has a claimed 50 micron tolerance, however i've found lumps up to 2mm in diameter which is causes the filament to jam at the extruder. I wasted nearly half a roll trying to print with it. I came up with a dual purpose filament jam/runout sensor that uses a single microswitch to pause the printer if a lump is detected.
Now imagine if that was actually hooked into the printing process and automatically adjusted flow rates on the go to compensate for the change in diameter!
As far as I understand it, Marlin expects an analog input currently (1mV = 1um), which my sensor is designed to output. However, I would much prefer I2C, as the sampling is already done on the ATTiny. No need to go back to and from analog again.
@@MadeWithLayers As soon as there is a supply for a sensor at the horizon the software support will be ready within a few days. Marlin 2 is wonderfully modular. I almost wrote an extension for an external fam controller to have proper control over all fans with an SKR 1.4.
That was something that was widely talked about in the days of RepRap when we were all building our own printers instead of buying china made printers lol Some companies like mosaic used the idea of that and made the palette.
Awesome!!! We are also working to adapt to a future industrial printers a system to equalize the flow and solve one of the biggest problems for the accuracy of FDM printers.
I have seen quite a few filament spools where the filament is rotated and jagged,that can also make almost square sections,not just oval, usually cheaper brands, even local german, not even the chinese manufacutrers
Quick thought/suggestion: I know that there are many different varieties of 608 bearings, so they're all over the place. Most 608 bearings I've seen have quite a bit of internal clearance, with seven balls spaced around the race. Because of the internal clearance, I would expect there to be a "bump" every time a ball rolls by (inside the bearing). Some 608 bearings have pre-load in them, so they have much reduced "clearance." I think I heard one skateboard reviewer saying that Lucky ABEC7 brand bearings have the least-clearance / most-preload he's seen. I think their ABEC7 versions are no longer manufactured, but you might check Lucky's other versions. Also, you could try testing a length of precision drill-rod and see if there is any periodic error that matches the spacing of the bearing balls. Finally, if you have a friend who is a machinist, they may have a micrometer with a measuring head designed for measuring the thickness of pipes. You *might* be able to use that to measure the "thickness" of the bearing directly over a ball, and compare that to the "thickness" between balls. But... it might also be really tricky unless you have excellent micrometer technique, because the balls and races can all move around.
Excellent!! Good job! I’m totally in this project, thanks to share! In my country (Brazil) there is some older filament brands such as 3dFila and 3dLab that sells low quality at a premium price... this will help a lot to reveal those ones that invest in market (youtubers and adds) but not in quality. It’s also good to reveal bad advertising such as 0.02 max tolerance... would love to see Sunlu in those tests as they advertise it. Cheers!
flow compensation will produce better results in that case but lots of diameter variation stuck filament into bowden tubes... and that’s normal in some markets that uses cheapest and backyard productions.
Nice setup. From my experience I think that the filament diameter being consistent are allot more important then most users or manufacturer think. I have had good result with addnorth filament. Removing the need for constant measuring the filament diameter
Regarding the question at the end: I mostly use the cheapest PLA filament for my prints, which is usually from Filamentwerk/I-Filament/OWL. Their filament often has some diameter variations but only maybe 1 out of 30 spools is really bad and the filament gets stuck in my generously sized bowden tube. If you like I can send the next bad spool to you.
How do you track the travel-direction (horizontal axis in your graph) while measuring? Do you have a capstan of some sort? Or an encoder to measure the travel-length, perhaps?
I love alot of your filament videos. I do wish there was more of these types with the cheaper filaments. Would be a good compare for the more "name brand", if you would, vs what alot of people buy based on price.
Nice work! Independent verification is always nice, and I look forward to the addition to your reviews. I've personally had some issues with Inland Filament from Micro Center here in the US. I've had a few of these work fine for the first 50-70% of the roll, then serious extrusion issues and clogging show up, so much so that I don't dare use these for MMU based projects. I wish it worked better than my usual choices (Atomic Filament and Push Plastic) since it's half the price and local to me.
Tom, I appreciate the work you did with this video and I am waiting for your designs so that I can, in the near future, replicate it and experiment on some different types and brands of filaments. I agree that initially the quality of filament, in the days of the early RepRap Darwins and Cupcake machines, as to is roundness, consistency of the diameter, etc was far inferior to what we have today and really am enjoying the quality and variety of filaments that are available today. I remember that you only could get 3 mm filament in a very limited range of colours for use in your 3D printer and the revolution that came about with the advent of PLA to be able to print on a non-heated bed without any shrouding to protect the printer and still getting the prints to come out without warping.
