This is exactly the kind of ridiculous, non sensicle, unreasonable over engineering that inspires so many people to innovate and make awesome things. Please keep doing this!
Damn... I can barely find things to watch to keep my attention but I was mesmerized for this whole 77 minutes. Even at 1x speed. You have a delightful balance of masochism and intelligence to the effect of "I'm going to bang my head so hard and repeatedly against this problem that it simply will be solved, it is only a matter of time." This was the first video of yours I've watched, and it wasn't until the end when I saw your build stage that I recognized I would be unwise to bet against you printing a benchy in under a minute.
29:00 a random thought to a:b test it - bead blast 2 channels different amounts, leave one normal, polish one - then load different colours in each. Run 4 colours of filament, with the motors at their absolute peak (skipping); then do a cross section analysis of what comes out to see if they have different quantities.
with 17 grams of moving mass, do you anticipate having to dynamically adjust the speed/accel as the build plate gets nearly 50% heavier as the benchy is deposited? This is such a rad project. Those spools are ZOOOOOMIN!!
ok, lets face it, we have reached a pooint where ii have to write it, please load RGB colors and create a code that does mix different amounts for different colors....
You should definitely keep the name for the second heating stage sub assembly "Just the tip" (and the cooling assembly "Only from behind"). (But we shouldn't get ahead of ourselves before we draw any cockclusions. SCNR)
Have you thought about using warm-hot filament spools to reduce the amount of energy needed to melt it once it gets to the hotend? Maybe a silicone sock as well.
Crazy idea for a crazy project: have you considered direct heating the block by shorting say 200a through the block at low voltage? Might be possible to control heating location by mapping resistance and should be very responsive.
That's just crazy enough that it might actually work. But 200A would not be close to enough to cause any significant heating in a copper block of this cross section. Also keep in mind, that the connection "wires" would have to have a significantly bigger cross section than the heating block (otherwise we'd heat the wires, not the block)... But maybe using induction could work!
Ooooooh so excited for a NeedItMakeIt collab of some kind! Love Mike's videos. It's crazy to watch the extruding filament shake when a motor skips! Pressure must be huge. Super glad you are sticking with this and I loved the long detailed video. Keep at it!
for the heater control situation - you know how much filament you're trying to extrude, so with some simple thermodynamics you could make a simple model for how much heat is needed to maintain steady-state and use that model to add a feedforward term to the temperature controller.
Not sure how well it'd work for the larger section but to polish some of my copper hotend parts I use a piece of TPU filament mounted in a rotary tool with a little compound. Of course that'd only work with both sides together. It's slow but moving up and down the path worked amazingly well. This is such a cool project. I have no idea if it'll actually work but I give you badassery score of 10.
So happy to see Martin and the guys at Bontech hooking you up with the best extruders out there. I have known Martin for many years. He's always done a lot to help people in more demanding applications. And as you said... if yours isn't a demanding application, whose is 🙂
30:39 At my old workplace we would polish the inside of our Hotends to a near mirror finish, as we had lots of problems with inconsistent or bad flow rates with the machined surface we got from our supplier.
I think if you just tapered the 2mm holes into 1mm wide slots before joining you could dramatically improve the heat transfer without ever running into what I call the "cht wall" where the unmelted filament reaches the splitter.
Yes, that's an interesting thought. But that comes also with a very drastic and sudden change in melt velocity. (Twice actually). I have no idea how much trouble this is going to cause...
@@Roetz40 Everything is so laminar regardless of surface finish that there's no issue with speed fluctuations in the melt pool. Also, if you *are* concerned with that, simply make the narrower slots deeper to maintain cross-sectional area.
Ich möchte hier mal deine Lesezeichenleiste erwähnen. Nicht das ich neugierig wäre aber als mir 3D Pimmel aufgefallen ist musste ich erstmal lachen 😂. Trotz allem hat mich alles umgehauen wie viel arbeit du in allem gesteckt hast und was dabei heraus kam. Klasse Video 💪🏻
I don’t know if you already mentioned this but adding less than a drop of cooking oil to the polished section of the hotend can help with flow. It makes the walls of the hotend behave like a nonstick iron pan.
Hello Jan, love your content 👍🏻 A little "off topic" but is it possible to do a test mixing the filament with this model and create different colors (RGB) while printing? Very curious about this.... And could be a gamechanger...🤓🍻
@@Roetz40 Please, please do some experiments with the RGB mixing out of the sub 1 min benchy scope. Totally agree that it could be a gamechanger. Actually I signed up for your channel when I saw the hotend design. I immediately saw the color mixing potential and was waiting to see the hotend in action.
