The exclude object part that BambuLab uses is also available in octoprint and klipper. It relies on the slicer "segmenting" the GCode with object start and end markers and just skipping those during GCode execution.
Another thing that you might have overlooked is that the printing orientation of the gear shafts. The shaft printed in that orientation are generally weaker due to the layers. And it can be seen as the main gear broke off several times. I hope this gets fixed in V4
This is the problem with printing gears. you have to either pick an orientation that leaves you with a weak shaft but strong teeth, or teeth that WILL guaranteed snap off an otherwise sturdy shaft. A decent solution to this is printing slower with higher temps and slightly higher extrusion rates along with bumping up the number of walls or just printing the gears thicker overall. Running hotter temps at slower speeds (with no cooling fan, obviously!) allows the plastic to fuse MUCH better and is a similar technique when making watertight models; the increased adhesion between the layers prevents them from shearing off. This is what needs to be in V4.
Your videos are fantastic! You are the reason I own an A1 mini. It was your last video, April 12th, that finally propelled me to get into 3d printing and try building your eye mechanism. And I loved using the mini so much I ended up also buying the A1 with AMS lite right when it came back in stock. Regrettable timing as I would have used your link!!!
Brilliant vid, thanks - so interesting to see what didn't work, as well as your ideas for v4. I'm rubbish when it comes to mechanical design, but I hope to get there and really appreciate guys like you pushing the boundaries, approaching print-in-place from 'different' angles, and doing genuinely novel design - even if I can't come up with such original solutions myself!
You can use fasteners for hinges two ways: use shoulder screws, which are only threaded partially at one end, or run the screws through a sleeve/bushing. The bolted pivots on a full suspension mountain bike are an obvious example.
I just wanna say how much you've inspired me to get working on robotics. If possible I'd love if you did some tutorials on how you think about CAD too. Keep up the awesome work man!
Very cool exploration -- definitely got my "gears turning", haha! I have made some very durable designs with cams and linkages in PLA by using finish nail shafts as low-cost pins, sandwiched in clamshell assemblies to support both ends. Where unsupported printed hinge pins are needed, biasing the contact patches toward the bases helps longevity by reducing leverage on them. Thin raised "rub rings" to reduce the contact patches between sliding and spinning parts also help -- it looks like you did this, but I prefer to bias mine toward the outer diameter for more stability and reduced contact force. I do have to say, this design seems like a big ask of any motor since there's so much rotating friction (it would be a lot even with ball bearings), and I wonder if you'll end up needing supplemental power sprinkled along the chain. Looking forward to the next version!
I should also note that one of the best additions I ever made to my vitamin selection was a 100-pack of cheap 6x3x2.5 microbearings. They're perfect for compact, low-RPM applications like this even when they're one step removed from actual garbage on the quality scale. Designing without them for an assembly like a centipede with a massive number of rotating parts makes sense from a DFM/DFA standpoint, in some ways, but when you're already resorting to nested insert-during-printing assemblies to combat friction and robustness issues, I wonder if resorting to embedded ball bearings might be the better part of valor this time :P
If the gcode is static (which it is), then you can store a special part code marking the start of each part at each layer. When running through the gcode command array during a print, and you come across one of these codes, you can skip it and move to the next. I don't know if they do this, but it would work at print-time. That the print job is accompanied by a picture, it implies that the printer receives more information than the g-code alone. I have a Bambu Labs printer. It's not that they're hyper advanced, just that they've capitalized on low-hanging-fruit for quality of life software features that could be a part of ANY printer (even the ultra-budget ones). They have also made some good design decisions here and there (eg. no-tool hotend swaps).
You are correct. Its add a Header and Footer to each section it does for a object on the bed. It looks at each object at slicing. So if you had it as 1 file not split up into parts or objects, it doesn't work. Then the firmware is just picking it up based on Gcode Comments Here is an example: ; stop printing object EC_R1_2a_r.stl id:4764331004008807296 copy 0 ; printing object EC_R1_2a_r.stl id:453636 copy 0
Yeah, the slicer adds comments mentioning which object it's going to print now. However, even without those (non-standardized) comments, the gcode could be analyzed to identify separate objects. The program could generate a map of where plastic gets extruded, and identify "islands" that aren't connected to each other. The thumbnail picture of the print job is in the gcode file! It's a small picture file that's been Base64-encoded into text and put into a comment.
