Hey everyone! I hope you enjoyed this deep dive into the design process of the high-performance linear actuators for my juggling robot. I'd love to hear your thoughts on the design! What improvements would you suggest? Have you ever built or used linear actuators in your own projects? Share your experiences and tips in the comments below! 😃🤖🔧
As others have mentioned, one or two PTFE or Rulon J linear sleeve bearings would be lighter and probably stiffer than 6 6001RS steel bearings. You could also swap to acetal plastic ball bearings instead of steel to lighten them up and reduce wear. Other than that you could replace the push and pull strings with a single string and have a tensionning spring at one end to remove backlash. You could also have the pulleys be tensioned against each other to acheve the same result but that sounds clunkier.
For the string you might also look into braided uhmw, commonly used for high performance kites. It is not degraded by UV (if your robot has to perform outdoors) doesn't absorb water, and has far better abrasion resistance and has less friction than kevlar, also has nearly zero stretch for low backlash. Pretty cheap too. Excellent vid. Thanks for posting!
UHMWPE does have problems with creep, though. So you'd want to be sure that it's not stored under tension when not in use. Otherwise slack will creep in, and you'd need to re-tighten everything before you restart
You could probably fit the ptfe tubes inside the brake cable housing for a bicycle. Alternatively, you could design an print a modular reinforcement part that surrounds the tube and can carry the compressive load. This is a great design, very cool project!
Thanks for the kind words! Re. PTFE ideas: a big consideration for these actuators in this project is that they need to be able to move around fairly freely, and I'm worried that bike brake sheaths would be too stiff, and I can't visualise how a printed support structure would be both flexible (in bending) and stiff (in compression). I would love to know if you have any suggestions on that front! FWIW, a few people have suggested trying brake cables and I'll be testing that (and a few other) ideas soon. I'm sure we can eventually find the perfect tube! (or completely get rid of them, as Eric McGraw suggested in another comment. So many options!)
I was thinking about this while falling asleep last night, and I realized that simple beads-on-a-string will provide resistance against compression, but still allow flexibility while not under compression. Anything that allows the tube to freely pass through it, with a rounded edge to allow flexibility, would probably work. However, if it needs to remain flexible _while under compression_ it may be more of a challenge.
@@James02876 "beads-on-a-string" is basically the idea behind Elite Link and similar products. They allow smaller radius than conventional bowden tubes and at least one of such (of which I just can not remember the name) claims to be flexible under load… so further investigation may be worth it. I never used such myself because on all of my vehicles, classic bowden works just fine.
Of you go down the bicycle cable option. Try with gear cable and gear cable housing rather than brake cable (thinner diameter cable and come with far more housing and connector options)
Nice project! One little trick for getting strings (and wires) through tubes is to use a hoover to simply suck the string through! Works incredibly well ime 🙂
Lots of great comments and ideas on this project here in the comments. So impressed by the ingenuity of this design. Can’t wait to see future versions of this as you keep developing this.
Really nice. I wonder if it would be possible to route the Bowden string directly inside two of the carbon rod... This would make it looks like it's moving magically.
@@nemesis1588 Maybe turn the whole system inside out, where the actuator is a large tube on the outside, and the bearings are on the inside mounted on a smaller rod. Hide everything. Meh ... may make it too heavy.
Wow this project presentation was EXCELENT! Besides the design being impressive, the presentation was beyond what you can expect from the average engineering video. It was very easy to follow, well paced, and most crucially it covered all* your design constraints which produced the end design. That piece of context is so incredibly important to make the listener understand. It demonstrated a deep understanding of the issue, it's multiple solutions, and that your design really *is* near the ideal design for your particular use case. I felt like I actually understood the design, I didn't just observe it. That level of maturity in design presentation just blew me away. Personally I think you outdid "Stuff Made Here", which in my eyes is the gold standard of engineering videos on youtube. Although the comparison is nto entirely fair to "Stuff Made Here", as he obviously targets a more casual viewer.