As you're driving the reel with CNC anyway, you could add an axis to lay the filament neatly onto the spool. Do you have any feel for how much skew is introduced to the results due to non-linearity of the feed speed caused by the steadily increasing diameter? I'm assuming the Prusa data is linear as it's the output from the extruder?
Love it! I have one spool, (I think it's the AMZ3D but it's very old) I can't use because every couple of meters or so it goes large enough to jam. It would be easy enough to add another axis on the pickup spool to wind it evenly
Hey Tom, could you maybe do a follow-up episode with really cheap filament you can buy on amazon or ebay? Could be interesting how well the no-name or knock-off manufacturers are doing, compared to the reputable sources. I too am from Germany and have tried some no-name eBay PLA filaments with varying degrees of success. But most of them were completely usable. So I'd like to see how much filament diameter really does impact the prints. You could also maybe team up with Stefan from CNC kitchen, to do a kind of collab, where you measure the filaments and he tries do evaluate them for mechanical and printing properties. Some in-depth analysis of different filaments would be really interesting I think :)
I have ESUN White Standard PLA and for some reason my vertical walls are inconsistent. At first I though it was the setting but after switching out filaments my vertical walls were perfect. Not sure if it's due to inconsistency's by I just cannot get decent prints with it. Other ESUN filaments seem fine, might just be the one roll.
That's a genius idea, Tom! well done, Fascinating to see it come together and work. I would like to see the diameter consistency of GST3D filament. Would be interesting to see. :)
Hey Tom,. Set up a servo on the end to control an arm to wind the filament specifically onto the spool in a neat way? That way it can double as a rewinder. Maybe fit a Filadry in there and it would become a DIY filament certifier... ;)
Great project but I have a question: with the increase in the amount of filament on the towing spool does not increase the radius and therefore the tangential speed? If the frequency of detection of the diameter remains unchanged, is there not a risk of reducing the accuracy of the measurement?
A wool winder might be a place to look when designing an automated winding arm? There's a chance a big one might just work as is, but they look to cross the winding over it's self a lot? Awesome job! Love the drill bit calibration :)
I recently had a few issues with Purefil PLA-filament (a Swiss filament company), which states tolerances of 0.05mm but I measured a bit more than 0.1mm by using a Mitutoyo micrometer, which should be quite accurate. If you could measure a spool from them, I would be very interested in that.
Amazing work! Nice idea measuring the diameter using hall sensors. As for filament that could be interesting to test: OWL-Filament is an incredibly cheap (12.90/kg for their b-stock or a-stock if buying in bulk) filament brand that's actually made in germany. I've read user reports that their filaments sometimes vary between 1.5 and 2mm in diameter, which sounds insane, but others seem to be happy with it. I actually ordered a few kgs just yesterday to just print a boatload of drawer organizers for which I don't particularly care for the quality. If the storys of >0.5mm variance turn out to be true I'd be happy to send you one of the rolls, might be interesting to have a legitimately terrible result for comparison.
I use lots of their filament. Most spools are okay. You can see the variations when looking at an angle over printed walls but for functional parts it is perfectly fine. Only every maybe 30 spools there is one that is so bad that the filament gets stuck in the bowden tube. For me this is okay for the price. If I need to print something beautiful I can always use a more expensive brand.
@@MetalheadAndNerd Just opened and started printig with the first spool. I'm quite impressed so far, obviously I'm still at the beginning of the spool but I measured a few times using callipers and measured diameters between 1.76 and 1.79mm, so variance seem very acceptable, and after calibrating the flow rate the prints look really good.
Really awesome project. Thanks. From using a winch I learned a small amount of drag will help the plastic spool neater also maybe purring it on its side? Hope I don't come across as being critical. Big fan
I'm fairly confident that your measured diameter was reading consistently low because you used ferromagnetic steel rods to calibrate your magnetic field Hall Effect sensors. The steel conducts the magnetic field better than plastic or other non-magnetic materials so the same distance will read a higher field through steel than plastic. Like if you were using the resistance of a wire to calculate it's length, and your calibration was done on copper wire and you went on to run the test on aluminum wire. If you recalibrate with something non-magnetic I'm pretty sure your error will go down.
I have some old filament (a clear nylon) that got very oval over time. When it was new, it would print just fine, but after a few years, it's now so oval (checked with calipers) that i can't feed it thru a piece of teflon tube anymore. Interesting device you have built tho, all it needs is a better filament winder with an extra axis that slowly guides the winding up and down along the width of a spool. And having a pickup that reads the length of filament running thru the sensors to give you an absolute position might be another nice addon idea.