@@andreamitchell4758 As I understand hueforge has a db about different filaments and calculates the opacity based on the layers. With Jan's machine the test would look like to fill in the most rgb close filaments in a pair, like two extruders filled with blue two with red and see how consistent the outcome is, I mean if the colors get mixed in the chamber or not. If you check the diamond hotend for rgb mixing, that's the same I'm talking about. So for proper color mixing you have to adjust the extrusion speed of the different colors. That would be the ultimate color mixing, but such a hotend is too heavy for a core XY IMHO.
I love watching tool changers, when i programmed for Lockheed, i added in a tool change to tool station 3 as my signature on all of my programs. I could always tell if my program was being run because it was switch to tool 3 and back to the proper tool really quickly. It was an extra couple of seconds but it was fun to watch. I was there a while back to do a document review and i watched one of my old programs running from 15 years ago. It was just cutting down billets but it was funny to see.
31:40 thought question about the surface finish - cant wait to see the results. One hand hand, having a rough surface increases surface area, yet maybe less contact area for un-molten filament. On the other hand I see nothing wrong with your approach and argument either.
This exceeded my expectations! I love how you're making the impossible possible step by step I was also thinking it would be amazing to be able to lap the channels via *abrasive* *flow* *machining* Can be done with a hydraulic press. looking forwards to the print!
@@asusplayer_ You'd have enough "channels" in this to have cyan, magenta, yellow and the base white, right? In all honesty I don't have the correct background to assess this idea properly, but the 4 inputs made me think of it.
Cut an X in the end of the toothpick, it will act as a reservoir for the paste. It gives it a place to go and it hols very well. I use it for polishing firing pin channels on firearms.
Ideally you have an additional input in your control loop function: The reported flow rate. This way you can have a dependence of the heating power on the expected flow. Something like f(Td, Dm), where Td is the temperature difference (target - measured) and Dm (is the expected flow, as reported by the control software). I am not entirely sure if you can simply use two or 3 PID loops, depending if the two variables are independent: 2 if they are, otherwise 3 (upper tri-diagonal matrix). You would basically build an estimator of the expected temperature, given two inputs. If you want to go full banana, this could easily be done with machine learning but I'd personally prefer classic control theory. The devil obviously lies in the detail: tuning 3 PIDs might be difficult.
That hot end is a crazy work of art. Great job. Have you considered insulating the melt zone block to help retain some the radiant heat lost so more goes into the filament?
This extruder/hotend combo is the prettiest I think I've ever seen. Also, as soon as I saw the nozzle cooler, I immediately thought of Mike from Need It Make It. I think he's probably the best guy to work with on nozzle cooling right now. By the way, I have issues watching an hour plus long video when the content is so interesting, and I'm sure many out there feel the same!
Nozzles should be designed with a bevel on the end to match a bevel on the hot end, similar to a swagelok fitting that can handle 20,000+ psi without leaking.
I think you're exceeding the G limit of what filaments can be expected to hold not only against the bed but when deposited against itself at 'glass' temperature. I think delta style printers have the potential to be the fastest because of the stationary bed and workpiece.
Your theory of stick-slip in the top section is interesting, but if it’s slipping it is effectively decreasing the friction loss and would increase the pressure downstream where you need it, at the nozzle.
I love this project so much, is is so innovative compared to the usual route to getting a fast benchy. I am now building my second custom printer to print fast thanks to you
Theoretisch könnte sich das Fielament beim Polierten Port den leichtesten Weg ja aussuchen. Die harte Seele muss bei der groben Oberfläche aber an der gerade so plastifizierten Randschicht gleiten was am Ende ja mehr Reibung bedeuten würde. Stick Slip gibt es ja eher bei festen Gleitflächen. Bei Zähflüssigen Flüssigkeiten spielt eher Wandreibung eine Rolle in der die Flüssigkeit am Rand fast "steht" und zur Leitungsmitte dann an Geschwindigkeit zunimmt. Damit wäre eine schnelle Plastifizierung Key...aber das ist mit diesem Juwel kein Problem bzw kaum zu verbessern... Respekt für die Konstruktion und Fertigung! Am Ende denke ich macht es aber warscheinlich keinen merkbaren Unterschied ob Poliert oder Sandgestrahlt.
Ich würde tippen, dass die rauhe Oberfläche turbulenten Fluss ermöglicht. Das würde eine bessere Durchmischung des Filaments und schnellere Erhitzung begünstigen. Fluiddynamik ist jetzt nicht mein Gebiet, kann daher auch falsch liegen.