This is absolutely brilliant work. I came up with a similar mechanism for my vacuum former. I needed to plunge the part piston rapidly to create a sort of surrounding bubble of soft plastic, before activating the vacuum. I'm surprised there aren't commercial products using this technique. Keep up the good work, I love your projects!
Cheers for the kind words! I agree that it's remarkable that this mechanism doesn't seem to exist on the internet anywhere else (at least not that I could find). Hopefully anyone else in the future who needs to build themselves a speedy actuator can now save themselves the ~1.5 yrs of development that I had to go through to get these working!
Wow man, it's so rare to find a channel this small with your level of production, presentation and co tent quality. Really amazing stuff. Seriously, not many channels with 4k subs who as articulatly and succinctly get complex stuff across like you. Keep going and you won't be small much longer. :):)
Consider mounting one bearing of each triplet on a small flexure preload spring. That way you don't need to adjust each guide. Stiffness up-to the preload will be determined by the remaining four fixed bearings. :)
Interesting idea! I think this might have the same issue as what Charlie Wynn was suggesting in another comment - the central tube should be perfectly centred in the actuator for what I'm using it for. Precision will be very important for Jugglebot and I want to try and keep everything as concentric as possible. Thankfully the spacers/guides are really easy to swap out and once you find a size that works, there's no need to change them again
@@harrisonlow if you’re worried about accuracy, I would adjust your attachment points on the rod for your strings. They look like they are at an angle in your video. Combined with the fact you’re only measuring distance on your motor, you’ve introduced potential backlash in the system, AND you’re measuring the change of the hypotenuse of the triangle you create with the string instead of the actual linear, parallel distance. It is probably not a lot at all, but in belt driven systems that need accuracy, the small angle can definitely throw off measurements.
@@JustTryGambling Good thinking! Something I didn't explicitly say in this video is that these actuators (there will be 6 of them) are being used to power a Stewart Platform, so the actuators need to move around quite a bit. The PTFE/string attachment points are angled to (hopefully) lessen the degree to which the PTFE inhibits the platform from moving around
Bicycle shifter cable and housing is super cheap and cand stand up to A LOT of tension. You could even include a barrel adjust in the system and easily adjust the tension of the cable on the fly.
Great concept! An easy fix for the string getting pinched would be a small screw eyelet. Have you considered using Bowden cables? That would likely need you to change from spools to cams to drive it though.
Hello, one of the string could go through the center of the other red printed part (the one attached to the base of the printer at the bottom of the 3 small carbon tubes), and go through the big carbon tube to the top. It shoud fix your issue with the string under the bearing, and use a more direct path :)
Good thinking! The main reason I haven't done that is that these actuators (there will be 6 of them) will be powering a Stewart Platform, so the actuators need to move around quite a bit, meaning it's tricky to route the string through the central tube since the central tube moves so much. I love the idea though, and it'd certainly make the design a lot cleaner. Cheers!
For the string running between the bearing and the slide, I would suggest 'v-groove' bearings, which as the name implies, have a groove along the face of the bearing. Maybe someone already suggested it, but just my 2 cents (or pence) :)
Super awesome! Another alternative to the bearings are igus drylin type r bearings. They are a split sleeve that rides very nice on carbon tubes and are reasonably affordable at around 5 dollars. We use them all the time at work for that purpose. If you put a slot and adjusting bolt in the 3d printed part you can set the preload/slop. You could probably even make a 3d printed spring feature to auto set it
Hello, try wrapping your ptfe tubes with a metal spring. will add some support about the length of the tubes in sections that demand a turn for fitment if you want to keep the cost low. otherwise you could consider a hybrid setup using metal cable bowden sections and then connecting the strings in the “interface” at the pushrod side and motor side. this should reduce stretch and keep the weight where it needs to be and where it shouldnt. keep in mind the material compatibility here, kevlar is strong and the ptfe will handle the abrasion for a little while, so periodic change outs may be required. nice job!