I also struggled with proper hinges. I came up with using shoulder bolts with heated inserts. They come in a range from sizes and are super strong. Maybe right for this application
Try PLA Tough, all the strength of PLA with none of the brittleness issues. I use it exclusively for everything and stopped bothering with ASA/PETG/etc.
Hmm... are you designing in proper tolerances into your parts? Getting into the habit of throwing parameters in for tolerance and then being able to adjust it at the end parametrically is, IMO, a good practice to have for minimizing parts not fitting or working as the stimulation did.
I’ve never seen inserting parts during a print taken to this extreme, it definitely seems like it could be useful in some situations. Shafts really don’t do well printed vertically though, most 3D printing materials are weak in the Z direction due to layer adhesion. A technique I have seen to strengthen prints is to insert a bolt through the middle of the shaft and a nut at the other end, this adds reinforcement, the same could be done with any metal pin or bolt really.
Have you tried printing the parts which require a lot of support, from petg? If you set pla as interface layer, you can set the distance between support and model to zero to get a close to perfect overhang, and the pla snaps right off, allowing the support to be removed very easily.
When you go from V1 to V2, do you start over, or try to adopt? I always find that even though I'm using parametric design I end up with a small change causing huge issues down the timeline and it's either easier to start over or just do the non-parametric thing and update the design at the end (instead of changing the diameter from 3mm to 4mm, I start a new sketch and do a new extrusion to add the extra mm).
You mentioned friction issues multiple times, have you attempted 3d printed bearings with graphite powder as a lubricant? It does introduce a little slop but putting two in a gear on top and bottom can mitigate that for the most part This is a cool project and I can't wait to see where it goes
Excuse my ignorance, but doesnt the friction of the plastic of the gear in part cause to the breaking? Would a simple small bearing not address this issue?
The neck of the gears are breaking along the layers too. Another reason for there to be a shaft going through the gears instead of printing them all vertically.
Maybe if you skip the whole rotating shaft and design the legs to operate based on pulling a string you could make the whole leg mechanism a compliant mechanism with no assembly at all. This would also allow you to make the centepede a lot smaller. Ypu could probably get a ”wave pattern” in the legs by having a bit of slack in the string between each leg module, but might be difficult yo achieve, especially if you want the wave to contain multiple wavelengths along it’s body. Might be able to design some compliant mechanism that does multiple oscillations from a single pull, but that’s probably difficult. Would be really cool though to have a basically life sized remote controlled centepede :p
Interesting experiments! I've been thinking of playing more with mid-print inserted parts too, but it seems to be quite problematic. Prints detaching from the build plate, especially. I'm beginning to think it's wiser to just design the parts to be glued together afterwards!
2:32 please please what is this software. Im designing painfully piano mecanisms on Solidworks and I cannot find a simple software that permits me to try simple mecanisms on a very basic level. Im obliged to get super high in details to just see how it works and it's a pain to change numbers
Hey there! I found this video very inspiring, the multi stage printing sound like a very interesting idea. I have been designing mechanical parts for a while but never really got in depth with gearing systems, and fusion360 doesn't seem like the ideal program to explore them. Which program are you using to design these components or suggest using? Thank you for your help!
The centipede could have 2 motors for each leg for steering control, and if the design had some PID adjusting capabilities, it could walk along and turn in uneven terrain.
cool project. It broke a fundamental rule that made everything more tortuous: you over optimized cool hinges with overengineered joints. which are fun, but you could have done that after you tested the most likely failure point: the gearbox. you wasted too much brain in the cool factor, which made you negligent to critical parts. but again, really amazing project. and the over engineered parts are kind of fun, and I am quite sure that they will sell (in youtube) better than heartless efficient engineering. I will be waiting a part 2.