Awesome test! I think the bumpiness definitely effects the reading. The free area within the sensor should be as short as possible. Maybe it also helps a bit if the source spool isn't spinning completely free, so you get a bit of tension in the filament. And for winding: you might need a second stepper increasing Z by filament diameter by each rotation of the winding stepper, and the return the same way after every layer.
These sensor concept are really verry interesting. I considert building something difrent for filament diameter with a spring loaded linear potentiometer with a berring in it.
I thought of doing this inline for extrusion multiplier adjustment in real time using linear image sensors (one in each axis). Just needs an opposing LED light source with diffuser and do a binary threshold on the image. Joseph Prusa wasn't convinced when I suggested it to him maybe 6 years ago.
Great video! I have a suggestion for future video on this topic - if you happen to find older spools of filament you have, check those? It may be interesting to see what kind of progress there was, even you mention you wanted to tackle this problem by measuring diameter real-time. Oh and please do cheapest spool from ebay 😂
Couple notes on the results:
- I'm currently not tracking how quickly filament is being pulled through the sensors. While this won't affect the readings themselves, it does "squish" the graph in its X-axis.
- My graph lines up with the graph from Prusa because I had already printed some filament off it (which stretches my graph back into shape). That's also why it seems too mismatch at the start of the graphs.
- The way I calculate un-roundness is not the same as what Prusa uses to measure ovality. The two values are not comparable.
- The sensors themselves are way exceeding my expectations, but I've identified a few spots that can still be improved. I'll rework the design and publish it as CC-0 when I know it's done.
- Filament winding was a massive challenge. Thinking about putting an automated winder arm on there at some point - unless there's already better solutions out there than that?
But overall, I'm not just impressed by how well the setup worked, but also by how well these filaments are made!
Have you thought about using pairs of light sources and filtered photo transistors as sensors? With a wavelength that is absorbed by all filaments this should work as a diameter sensor without moving parts. The filters are important if the filament has a fluorescent dye in it.
Do you need some filament with bigger diameter tolerances for testing?
Does using a drill bit shaft affect your hall sensor calibration in any way, cause it's steel (seams not the case)
Please do ones for Amazon basics, polymaker, and more protopasta. I'd love to see how different and more expensive filaments vary but also the cheapest reasonable quality like amazing basics. P.s. I miss your filoween episodes sooo much! I love this return to filament testing and I hope it stays!
I was thinking about point 1. It's probably not worth the effort to fix, but you could feed through an old extruder gear and wind with a spring loaded spool and motorized winder with a stop switch when the spring is tensioned.
This would give you a constant speed.
Not gonna lie, I am jealous of your stack of aluminium profiles.
SAME
Me too! I thought "damn that would be nice to have"
But why don't you just order stack of ones if you're jealous? I mean very basic profiles like 2020 or 4040 are really dirt cheap per meter if you do some browsing. I have about I guess 6 meters on my stock all the time and as said it didn't cost much.
Also jealous that where Tom lives, there's no need to deal with both metric AND imperial profiles and hardware. Grrrrrr!
@@jothain mitsumi 2020 material, 75 day ship right now. With the CNY, it could be months before materials arrived even if ordered today. I have some 2020 arriving soon, but it took some digging to find stock. Nice pile of stock you have Tom. :p
love the editing in this video vs all the others ive seen so far. you're killin it!
Yup! Spinning the spools like a DJ was awesome! Hahahaha nice!
Yeah i noticed that too :)
Agreed, editing change is welcome. I like seeing the build up process to the final project too. Great work as always.
For months I have been building a production line for the production of filament, this is the video I needed :D
Same here ☺️ it’s not as easy as it looks. Ofc you can just follow the basics to make filament (Pellets+ colorant) but I am sure we need more additives to make higher quality.
Any luck in finding the almost perfect composition?
@@rastyle11 So far, after many modifications, I have a well-functioning extruder. For several weeks I have been installing the entire further process, cooling, measuring, filament stresses and filament winding. Today I start preparing electronics and programming all the regulators at the same time.
@@ukaszczupakstudia3085 check out how prusa does the diameter measurement for Prusament. I believe their system is all optical
@@ukaszczupakstudia3085 that cool. Hope you can make it 😁. I am more into searching about the composition of the raw materials to make filament. Yes the diameter is important but also the additives (Plasticizers, Impact modifiers, lubricants etc) to make a better quality. I still can’t figure that out.
@@Trent-tr2nx At the moment, I do not want to spend tens of thousands in a project that may not work, so far it is at the threshold of thousands. I want to safely invest in the machine, and if it works very well then I'll start investing in sensors etc.
At the moment, I use a dial gauge with an accuracy of 0.001mm.
Love the design of the sensor Tom.