It can also be equipped to shake a quart or gallon of paint. It properly programmed and dressed, it can do the MJ Moon Walk! Thanks kindly for sharing!
Loving the progress! You can get rid of the bowden tubes and also stiffen your transmission coupling by guiding the cable around pulleys which rotate coaxially with your gimbal joints. The two cables need to wrap in opposite directions so that the total cable length remains constant. Normally this would slightly kinematically couple your linear motion with the rotation, but you can cancel this geometrically by offsetting the linear axis from the gimbal axis by the same radius as the pulleys. You can duplicate this concept rotated 90 degrees to get the other DOF of rotation. If you need help visualizing/designing the pulley layout, let me know, but I think it should be a feasible way of getting well-constrained cable motion to the actuator with almost zero flex/compliance or motion coupling. The pulleys could actually be a single, dual groove pulley since they will rotate in tandem, but you can also just use 2 parallel pulleys or one on each side of the joint itself.
Thanks! Hmm, I'm having difficulty visualising what you're describing. Do you know of any real-world systems that use something similar? I do quite like the sound of a tighter transmission 😁
@@harrisonlow Skyentific has shown a mechanism using pulleys co-axial with arm rotation, but for the purpose of actuating the arm itself. ua-cam.com/video/utDagouxM5U/v-deo.html The idea I'm suggesting simply uses such pulleys to guide the cable onto and back off of the rotating body. I will try drawing the idea a little later and sending a picture if you would find that helpful.
@@ericcmcgraw Very interesting! I just watched that video and I think I have a better idea of what you're describing. I have a few thoughts: 1) I'm quite averse to shifting the linear axis away from the centre of rotation as that makes the inverse kinematics for the stewart platform (what these actuators will be used for) significantly more difficult. I don't know exactly *how* much more difficult, but I started working through it a while ago and after a while decided to just make the joint axes coincident with the linear axis 😅 2) This design looks like it'd require quite a few parts, and at the scale I'm working at, they'd have to be fairly small. I'm trying to cut down on complexity as much as possible (especially where it concerns manufacturing). 3) As a slight (maybe) spoiler, I'm actually thinking I'll change the lower universal joint for a magnetic joint, in a similar fashion to the upper joints (this video if you haven't seen it already: ua-cam.com/video/ACidat_EQ3Q/v-deo.html ). I'm not sure if this would affect the design you're describing, but it seems relevant to mention. Again, thanks for the idea! I'm interested to know what you think about the above points 😊
@@ericcmcgraw Grr, I think UA-cam may have removed your last comment. I can see some of it in my notification, but nothing past a "small error" re. shifting the linear axis. Did you include any links? UA-cam seems to dislike links to external sites 😡
@@harrisonlow Yep, I tried obfuscating the link by spelling it out, but still no luck. Here's the comment with the link removed: 1) I thought of that too, so I played around in Geogebra and found that you could still assume the linear axis rotates through a single point and you'd have relatively small error compared to reality. Assuming ±45 degree range of rotation, and 10mm diameter pulleys, the actual length error would only be ±0.1 mm throughout the range of motion. Here's the Geogebra file if you want to play around with it: (redacted) There do exist solutions which would cancel this error exactly, but the complexity is higher still. Let me know if you're interested in them. 2) I've found some design possibilities which add only a few additional pulleys to the part count you already have. Complexity is a little higher, yes, but the benefits are: no sliding friction in bowden, no risk of bowden buckling, higher tension and higher stiffness capability. It's up to you whether it's worth it. 3) Yes I saw those ball joints, it's a pretty elegant solution and I like how it makes the actuators detachable and modular. Of course the only drawback, as you mention, is the pulling force capability. As you start to push the limits of acceleration (how many balls can it juggle at a time, etc.) you may want to be able to apply high forces in both directions without worrying about the joints falling apart. Again, it's up to your own judgement (and level of ambition 😁)
Heat the CF tubes up to the glass transition temperature of the filament you used (e.g. PETG = ~85°C), then push them in slow and steady. That should make things less violent! ;-) Instead of the 6 axial bearings, maybe look into using just one linear ball bearing (like LM8UU, but with larger inner diameter)? That should save a lot of weight, as you'll only need one. Centering it could be done via 3 bolts then. Use PTFE tubes with a smaller inner diameter and more wall thickness, they should get fairly rigid.