Pla is stronger than asa. Pla is one of the strongest filaments available. Only being best by polycarbonate. It is super brittle and lacks any tolerance to heat though.
Is is only me that felt audio was really f'd up!? You probably need a multi material printer with a multiple nozzle like Prusa Xl to get the results than AM's which is more of multi color.
Yup. Although there's a very niche time and place for these kind of assemblies. If your design is so utterly space or material constrained, it can be a cool optimization. 98% of the time, there's a better way, but this is a nice exploration of the possibilities of integrated assemblies nevertheless.
This is exactly what I should be watching at 2 am
The exclude object part that BambuLab uses is also available in octoprint and klipper. It relies on the slicer "segmenting" the GCode with object start and end markers and just skipping those during GCode execution.
Another thing that you might have overlooked is that the printing orientation of the gear shafts. The shaft printed in that orientation are generally weaker due to the layers. And it can be seen as the main gear broke off several times. I hope this gets fixed in V4
This is the problem with printing gears. you have to either pick an orientation that leaves you with a weak shaft but strong teeth, or teeth that WILL guaranteed snap off an otherwise sturdy shaft. A decent solution to this is printing slower with higher temps and slightly higher extrusion rates along with bumping up the number of walls or just printing the gears thicker overall. Running hotter temps at slower speeds (with no cooling fan, obviously!) allows the plastic to fuse MUCH better and is a similar technique when making watertight models; the increased adhesion between the layers prevents them from shearing off. This is what needs to be in V4.
I never thought about printing a gearbox like that, nice work!
Your videos are fantastic! You are the reason I own an A1 mini. It was your last video, April 12th, that finally propelled me to get into 3d printing and try building your eye mechanism. And I loved using the mini so much I ended up also buying the A1 with AMS lite right when it came back in stock. Regrettable timing as I would have used your link!!!
Brilliant vid, thanks - so interesting to see what didn't work, as well as your ideas for v4. I'm rubbish when it comes to mechanical design, but I hope to get there and really appreciate guys like you pushing the boundaries, approaching print-in-place from 'different' angles, and doing genuinely novel design - even if I can't come up with such original solutions myself!
You can use fasteners for hinges two ways: use shoulder screws, which are only threaded partially at one end, or run the screws through a sleeve/bushing. The bolted pivots on a full suspension mountain bike are an obvious example.
I just wanna say how much you've inspired me to get working on robotics. If possible I'd love if you did some tutorials on how you think about CAD too. Keep up the awesome work man!
Great print-in-place experiments. Don't give up!
welcome back Will. great project
Man all of your projects and videos are gold
When i ignored this video in my feed countless times i had no idea this was a will cogley video, definitely made me rethink and click :)
Very cool exploration -- definitely got my "gears turning", haha! I have made some very durable designs with cams and linkages in PLA by using finish nail shafts as low-cost pins, sandwiched in clamshell assemblies to support both ends. Where unsupported printed hinge pins are needed, biasing the contact patches toward the bases helps longevity by reducing leverage on them. Thin raised "rub rings" to reduce the contact patches between sliding and spinning parts also help -- it looks like you did this, but I prefer to bias mine toward the outer diameter for more stability and reduced contact force. I do have to say, this design seems like a big ask of any motor since there's so much rotating friction (it would be a lot even with ball bearings), and I wonder if you'll end up needing supplemental power sprinkled along the chain. Looking forward to the next version!
I should also note that one of the best additions I ever made to my vitamin selection was a 100-pack of cheap 6x3x2.5 microbearings. They're perfect for compact, low-RPM applications like this even when they're one step removed from actual garbage on the quality scale. Designing without them for an assembly like a centipede with a massive number of rotating parts makes sense from a DFM/DFA standpoint, in some ways, but when you're already resorting to nested insert-during-printing assemblies to combat friction and robustness issues, I wonder if resorting to embedded ball bearings might be the better part of valor this time :P
really interesting approach! as always, and indeed the bambu printers are impressive as well
Damn, I watched that whole thing expecting to see the functioning centipede at the end! I was so surprised when it ended lol
The bad way to tell the printer recommendation.