Control system engineer here. You are thinking along the right lines for temperature control of the main heat block, but I'd go one step further and do full system inversion for it. PID control loops struggle with pure time delays, and while there are strategies for compensating it's really not the right tool for this job. The amount of heat you need to inject is predictable from the target temperature and quantity of material pushed through the extruders, so with a little maths based on extrusion rate, filament heat capacity (including latent heat for the phase change to liquid), and some losses to atmosphere you can calculate heater duty cycle and not rely on chasing errors with PID. You would still use an integral function for temperature trim, but with a good PID loop on the hot end to do fine control you should have no trouble accurately tracking target temps. Happy to help further if you want it, feel free to DM or otherwise get in contact :)
what i dont get with direct extruders is why use contact wheels when you could get a better grip over a longer area using a track/belt system to grip the filament from both sides would give huge friction without damaging the filament any way super impressive work. i think you have a solid theory on surface melt look forward to see the results. thanks
You could nickel coat the bores to make it last longer, i did alot of thermal formed ploymer molds and they are very abrasive. They will rough it up by itself. Roughing it will create micro pockets of pressurized steam from the humidity inside the filament.
There's companies that pump a slurry into manifolds to internally hone surfaces. Another option is vapor blasting, essentially wet bead blasting, can get a finer finish than dry bead but more importantly, it peens the surface.
Regarding the heat drop problem, would it be possible to have multiple sensors at multiple locations, averaging them or using the lowest, median or what works best?
I think you may have created a way to make multi color prints. If you have 5 instead of 4 extruders into the one hot-end. White, black, red, blue, green. those be the base colors to make most any other color. And if that works well, more extruders could be added for more of a round hot-end, for more color combinations, such as brown, orange, yellow, pink, purple, and so on.
I remember an interview with the guy from E3D and he said that they leave a little bit of roughness to their nozzles rather than have absolutely polished smooth finish because it worked better. He didn’t elaborate on it, but I always thought it was interesting.
Maybe you mentioned this, but the mass of the hotend would require some kind of massive stepper motor to move the hotend quickly. If you haven't thought about it, would you consider keeping the hotend stationary and moving the bed itself on the XYZ axes? The bed would likely have less mass than the hotend.
Have you looked into more detailed heater control in klipper? It looks like you have a couple of heaters in the big block, that you just control as one in the context of a bed heater. Klipper should be able to control a bunch of separate heaters, you could set them up all separately. You would need to drill a separate hole for each heater and install a thermocouple. In clipper you could then set them up as extruders oder other types of heaters. In the slicer you would just add a couple of commands to set them to the correct temperature.
Pla can be printed under water, I would do water mist cooling or something of the sort, preferably chilled with ice or something. Nice work on the hot end 👍
Have you thought about slightly widening the split in the filament path so that you can cut a thin slit in each of them to put a ceramic heater right In the center of each split? Also, im sure someone else has pointed it out. But it looks like your filament is super moist, all that bubbling and the fact that it snapped at the spool at one point. Youd probably get more accurate extruded volume without the water expanding creating voids and messing with the pressure
Great work, very impressive design. Why do you split the filament to increase the area that makes contact with filament? The centre is harder. Wouldn’t it be better to make the channel wider but not as deep to increase the contact area? Also, is copper the best material in terms of surface erosion over time, especially at the high temperatures you want to reach. I suppose the exposed parts can be plated if necessary.
I’d be very curious to see this as a possible injection molding extruder for small scale applications, where traditionally maybe a screw extruder would’ve been used. I think the excess heat capacity there would actually be a good thing because you want it to be liquid going into the die for the shot, and you’re melting it fast enough that you could do it in a relatively quick time as well. Also, it’s just insane to me watching how much that filament expands coming out of a .5 mm nozzle 😂
If you can make ‘the Nozzle’ move XY in (for example) 5cm incremental steps quickly, then your ‘movable buildplate’ only has to move quickly 5cm x 5cm. Maybe: If then the end tip of ‘the Nozzle’ is long enough to not bump to already printed parts, you can build a couple of layers at a time before advancing to the next 5x5 spot on the grid.
Just before you said “I’m not a native speaker”, I was thinking about how well you were explaining your reasons for the differences in finishes along the path of the filament!
If you print 10X as fast as a normal printer, cooling might become an issue. How much time will one layer need to cool before the fan can cool the surface of a particular spot?
I see this a the birth of a multicolor hotend with zero filament waste. I see a circular version in stead of a flat version, so a 2 by 2 instead of 4 by 1. I think this is very cool! When using with a ‘normal’ bed, is there a way to move this thing in two resolutions? A quick and small resolution where the end nozzle is pivoted 1cmx1cm and a bigger step where the whole ‘cyclone nozzle’ is moved in cm increased steps.