It’s an excellent project 👍
I had been working on a filament diameter sensor that used optical mouse parts, but this is more practical. The thing is, I have extensive experience with Hall-effect sensors, and am a bit confused why I didn't start there in my design. Doh!
My initial plan was to integrate real-time filament diameter sensing with the Klipper firmware. However, the overall quality improvement across the entire 3D printing ecosystem improved faster than my project was progressing, making Klipper much less of a need.
I finally shelved my project when I got better at performing extrusion tests that showed filament diameter was NOT a significant factor compared to all the others. For example, filament feed accuracy made the largest contribution, particularly when the hobbed gear teeth get filled with filament.
Nozzle quality also makes a difference. It's not just nozzle diameter errors: I think the smoothness of the passage to the outlet is also a factor.
The extrusion tests I performed worked backwards, allowing for and eliminating measured factors to determine how much was left to be accounted for by things not yet measured. The amount left that could be due to variations in filament diameter were too small to be worth measuring!
Those nozzles, though. What a mess, given that I use cheap ones. What I'd like to try next is to get smaller nozzles then polish them to the next larger size. Thinking of using thread and jeweler's rouge, but haven't tried anything yet mainly because of measurement issues: How will I know when to stop polishing? What I may do is settle for polishing until the ID is "smooth enough", then back-calculate the actual final diameter.
You will soon make the million subs.
Plot twist: Tom's full name is Tomato
Now this is more like it! Great content Tom. It's really nice to see more making as opposed to just the same-old printing and reviewing. Now, how about a video showing us the end-to-end process and tools you used to get those PCBs designed and made?
This is incredible content. Tom is a genius. His results scaled on to the prusa results is unbelievably good
That build montage was awesome. Really enjoy that kind of content and the result looks great!
I just want to say, in only a few minutes into your video and I decided to subscribe simply because of the organizational detail you put into your workshop! Outstanding!
15:24 well now I REALLY want to know if “cheap” non brand name filament is any good. 🧐
I never saw cheap "no name" filament, but I do have my suppliers of inexpensive filaments in Europe: Devil Design and Azurefilm. I get PETG from Devil and A85 TPU from Azure.
Watching an assembly it is incredible how nice music can be mixed using basic two spool deck player ;).
The topic just hit my problems and thinkering. Great!
This is the quality content I subscribe for.
MASTERPIECE - wow that’s the most impressive project around 3d-Printing I have seen so far. Not solely the project itself, also the documentation and video quality is outstanding - CONGRATULATIONS Thomas ⭐️⭐️⭐️⭐️⭐️
would be cool to compare the cheapest filament you can find and the most expensive, and the cheapest reccomended filament (imo yoyi)
Edit: seems I got lucky on pricing with yoyi filament, so OWL filament seems to unanimously be cheapest
Yeah definitely. The Amazon house brand and the direct from China brands, would be interesting also. There aren't that many places in the world that have the extruder plant to make these filaments. who they are and where they are might be difficult to identify in China but in the US and Europe probably not so hard to put your finger on what plant has what level of quality.
Like most 3D print youtubers Tom refuses to acknowledge the mere existence of cheap filament.
BTW: Yoyi costs 40€ a spool.
I pay 17-20€ for precise filament from Sunlu or eSun and 11-14€ for the cheapest stuff from OWL/Filamentwerk.
@@MetalheadAndNerd +1, nobody ever talks about cheap filament but they're probably the most used. Yeah filament is nice but even cheap filament is expensive, but I won't spend the double, even if it can be a bit better.
i am almost positive there would not be a big enough difference to even talk about. if one could magically go back in time and get a roll of filament from the 2016 ish lol then you would get your socks knocked off.
@@patprop74 There are visible differences. But the question is: Do I care?
I do lots of my prints with 0.3mm layer height, 80mm/s outer wall speed and 120mm/s infill and inner wall speed. I don't care if the walls are not perfectly smooth since at these speeds there are other artifacts as well.
Since Marlin already has support for a filament diameter sensor this could be the next big feature in FFF printers.
I kinda feel the sensor is noisier than the actual filament diameter...
@@SianaGearz That would be easy to filter out.
@@renevile There are cheap filament brands with very visible diameter tolerances. I would definitely buy a sensor that could mostly eradicate the differences in print quality between cheap filament and not so cheap filament.
It'll eventually be incorporated into newer machines i guarantee it. Much like auto bed level, filament diameter sensors will be able to adjust on the fly. No reason why something like this couldn't be achieved and achieved soon.
I was talking about that maybe 6 years ago. I even suggested it to Joseph Prusa when I met at Bay Area Maker Faire, but he wasn't convinced.