If the gcode is static (which it is), then you can store a special part code marking the start of each part at each layer. When running through the gcode command array during a print, and you come across one of these codes, you can skip it and move to the next.
I don't know if they do this, but it would work at print-time. That the print job is accompanied by a picture, it implies that the printer receives more information than the g-code alone.
I have a Bambu Labs printer. It's not that they're hyper advanced, just that they've capitalized on low-hanging-fruit for quality of life software features that could be a part of ANY printer (even the ultra-budget ones). They have also made some good design decisions here and there (eg. no-tool hotend swaps).
You are correct. Its add a Header and Footer to each section it does for a object on the bed. It looks at each object at slicing. So if you had it as 1 file not split up into parts or objects, it doesn't work.
Then the firmware is just picking it up based on Gcode Comments
Here is an example:
; stop printing object EC_R1_2a_r.stl id:4764331004008807296 copy 0
; printing object EC_R1_2a_r.stl id:453636 copy 0
Any printer running Klipper firmware can do part skipping. Not sure if marlin has this these days
Yeah, the slicer adds comments mentioning which object it's going to print now. However, even without those (non-standardized) comments, the gcode could be analyzed to identify separate objects. The program could generate a map of where plastic gets extruded, and identify "islands" that aren't connected to each other.
The thumbnail picture of the print job is in the gcode file! It's a small picture file that's been Base64-encoded into text and put into a comment.
@@jexomyess. Also Octoprint had a plug-in for years
@@jexom Marlin can do it when combined with octoprint
oh my, stumbled onto this in my feed today. did not dissapoint - new sub!
I also struggled with proper hinges. I came up with using shoulder bolts with heated inserts. They come in a range from sizes and are super strong. Maybe right for this application
Fantastic work. Great ecoerimental thinking and quite fascinating.
“Take that maths!” 😂I love it. These videos always blow my mind. Well done Will ❤
Try PLA Tough, all the strength of PLA with none of the brittleness issues. I use it exclusively for everything and stopped bothering with ASA/PETG/etc.
Very impressive work, hoping V4 works right out of the gate. Also the X1 Carbon is really good.
Cool projects and dedication.
Hmm... are you designing in proper tolerances into your parts? Getting into the habit of throwing parameters in for tolerance and then being able to adjust it at the end parametrically is, IMO, a good practice to have for minimizing parts not fitting or working as the stimulation did.
I’ve never seen inserting parts during a print taken to this extreme, it definitely seems like it could be useful in some situations.
Shafts really don’t do well printed vertically though, most 3D printing materials are weak in the Z direction due to layer adhesion. A technique I have seen to strengthen prints is to insert a bolt through the middle of the shaft and a nut at the other end, this adds reinforcement, the same could be done with any metal pin or bolt really.
Have you tried printing the parts which require a lot of support, from petg? If you set pla as interface layer, you can set the distance between support and model to zero to get a close to perfect overhang, and the pla snaps right off, allowing the support to be removed very easily.
Fantastic as always
You can solvent weld ASA and ABS with acetone if that will help stiffen your assembly.
Very good video, testing ideas and reflecting on the results. So I can learn. New subsciber.
When you go from V1 to V2, do you start over, or try to adopt?
I always find that even though I'm using parametric design I end up with a small change causing huge issues down the timeline and it's either easier to start over or just do the non-parametric thing and update the design at the end (instead of changing the diameter from 3mm to 4mm, I start a new sketch and do a new extrusion to add the extra mm).
You mentioned friction issues multiple times, have you attempted 3d printed bearings with graphite powder as a lubricant? It does introduce a little slop but putting two in a gear on top and bottom can mitigate that for the most part
This is a cool project and I can't wait to see where it goes
Excuse my ignorance, but doesnt the friction of the plastic of the gear in part cause to the breaking? Would a simple small bearing not address this issue?
The neck of the gears are breaking along the layers too.