Concerning the air cooling system since you plan to have a limited flow with pressurized air it could be interesting to have a conical nozzle with the outlet being thinner than the body. This should force the air to se a large variation of pressure when exiting the nozzle and loosing a few degrees at the same time. With conical nozzles, the change in pressure takes place mostly when entering the nozzle (at the bottleneck) and the air may heat up again in contact of the nozzle. It may not worth the pain of finding/making conical nozzles though...
Just a constructive comment/criticism on your comment at about 36:20 in the video "I didn't receive additional money to review this". He had several tv and radio programs where he reviewed products from phones, computers, printers, routers, audio, television, projectors ... Every single product he reviewed he purchased and used himself so that he had "money in the game" just like a consumer. Leo LaPorte: I would call Leo the GOAT in introducing computers, software, firmware, OSes to the masses in the 1980 Leo LaPorte. He was responsible for the radio broadcast "Dvorak on Computers" in January 1991. Leo co-created the tv series Screen Savers 1998.
Would "preheating" the filament before the extruder help at all? To 50*c or so, below the glass transition temp. Could probably do this with heated spool containers and the PTFE tube running all the way to them, maybe with an insulating layer around it. Bonus is it would help dry the filament out too.
Looks great. I don’t agree with the rough inner surface for forcing a shearing flow. The extruder gears are pushing the cold filament as one slice, this will not shear like how you described. I would love to see a custom pcb that controlls all the extruders, heaters and temp sensors right on the extruder. Definitely follow up on how the solid “sock” works out!
Anyone got a good name for the hotend? Post them as a reply here! 🎉
Bob
Therminator
The Beast
Peacock
ShelterSmelter
This is exactly the kind of ridiculous, non sensicle, unreasonable over engineering that inspires so many people to innovate and make awesome things. Please keep doing this!
I call it "Fuck around and find out" engineering lol. I love it
I wouldn't even call it overengineering. It's way too simple for that. And it's beautiful.
About the temperature control loop:
Danger Klipper has MPC as an option other than PID, which works just like how you described.
This is absolutely the way to go
I love this project!! Looking so forward to the day of an actual run.
Damn... I can barely find things to watch to keep my attention but I was mesmerized for this whole 77 minutes. Even at 1x speed. You have a delightful balance of masochism and intelligence to the effect of "I'm going to bang my head so hard and repeatedly against this problem that it simply will be solved, it is only a matter of time."
This was the first video of yours I've watched, and it wasn't until the end when I saw your build stage that I recognized I would be unwise to bet against you printing a benchy in under a minute.
29:00 a random thought to a:b test it - bead blast 2 channels different amounts, leave one normal, polish one - then load different colours in each. Run 4 colours of filament, with the motors at their absolute peak (skipping); then do a cross section analysis of what comes out to see if they have different quantities.
with 17 grams of moving mass, do you anticipate having to dynamically adjust the speed/accel as the build plate gets nearly 50% heavier as the benchy is deposited? This is such a rad project. Those spools are ZOOOOOMIN!!
ok, lets face it, we have reached a pooint where ii have to write it, please load RGB colors and create a code that does mix different amounts for different colors....
Yes excellent idea. RGBW
Selectable colors and for highest flow rate it enters poop mode, when the filament comes out much faster and is brown 💩
Wouldn't it need to be YMCK color?
The pigments don’t mix well. You would just end up with a print that looks like different colors depending upon the direction you view the part from.
@@jimmyaveragetoes pretty sure you can create a mixing chamber for the 4 extruders
@@SuperPhilopeCMYKW
You should definitely keep the name for the second heating stage sub assembly "Just the tip" (and the cooling assembly "Only from behind").
(But we shouldn't get ahead of ourselves before we draw any cockclusions. SCNR)
Have you thought about using warm-hot filament spools to reduce the amount of energy needed to melt it once it gets to the hotend? Maybe a silicone sock as well.
Crazy idea for a crazy project: have you considered direct heating the block by shorting say 200a through the block at low voltage? Might be possible to control heating location by mapping resistance and should be very responsive.
I was thinking the same, It would be epic.
Surely somebody has tried using metal filled filament as the heating resistor by now right??!
@@phlebas-uq6fv that is so smart
That's just crazy enough that it might actually work. But 200A would not be close to enough to cause any significant heating in a copper block of this cross section. Also keep in mind, that the connection "wires" would have to have a significantly bigger cross section than the heating block (otherwise we'd heat the wires, not the block)...
But maybe using induction could work!