My idea was to use linear image sensors to scan the filament in two axes.
Next-level content and providing a way for the community to test it ourselves as well. Thank you for your hard work.
The testing is great but seeing a bit of your workshop was even better, lol. I'm super jealous that you have all that raw stock of AL extrusions and nicely sorted fasteners.
I love this kind of high-quality DIY objective measurement of commercial products, Outstanding work, Tom!
I once used filament that a colleague of mine created out of pure pla from a chemistry supply store. As he did it by hand, the diameter was all over the place. I ended up cutting the filament into pieces with similar diameter, measured it with calipers and scaled the extrusion multiplier in my slicer to compensate for the diameter. Then I fed the short pieces (like 20cm each) one after the other into the printer. Fortunately it was only for a few test prints, and it only cost me half an afternoon. Good times ;)
From my experience on modding the hangprinter. Spool winding improves if theres less than 1º-2º between top and bottom of winding. Just make sure the first layer is correct, the rest just spools up nicely.
Thomas, that sensor is brilliant! Thanks for sharing that! I'm gonna build a similar system that will run inline with my printer! So cool!
I thought about measuring filament in 3 planes literally yesterday evening. Good timing Thomas. Thank you.
I am so jealous of all the materials (Al extrusions and hardware) you have in your shop. I have so many ideas for designs and would never leave the shop.
Would be interesting to see eSun filament tested with this machine, it's the cheapest I can get around here, but from what I can tell it's fine. Would be interesting to see how good/bad it actually is.
Same with Printed Solid's Jessie PLA. I had a couple of great spools but this latest one I bought from them kept causing extruder skipping.
I'd like to see Devil Design and also Azurefilm filaments to be tested.
Fascinating video and not something I have considered or seen before I suppose because I’ve never had a diameter problem. Thanks I enjoyed this
I’ve been interested in something like this, but wasn’t sure how to go about it. Looks like I’ve got a great starting point now. Excellent!! Thank you!!!
Just have to say that I love your hoodie! Lol. I just bought my first 3d printer and got it set up thanks to yours and a few other's videos. Thanks for the content. I have learned so much about 3d printing but still need to learn soooo much more!
I hope to see you test more brands (and maybe more spool from the same brand) in the future. Great idea and video.
Great test and ingenious testing rig. Kudos.
Yes, I had problems with one spool of PLA. One 1kg spool of Anycubic I got last year gave me trouble. It managed to get stuck(!) in my Ender 3 Pro's regular bowden tubing. Most of it was fine but it exceeded the inner diameter of my tubing in spots. I even had to use pliers in order to get it out. TBF, that was one spool in five, but still...
That build edit is sooo amazing 😄👍 Awesome music choice too ^^
The (re-)winding machine alone is already super interesting. I talked with some twitter friends recently about "group buying" interesting filaments and then sending each other "samples" or 250g spools.
I don‘t comment often on videos but I have to say that this was some really nice engineering work and a good scientific approach! Good job Tom!
Hitting it out of the park as usual. Between you and Stefan @ CNC Kitchen you guys are doing great technical work for the 3d printing community, go Germany.
Very impressive. Having Prusa's data to compare as an independent verification adds a level of validity that makes this worth some serious study. You are now positioned to do some interesting filament analysis. Testing the effects of humidity and drying systems on filament?
To answer your inquiry, the only problem I've ever had was after replacing the bowden tube on my Ender 3 Pro. I think the tubing was a little too tight. During a really humid month, I had a spool of ASA that kept jamming.
Switching out to some Capricorn tubing fixed the issue, as well we adding a heater to my cabinet, to keep the humidity down.
This is really super interesting, I'd really love to see more designs of open source 'stuff', its great to see the design process. Maybe you could build an open source CnC machine or something else and then show improvements. You could even do a design sprint as a livestream and collaborate with viewers
This is a very cool testing rig. I love the use of hall-effect sensors here, very clever! I haven't really had any issue personally with filament inconsistencies, but it would be interesting to compare two spools of the same filament, but from different batches. Thanks for the video tom, I really like your experiment videos! 🧡💙
This is the best video I have seen from you in over a year, great engineering, it was very enjoyable watching you put this all together and have it work.
Great video!!!! Would love to see a follow-up video measuring the diameter of the the same spools after storing them on ambient air for a few days, to see how absorbed moisture is going to affect the quality of the filament.
If you fancy experimenting with the arrangement of hall sensor and magnet, there's a more sensitive configuration where either a thin magnet is placed edge-on to the sensor, or a pair of magnets used side by side. Either way, there is a very sharp transition from north to south. What's more, the response is nearly perfectly linear at the crossover point. This sort of configuration has been used in seismometers, and I once used it in a dilatometer which managed about 10nm resolution.