Another reason for there to be a shaft going through the gears instead of printing them all vertically.
0:24 prehensile means that it can grab something ! Not bend in any direction.
Love the face cam!
Maybe if you skip the whole rotating shaft and design the legs to operate based on pulling a string you could make the whole leg mechanism a compliant mechanism with no assembly at all. This would also allow you to make the centepede a lot smaller.
Ypu could probably get a ”wave pattern” in the legs by having a bit of slack in the string between each leg module, but might be difficult yo achieve, especially if you want the wave to contain multiple wavelengths along it’s body. Might be able to design some compliant mechanism that does multiple oscillations from a single pull, but that’s probably difficult.
Would be really cool though to have a basically life sized remote controlled centepede :p
Interesting experiments! I've been thinking of playing more with mid-print inserted parts too, but it seems to be quite problematic. Prints detaching from the build plate, especially. I'm beginning to think it's wiser to just design the parts to be glued together afterwards!
great vid. also was any lubrication used for the gears? feel like that can be underrated in mechanical prints.
2:20 - what is there with your gears?
Binder pins also known as Chicago Screws :)
hmm maybe I should totally buy another X1C..or is there something else coming?
2:32 please please what is this software. Im designing painfully piano mecanisms on Solidworks and I cannot find a simple software that permits me to try simple mecanisms on a very basic level. Im obliged to get super high in details to just see how it works and it's a pain to change numbers
can i get a link to that chao print that was adorable and dope printing man
Yes! Forgot to upload it. Here you go: nilheim-mechatronics.notion.site/Baby-Chao-2a44c4763a964eab900b8d15cc3acd3e
neat ideas, thanks for the video
Welcome cogs is a risky opener but I think you could get away with it.
Hey there! I found this video very inspiring, the multi stage printing sound like a very interesting idea. I have been designing mechanical parts for a while but never really got in depth with gearing systems, and fusion360 doesn't seem like the ideal program to explore them. Which program are you using to design these components or suggest using? Thank you for your help!
Finally, my childhood dream of witnessing the arise of real life Zoids is ever closer
The centipede could have 2 motors for each leg for steering control, and if the design had some PID adjusting capabilities, it could walk along and turn in uneven terrain.
Two flexible shafts could also work. Also balances out the moments
Very cool video!
Would be interesting to do similar with 3D printed compliant mechanisms. Reduce part count.
cool project. It broke a fundamental rule that made everything more tortuous: you over optimized cool hinges with overengineered joints. which are fun, but you could have done that after you tested the most likely failure point: the gearbox. you wasted too much brain in the cool factor, which made you negligent to critical parts.
but again, really amazing project. and the over engineered parts are kind of fun, and I am quite sure that they will sell (in youtube) better than heartless efficient engineering. I will be waiting a part 2.
Wow I need to get that printer!
comparte el stl por favor
Take a 0.2mm nozzle for such small gears
Pla is stronger than asa. Pla is one of the strongest filaments available. Only being best by polycarbonate. It is super brittle and lacks any tolerance to heat though.
It's a 3D printed Inception, prints within prints within prints. Just need a kick to get it off the build plate 🤣
Is is only me that felt audio was really f'd up!?
You probably need a multi material printer with a multiple nozzle like Prusa Xl to get the results than AM's which is more of multi color.
Joseph Claw, Clan or Clown... I can't find info on that leg and subtitles aren't helping
Klaan!
Cool project, but not cool advertising for a company that's blocking up the 3D printing industry with patents again.
Wd40
*Calm music intro*
Welcome cOckS
Horrible to think about the people watching with a background in design/production yelling
Yup. Although there's a very niche time and place for these kind of assemblies. If your design is so utterly space or material constrained, it can be a cool optimization. 98% of the time, there's a better way, but this is a nice exploration of the possibilities of integrated assemblies nevertheless.
If you think you are a designer just because you can use a cad/ cam then you soon learn that you are ignorant
Dang Bambu sends X1's to influencers like candy i better get to uploading
Next step a mechanical human centipede 😈