@@MarcoTedaldi I just saw someone did exactly this yesterday. CRAZY how fast this stuff progresses
sooo much work clearly went into this video. Absolutely amazing! Keep it up!
I see a LOT of moisture in that filament. All the bubbling out the hotend, especially around 1:09:40
Brahma is a god with four heads, believed to have originated from the four Vedas. The name for the hot end
That's a perfect name lol
Ooooooh so excited for a NeedItMakeIt collab of some kind! Love Mike's videos.
It's crazy to watch the extruding filament shake when a motor skips! Pressure must be huge.
Super glad you are sticking with this and I loved the long detailed video. Keep at it!
for the heater control situation - you know how much filament you're trying to extrude, so with some simple thermodynamics you could make a simple model for how much heat is needed to maintain steady-state and use that model to add a feedforward term to the temperature controller.
Not sure how well it'd work for the larger section but to polish some of my copper hotend parts I use a piece of TPU filament mounted in a rotary tool with a little compound. Of course that'd only work with both sides together. It's slow but moving up and down the path worked amazingly well.
This is such a cool project. I have no idea if it'll actually work but I give you badassery score of 10.
46:56 “…this is pretty much a complete unit.” One may even say it’s an Absolute Unit 😂😂😂
I love this idea for the name!
This extruder looks straight out of injection molding machine. I wonder what kind of pressure it can make
Wow! What a fantastic project you are developing, Jan! Thank you for sharing your progress; it is great fun and inspiring to follow your progress.
So happy to see Martin and the guys at Bontech hooking you up with the best extruders out there. I have known Martin for many years. He's always done a lot to help people in more demanding applications. And as you said... if yours isn't a demanding application, whose is 🙂
30:39 At my old workplace we would polish the inside of our Hotends to a near mirror finish, as we had lots of problems with inconsistent or bad flow rates with the machined surface we got from our supplier.
I think if you just tapered the 2mm holes into 1mm wide slots before joining you could dramatically improve the heat transfer without ever running into what I call the "cht wall" where the unmelted filament reaches the splitter.
Yes, that's an interesting thought. But that comes also with a very drastic and sudden change in melt velocity. (Twice actually). I have no idea how much trouble this is going to cause...
@@Roetz40 Everything is so laminar regardless of surface finish that there's no issue with speed fluctuations in the melt pool. Also, if you *are* concerned with that, simply make the narrower slots deeper to maintain cross-sectional area.
This slot design is used in the Nanoflow tungsten carbide nozzle.
Ich möchte hier mal deine Lesezeichenleiste erwähnen. Nicht das ich neugierig wäre aber als mir 3D Pimmel aufgefallen ist musste ich erstmal lachen 😂.
Trotz allem hat mich alles umgehauen wie viel arbeit du in allem gesteckt hast und was dabei heraus kam.
Klasse Video 💪🏻
and no benchys. ive been bamboozled.
I thought the same, but the title says that "he could print" not that he printed it.
Bambu-zeld? 😂
@markfen88 I knew it wasn't coming, so I just went to the end to make sure before I hoped for 27 min.
The margins of the 1 hour 16 min video were too narrow to contain the benchies.
@hippiemcfake6364 You mean the 30 second benchys? 😆
Everything about this video is bloody brilliant,including the silent bit. Don’t stop doing this!
I don’t know if you already mentioned this but adding less than a drop of cooking oil to the polished section of the hotend can help with flow. It makes the walls of the hotend behave like a nonstick iron pan.
Possible to feed RGBW or RGBK filaments with different flow rate to mix out all colors?
I don't see why not! I would be interested in just seeing what it would look like with 4 colors all at once! :)
Good idea, the correct colors would be CMYK
Das ist genau der Scheiss', auf den ich Bock hab'!
Great video! It felt like 30 minutes. Can't fault a single thing. Waiting for the next video! Keep them long.
Its perfect to make an art deco style 3D Printer! :D
Next step: Input shaping for Hotend heater
Hello Jan, love your content 👍🏻
A little "off topic" but is it possible to do a test mixing the filament with this model and create different colors (RGB) while printing?
Very curious about this.... And could be a gamechanger...🤓🍻
I don't plan on going RGB or anything but we will definitely get some color going on in the next prints. ;)
@@Roetz40 Please, please do some experiments with the RGB mixing out of the sub 1 min benchy scope. Totally agree that it could be a gamechanger. Actually I signed up for your channel when I saw the hotend design. I immediately saw the color mixing potential and was waiting to see the hotend in action.
@Mister_Mxyyzptlk
@@Mister_Mxyzptlkyou mean like hue forge but for 3D prints instead of 2D?