With this you could start including the data from this in filament reviews! Amazing work!
building montage with 80's style synth....classic
Those sensors are really cool. I can see farms maybe using them to monitor filament without costing a pretty penny. Also, it can be really good for future reviews of new companies producing filaments if any.
My only problem with Diameter has been the absolute Cheapest Black PLA on Amazon a while back, and I wasn't too miffed about it. PLA+ feels so nice to use.
Great Idea. What I miss in this build is a solution to wind up the tested Filament correctly and remove fails (knots) from the original spule.
The hall effect sensor is totally brilliant. I need to work with those more.
Trolleys for reeling garden hoses have a simple mechanism for evenly rolling the hose - a rod with a bi-directional thread and a hose guide wheel on it (the wheel changes direction when it reaches the end, the same at the other end). Similar mechanism could be used for filament winding.
Loving this Thomas, Great Manufacturing Skills shown, I love your workshop! Love your Channel! Keep up the good work!
Interesting machine and results. Nice workshop! BTW: the short hair cut really works for you. ⭐️😊👍
That idea of a printer auto-adjusting the filament diameter compensation is such a good idea😁😁🦆👍👌👏
very interesting Tom, you should test more of the cheap filaments.
The montage music was amazing.
13:22 Just WOW!!!!! Thomas Sanladerer...
I'm very very proud of yourself. The filament size sensor you design is truly (I've no words to say how impress I am).
Don't know if this is a German thing, but I would be so excited/demonstrative if I had got that level of precision at that price point!
And... You put your design free of right
And... We can build it for under 5$
All heroes don't wear cape.... some wear hall-effect sensor!
I've had big problems with Eryone Sparkly Silver PLA, Which has a claimed 50 micron tolerance, however i've found lumps up to 2mm in diameter which is causes the filament to jam at the extruder. I wasted nearly half a roll trying to print with it.
I came up with a dual purpose filament jam/runout sensor that uses a single microswitch to pause the printer if a lump is detected.
Wow, there are a lot of satisfying montages in this one!
Rip to that satisfying prusa spooling though haha
1:34
"we're gonna see how well they stack up"
Me: looks at the stack of filaments
I'd say they stack up pretty well
Now imagine if that was actually hooked into the printing process and automatically adjusted flow rates on the go to compensate for the change in diameter!
Marlin 2 already has this feature. Sitting there, waiting for someone to build a sensor.
As far as I understand it, Marlin expects an analog input currently (1mV = 1um), which my sensor is designed to output. However, I would much prefer I2C, as the sampling is already done on the ATTiny. No need to go back to and from analog again.
@@MadeWithLayers As soon as there is a supply for a sensor at the horizon the software support will be ready within a few days. Marlin 2 is wonderfully modular.
I almost wrote an extension for an external fam controller to have proper control over all fans with an SKR 1.4.
That was something that was widely talked about in the days of RepRap when we were all building our own printers instead of buying china made printers lol Some companies like mosaic used the idea of that and made the palette.
Marlin can do that already
Would be kind of fun to see a video of you testing the really cheap fillament to maybe see where they cut corners and does it show? :)
Awesome!!! We are also working to adapt to a future industrial printers a system to equalize the flow and solve one of the biggest problems for the accuracy of FDM printers.
I have seen quite a few filament spools where the filament is rotated and jagged,that can also make almost square sections,not just oval, usually cheaper brands, even local german, not even the chinese manufacutrers
Quick thought/suggestion:
I know that there are many different varieties of 608 bearings, so they're all over the place. Most 608 bearings I've seen have quite a bit of internal clearance, with seven balls spaced around the race. Because of the internal clearance, I would expect there to be a "bump" every time a ball rolls by (inside the bearing).
Some 608 bearings have pre-load in them, so they have much reduced "clearance." I think I heard one skateboard reviewer saying that Lucky ABEC7 brand bearings have the least-clearance / most-preload he's seen. I think their ABEC7 versions are no longer manufactured, but you might check Lucky's other versions.
Also, you could try testing a length of precision drill-rod and see if there is any periodic error that matches the spacing of the bearing balls.
Finally, if you have a friend who is a machinist, they may have a micrometer with a measuring head designed for measuring the thickness of pipes. You *might* be able to use that to measure the "thickness" of the bearing directly over a ball, and compare that to the "thickness" between balls. But... it might also be really tricky unless you have excellent micrometer technique, because the balls and races can all move around.
What an absolute cool test method! I'm very impressed!!!