@@andreamitchell4758 As I understand hueforge has a db about different filaments and calculates the opacity based on the layers. With Jan's machine the test would look like to fill in the most rgb close filaments in a pair, like two extruders filled with blue two with red and see how consistent the outcome is, I mean if the colors get mixed in the chamber or not. If you check the diamond hotend for rgb mixing, that's the same I'm talking about. So for proper color mixing you have to adjust the extrusion speed of the different colors. That would be the ultimate color mixing, but such a hotend is too heavy for a core XY IMHO.
5:30 be careful of using that splitting method, it's under a patent in US
I'm not in the US and I don't intend on making any money out of this. I don't care for parents in that regard. But thanks for mentioning ;)
I love watching tool changers, when i programmed for Lockheed, i added in a tool change to tool station 3 as my signature on all of my programs. I could always tell if my program was being run because it was switch to tool 3 and back to the proper tool really quickly. It was an extra couple of seconds but it was fun to watch. I was there a while back to do a document review and i watched one of my old programs running from 15 years ago. It was just cutting down billets but it was funny to see.
31:40 thought question about the surface finish - cant wait to see the results. One hand hand, having a rough surface increases surface area, yet maybe less contact area for un-molten filament. On the other hand I see nothing wrong with your approach and argument either.
This exceeded my expectations! I love how you're making the impossible possible step by step
I was also thinking it would be amazing to be able to lap the channels via *abrasive* *flow* *machining*
Can be done with a hydraulic press.
looking forwards to the print!
Would you be able to do CMYK color mixing with this if you had some sort of "rifling" in the final combined path?
Wouldn't you also need a 'base' white color? CMYK also depends on mixing with white, but yeah at least in theory this could work
@@asusplayer_ You'd have enough "channels" in this to have cyan, magenta, yellow and the base white, right? In all honesty I don't have the correct background to assess this idea properly, but the 4 inputs made me think of it.
That is absolutely wild. Impressive work. Congratulations. ❤
This hotend looks TOTALLY EPIC!!!
Cut an X in the end of the toothpick, it will act as a reservoir for the paste. It gives it a place to go and it hols very well. I use it for polishing firing pin channels on firearms.
10g extrusion in 60seconds. Fast and the Furious popped into mind. Going to need some overnight parts from Japan.😂
Nope. In 30 secs. That's more like topgun on speed.
Spectacular work! Seeing melted filament blasting out like a firehose is wild!
Ideally you have an additional input in your control loop function: The reported flow rate. This way you can have a dependence of the heating power on the expected flow.
Something like f(Td, Dm), where Td is the temperature difference (target - measured) and Dm (is the expected flow, as reported by the control software).
I am not entirely sure if you can simply use two or 3 PID loops, depending if the two variables are independent: 2 if they are, otherwise 3 (upper tri-diagonal matrix). You would basically build an estimator of the expected temperature, given two inputs. If you want to go full banana, this could easily be done with machine learning but I'd personally prefer classic control theory. The devil obviously lies in the detail: tuning 3 PIDs might be difficult.
That hot end is a crazy work of art. Great job. Have you considered insulating the melt zone block to help retain some the radiant heat lost so more goes into the filament?
This extruder/hotend combo is the prettiest I think I've ever seen. Also, as soon as I saw the nozzle cooler, I immediately thought of Mike from Need It Make It. I think he's probably the best guy to work with on nozzle cooling right now. By the way, I have issues watching an hour plus long video when the content is so interesting, and I'm sure many out there feel the same!
BTW the extruder hotend combo now looks like a mega weapon from a mecha anime and the color of the block after heating is phenomenal.
Nozzles should be designed with a bevel on the end to match a bevel on the hot end, similar to a swagelok fitting that can handle 20,000+ psi without leaking.
I think you're exceeding the G limit of what filaments can be expected to hold not only against the bed but when deposited against itself at 'glass' temperature. I think delta style printers have the potential to be the fastest because of the stationary bed and workpiece.
Your theory of stick-slip in the top section is interesting, but if it’s slipping it is effectively decreasing the friction loss and would increase the pressure downstream where you need it, at the nozzle.
1:10:22 Haha! You will need roller bearings on the spools🤣👍Great work and exciting af. o7 Ulf
I love this project so much, is is so innovative compared to the usual route to getting a fast benchy. I am now building my second custom printer to print fast thanks to you
Regarding the cooling with pressurized air you should take a look at Vortex cooling. This could allow you to passively decrease the air temp.
You can use it for injection molding too :D
That's what I was thinking. 🤣
Theoretisch könnte sich das Fielament beim Polierten Port den leichtesten Weg ja aussuchen. Die harte Seele muss bei der groben Oberfläche aber an der gerade so plastifizierten Randschicht gleiten was am Ende ja mehr Reibung bedeuten würde.
Stick Slip gibt es ja eher bei festen Gleitflächen. Bei Zähflüssigen Flüssigkeiten spielt eher Wandreibung eine Rolle in der die Flüssigkeit am Rand fast "steht" und zur Leitungsmitte dann an Geschwindigkeit zunimmt.
Damit wäre eine schnelle Plastifizierung Key...aber das ist mit diesem Juwel kein Problem bzw kaum zu verbessern... Respekt für die Konstruktion und Fertigung!
Am Ende denke ich macht es aber warscheinlich keinen merkbaren Unterschied ob Poliert oder Sandgestrahlt.
Ich würde tippen, dass die rauhe Oberfläche turbulenten Fluss ermöglicht. Das würde eine bessere Durchmischung des Filaments und schnellere Erhitzung begünstigen. Fluiddynamik ist jetzt nicht mein Gebiet, kann daher auch falsch liegen.
Control system engineer here. You are thinking along the right lines for temperature control of the main heat block, but I'd go one step further and do full system inversion for it.
PID control loops struggle with pure time delays, and while there are strategies for compensating it's really not the right tool for this job.
The amount of heat you need to inject is predictable from the target temperature and quantity of material pushed through the extruders, so with a little maths based on extrusion rate, filament heat capacity (including latent heat for the phase change to liquid), and some losses to atmosphere you can calculate heater duty cycle and not rely on chasing errors with PID. You would still use an integral function for temperature trim, but with a good PID loop on the hot end to do fine control you should have no trouble accurately tracking target temps.
Happy to help further if you want it, feel free to DM or otherwise get in contact :)
Want to join the discord and share some knowledge? ;) links in the video description
what i dont get with direct extruders is why use contact wheels when you could get a better grip over a longer area using a track/belt system to grip the filament from both sides would give huge friction without damaging the filament any way super impressive work. i think you have a solid theory on surface melt look forward to see the results. thanks
Check out proper printing. He created one and discovered some quirks. Really cool guy
You could nickel coat the bores to make it last longer, i did alot of thermal formed ploymer molds and they are very abrasive. They will rough it up by itself. Roughing it will create micro pockets of pressurized steam from the humidity inside the filament.
There's companies that pump a slurry into manifolds to internally hone surfaces. Another option is vapor blasting, essentially wet bead blasting, can get a finer finish than dry bead but more importantly, it peens the surface.
Regarding the heat drop problem, would it be possible to have multiple sensors at multiple locations, averaging them or using the lowest, median or what works best?
I think you may have created a way to make multi color prints. If you have 5 instead of 4 extruders into the one hot-end. White, black, red, blue, green. those be the base colors to make most any other color.
And if that works well, more extruders could be added for more of a round hot-end, for more color combinations, such as brown, orange, yellow, pink, purple, and so on.
Pretty cool to get Mike from NeedItMakeIt involved! I’ve been following the hot end cooling duct competition on his channel!
I remember an interview with the guy from E3D and he said that they leave a little bit of roughness to their nozzles rather than have absolutely polished smooth finish because it worked better. He didn’t elaborate on it, but I always thought it was interesting.
Beautiful craftsmanship
40:53 Wow! That looks pretty promising 😅👍. And nicely designed too. o7 Ulf
Maybe you mentioned this, but the mass of the hotend would require some kind of massive stepper motor to move the hotend quickly. If you haven't thought about it, would you consider keeping the hotend stationary and moving the bed itself on the XYZ axes? The bed would likely have less mass than the hotend.
All I could think of at the 41:05-mark is, that's an LGX Hydra taking form.
Have you looked into more detailed heater control in klipper? It looks like you have a couple of heaters in the big block, that you just control as one in the context of a bed heater. Klipper should be able to control a bunch of separate heaters, you could set them up all separately. You would need to drill a separate hole for each heater and install a thermocouple. In clipper you could then set them up as extruders oder other types of heaters. In the slicer you would just add a couple of commands to set them to the correct temperature.
Well done, no leaks!!’
Don’t group the ptfe tubes together that the top, you’re adding resistance
thanks for sharing your rational in details and all the numbers. super interesting.
Need to tune those retraction settings! :P
Pla can be printed under water, I would do water mist cooling or something of the sort, preferably chilled with ice or something. Nice work on the hot end 👍
Have you considered to run filament with different colors to see the mixing effect and how laminate the flow is? Great video though!