Excellent!! Good job! I’m totally in this project, thanks to share! In my country (Brazil) there is some older filament brands such as 3dFila and 3dLab that sells low quality at a premium price... this will help a lot to reveal those ones that invest in market (youtubers and adds) but not in quality. It’s also good to reveal bad advertising such as 0.02 max tolerance... would love to see Sunlu in those tests as they advertise it. Cheers!
flow compensation will produce better results in that case but lots of diameter variation stuck filament into bowden tubes... and that’s normal in some markets that uses cheapest and backyard productions.
only when you your own designs its possible to appreciate how well is this desgined ;] nice one
Nice setup.
From my experience I think that the filament diameter being consistent are allot more important then most users or manufacturer think.
I have had good result with addnorth filament.
Removing the need for constant measuring the filament diameter
Regarding the question at the end: I mostly use the cheapest PLA filament for my prints, which is usually from Filamentwerk/I-Filament/OWL.
Their filament often has some diameter variations but only maybe 1 out of 30 spools is really bad and the filament gets stuck in my generously sized bowden tube.
If you like I can send the next bad spool to you.
really wonder if filaments will change in diameter after they absorb some good amount of moisture
Definitely a good setup to test that. Dunk a spool and measure it, or measure a spool before and after dehydrating
How do you track the travel-direction (horizontal axis in your graph) while measuring? Do you have a capstan of some sort? Or an encoder to measure the travel-length, perhaps?
I love alot of your filament videos. I do wish there was more of these types with the cheaper filaments. Would be a good compare for the more "name brand", if you would, vs what alot of people buy based on price.
One of the reasons why I love Prusament They don’t try to hide anything and put the tolerance of that exact spool on the box.
Nice work! Independent verification is always nice, and I look forward to the addition to your reviews. I've personally had some issues with Inland Filament from Micro Center here in the US. I've had a few of these work fine for the first 50-70% of the roll, then serious extrusion issues and clogging show up, so much so that I don't dare use these for MMU based projects. I wish it worked better than my usual choices (Atomic Filament and Push Plastic) since it's half the price and local to me.
Tom, I appreciate the work you did with this video and I am waiting for your designs so that I can, in the near future, replicate it and experiment on some different types and brands of filaments.
I agree that initially the quality of filament, in the days of the early RepRap Darwins and Cupcake machines, as to is roundness, consistency of the diameter, etc was far inferior to what we have today and really am enjoying the quality and variety of filaments that are available today. I remember that you only could get 3 mm filament in a very limited range of colours for use in your 3D printer and the revolution that came about with the advent of PLA to be able to print on a non-heated bed without any shrouding to protect the printer and still getting the prints to come out without warping.
at the 1:55 mark "I see Thomas has raided the Creality warehouse again."
As you're driving the reel with CNC anyway, you could add an axis to lay the filament neatly onto the spool.
Do you have any feel for how much skew is introduced to the results due to non-linearity of the feed speed caused by the steadily increasing diameter? I'm assuming the Prusa data is linear as it's the output from the extruder?
Love it! I have one spool, (I think it's the AMZ3D but it's very old) I can't use because every couple of meters or so it goes large enough to jam.
It would be easy enough to add another axis on the pickup spool to wind it evenly
Consider adding a Level Wind mechanism to evenly distribute the filament onto the spool. It's the same mechanism used on bait caster fishing reels.
This is absolutely fantastic!!!
Great solution for measuring Filament😀
Thanks for the input
Klasse Arbeit
@Thomas Sanladerer what kind of cutter do you use to widen the t-slots for the screw heads of this cylinder head screws?
Hey Tom,
could you maybe do a follow-up episode with really cheap filament you can buy on amazon or ebay?
Could be interesting how well the no-name or knock-off manufacturers are doing, compared to the reputable sources.
I too am from Germany and have tried some no-name eBay PLA filaments with varying degrees of success. But most of them were completely usable. So I'd like to see how much filament diameter really does impact the prints.
You could also maybe team up with Stefan from CNC kitchen, to do a kind of collab, where you measure the filaments and he tries do evaluate them for mechanical and printing properties.
Some in-depth analysis of different filaments would be really interesting I think :)
Thank you so much,
. I have been wanting to do this for a while now. I think you are one of the best. Thanks and keep up the good work.
I have ESUN White Standard PLA and for some reason my vertical walls are inconsistent. At first I though it was the setting but after switching out filaments my vertical walls were perfect. Not sure if it's due to inconsistency's by I just cannot get decent prints with it.
Other ESUN filaments seem fine, might just be the one roll.