Have you thought about slightly widening the split in the filament path so that you can cut a thin slit in each of them to put a ceramic heater right In the center of each split? Also, im sure someone else has pointed it out. But it looks like your filament is super moist, all that bubbling and the fact that it snapped at the spool at one point. Youd probably get more accurate extruded volume without the water expanding creating voids and messing with the pressure
Great work, very impressive design. Why do you split the filament to increase the area that makes contact with filament? The centre is harder. Wouldn’t it be better to make the channel wider but not as deep to increase the contact area? Also, is copper the best material in terms of surface erosion over time, especially at the high temperatures you want to reach. I suppose the exposed parts can be plated if necessary.
I’d be very curious to see this as a possible injection molding extruder for small scale applications, where traditionally maybe a screw extruder would’ve been used. I think the excess heat capacity there would actually be a good thing because you want it to be liquid going into the die for the shot, and you’re melting it fast enough that you could do it in a relatively quick time as well.
Also, it’s just insane to me watching how much that filament expands coming out of a .5 mm nozzle 😂
If you can make ‘the Nozzle’ move XY in (for example) 5cm incremental steps quickly, then your ‘movable buildplate’ only has to move quickly 5cm x 5cm.
Maybe: If then the end tip of ‘the Nozzle’ is long enough to not bump to already printed parts, you can build a couple of layers at a time before advancing to the next 5x5 spot on the grid.
Are the main pathways balanced for length?
fucking awesome. thank you for showing the process with such detail
I heard yesterday that there is a conductive filament. I wonder if you could heat it by running a current through it?
Just before you said “I’m not a native speaker”, I was thinking about how well you were explaining your reasons for the differences in finishes along the path of the filament!
If you print 10X as fast as a normal printer, cooling might become an issue. How much time will one layer need to cool before the fan can cool the surface of a particular spot?
I would just add a question why didnt you use the 2.8mm filament?
I see this a the birth of a multicolor hotend with zero filament waste. I see a circular version in stead of a flat version, so a 2 by 2 instead of 4 by 1. I think this is very cool!
When using with a ‘normal’ bed, is there a way to move this thing in two resolutions? A quick and small resolution where the end nozzle is pivoted 1cmx1cm and a bigger step where the whole ‘cyclone nozzle’ is moved in cm increased steps.
10/10 if it looks good it performs good
I would say even 20/20!
Looks and performs... GOOD? 😂😂😂
Concerning the air cooling system since you plan to have a limited flow with pressurized air it could be interesting to have a conical nozzle with the outlet being thinner than the body. This should force the air to se a large variation of pressure when exiting the nozzle and loosing a few degrees at the same time. With conical nozzles, the change in pressure takes place mostly when entering the nozzle (at the bottleneck) and the air may heat up again in contact of the nozzle. It may not worth the pain of finding/making conical nozzles though...
Would you get any benefits if you cool the shop air
this is amazing ... well done mate
What motion system will it take to throw that weight around to match that flow rate?
Why don't you try ABS filament?
It has a better thermal conductivity
Wow, great engineering, very interesting 👍.
Just a constructive comment/criticism on your comment at about 36:20 in the video "I didn't receive additional money to review this". He had several tv and radio programs where he reviewed products from phones, computers, printers, routers, audio, television, projectors ... Every single product he reviewed he purchased and used himself so that he had "money in the game" just like a consumer.
Leo LaPorte: I would call Leo the GOAT in introducing computers, software, firmware, OSes to the masses in the 1980 Leo LaPorte. He was responsible for the radio broadcast "Dvorak on Computers" in January 1991. Leo co-created the tv series Screen Savers 1998.
Would "preheating" the filament before the extruder help at all? To 50*c or so, below the glass transition temp.
Could probably do this with heated spool containers and the PTFE tube running all the way to them, maybe with an insulating layer around it. Bonus is it would help dry the filament out too.
Looks great.
I don’t agree with the rough inner surface for forcing a shearing flow.
The extruder gears are pushing the cold filament as one slice, this will not shear like how you described.
I would love to see a custom pcb that controlls all the extruders, heaters and temp sensors right on the extruder.
Definitely follow up on how the solid “sock” works out!
oh and for a name its Frankenstruder cause that monster is built up from so many different parts :)
Super excited for the next episode!
Love it!! Will die swell not be a Problem to get a precise benchy shape?
We'll see. If it's becoming a problem we have to find a sweet spot between back pressure and a longer nozzle zone. ;)
@ looking forward to it! :)
How about pre-heating the filaments by having them in a temperature-controlled chamber ? (aka filament dryers just before the hotends)
I'm just curious to see how Benchy will handle the accelerations you need to print it at 300mm3/s
With this design there is another think you can do CMYK color mixing for a new type of colored 3d print