That's a genius idea, Tom! well done, Fascinating to see it come together and work. I would like to see the diameter consistency of GST3D filament. Would be interesting to see. :)
Hey Tom,. Set up a servo on the end to control an arm to wind the filament specifically onto the spool in a neat way? That way it can double as a rewinder. Maybe fit a Filadry in there and it would become a DIY filament certifier... ;)
Great project but I have a question: with the increase in the amount of filament on the towing spool does not increase the radius and therefore the tangential speed? If the frequency of detection of the diameter remains unchanged, is there not a risk of reducing the accuracy of the measurement?
This is just amazing results and a rely nice tool setup !!
Hi, great video. I was wondering if those sensors are accurate enough to be used to create a system to auto level the print bed.
A wool winder might be a place to look when designing an automated winding arm? There's a chance a big one might just work as is, but they look to cross the winding over it's self a lot? Awesome job! Love the drill bit calibration :)
I recently had a few issues with Purefil PLA-filament (a Swiss filament company), which states tolerances of 0.05mm but I measured a bit more than 0.1mm by using a Mitutoyo micrometer, which should be quite accurate. If you could measure a spool from them, I would be very interested in that.
Amazing work! Nice idea measuring the diameter using hall sensors.
As for filament that could be interesting to test: OWL-Filament is an incredibly cheap (12.90/kg for their b-stock or a-stock if buying in bulk) filament brand that's actually made in germany. I've read user reports that their filaments sometimes vary between 1.5 and 2mm in diameter, which sounds insane, but others seem to be happy with it. I actually ordered a few kgs just yesterday to just print a boatload of drawer organizers for which I don't particularly care for the quality. If the storys of >0.5mm variance turn out to be true I'd be happy to send you one of the rolls, might be interesting to have a legitimately terrible result for comparison.
I use lots of their filament. Most spools are okay. You can see the variations when looking at an angle over printed walls but for functional parts it is perfectly fine. Only every maybe 30 spools there is one that is so bad that the filament gets stuck in the bowden tube. For me this is okay for the price. If I need to print something beautiful I can always use a more expensive brand.
@@MetalheadAndNerd Just opened and started printig with the first spool. I'm quite impressed so far, obviously I'm still at the beginning of the spool but I measured a few times using callipers and measured diameters between 1.76 and 1.79mm, so variance seem very acceptable, and after calibrating the flow rate the prints look really good.
I this week had a roll of orange AmazonBasics PETG with a 10-15mm section of 2.1mm diameter. Fortunately I caught it before it entered the ptfe!
God i'm so jealous of your workshop!!
Still loving the LTT style Tom keep em coming
Really awesome project. Thanks. From using a winch I learned a small amount of drag will help the plastic spool neater also maybe purring it on its side? Hope I don't come across as being critical. Big fan
I'm fairly confident that your measured diameter was reading consistently low because you used ferromagnetic steel rods to calibrate your magnetic field Hall Effect sensors. The steel conducts the magnetic field better than plastic or other non-magnetic materials so the same distance will read a higher field through steel than plastic. Like if you were using the resistance of a wire to calculate it's length, and your calibration was done on copper wire and you went on to run the test on aluminum wire.
If you recalibrate with something non-magnetic I'm pretty sure your error will go down.
Props for keeping it open sourced 😘
I have some old filament (a clear nylon) that got very oval over time. When it was new, it would print just fine, but after a few years, it's now so oval (checked with calipers) that i can't feed it thru a piece of teflon tube anymore.
Interesting device you have built tho, all it needs is a better filament winder with an extra axis that slowly guides the winding up and down along the width of a spool. And having a pickup that reads the length of filament running thru the sensors to give you an absolute position might be another nice addon idea.
Awesome test! I think the bumpiness definitely effects the reading. The free area within the sensor should be as short as possible.
Maybe it also helps a bit if the source spool isn't spinning completely free, so you get a bit of tension in the filament.
And for winding: you might need a second stepper increasing Z by filament diameter by each rotation of the winding stepper, and the return the same way after every layer.
These sensor concept are really verry interesting. I considert building something difrent for filament diameter with a spring loaded linear potentiometer with a berring in it.
like the use of analog hall sensors -- would like to see more on those
I thought of doing this inline for extrusion multiplier adjustment in real time using linear image sensors (one in each axis). Just needs an opposing LED light source with diffuser and do a binary threshold on the image.
Joseph Prusa wasn't convinced when I suggested it to him maybe 6 years ago.
Should we be changing the filament diameter in our slicer? As it looks like all filaments tested are lower than 1.75 mm.
Great video! I have a suggestion for future video on this topic - if you happen to find older spools of filament you have, check those? It may be interesting to see what kind of progress there was, even you mention you wanted to tackle this problem by measuring diameter real-time. Oh and please do cheapest spool from ebay 😂