I wouldn’t recommend standard bearing or bushings on CF. You could go with a plastic bushing like the IGUS ones though and they seem to work okay, but there’s a bit of friction that needs to be overcome to get them moving. They also have to be preloaded quite heavy for tolerances to be good unless you go for their bushings that are already preloaded with a metal housing, which will add a bit more weight.
Sure, igus are great, i used them on one of my 3D printers (Prusa, got rid of that damn bed slinger). I didn't notice the friction, though - but maybe that was because they were on steel rods, not CF tubes, and had only a small contact surface at the top. But yeah, anything other than those axial bearings, or: any kind of a linear bearing, would be the perfect application for this - because that's what they're made for! 🥳
@@jangrewe Great ideas! Some thoughts: 1) I love the idea of heating up the CF tubes. I wish I saw this comment before I built the remaining actuators; some of them developed cracks from my... enthusiastic... hammering 😅Doesn't seem like anything structural was damaged, but heating would certainly be better. Cheers! 2) I previously tried a linear bearing instead of the 6 radial bearings, but the carbon fiber tubes that I'm using aren't specced for that level of tolerance; one manufacturers tubes had a loose fit, another didn't fit at all. Linear bearings would be awesome (smaller, lighter, more elegant) but I don't have the cash to get high tolerance, light tubes... yet 😉 3) Good idea for PTFE with larger wall thickness. Very obvious but never came to mind! I'll see if I can find any. Cheers for the ideas!
Harrison, great job! Suggestion; instead of using the PTFE tubes for cable guides, why not incorporate the cables into the carbon fiber structure tubes?
You'd still want cable guides (shifter cable housing seems like the best bet) into the bottom block but yes, using the structural tubing as part of the string guide is a great idea.
Cheers! Routing the strings through the CF tubes would probably work well for many applications, but I'm using these actuators (there will be 6 of them) to power a Stewart Platform and they need to move around quite a bit. Having the strings routed through the PTFE tubes means that the strings don't get in the way of the actuator moving around
@@harrisonlow i had the same idea, why not routing the strings through the CF tubes. I don't exactly undestand why would the strings get in way of the actuator?
This is just a simple idea, but you could keep the PTFE tubes and make some sort of either hard plastic or TPU sleeve to brace that goes around the tube to prevent buckling, just the parts that curve up to where the string meets the PTFE adapters/nuts so that the only part of the tube that has not bracing is just the straight-away section to the motor. Cool project nevertheless, some pretty cool innovation here! I always thought regular linear actuators were slow so this is the perfect design for some practical applications!
If you do decide on another iteration, you could use the three linear actuators to make a very large and potentially fast delta printer. Edited to add: Think Bike cables for the cable issues. The tubes have a twisted wire inside to ensure flex and length of housing unlike PTFE tubes that can ' crush ' under stress. You may be able to find similar gauge wires as the string or at least close enough for your purposes.
With the encoder for leg length measurement: it's going to be slightly non-linear due to the changing diameter of the bobbin/spindle, do you think it would need correction? If so, a simple lookup table could work well.
Haha I remember a question in high school math on this exact topic! Something about a toilet roll being unwound and dealing with the changing radius. I would be very interested to know how big of an effect that would have; my intuition says not much, but I've been wrong before 🤔
You can use a badge reel that winds and unwinds really small wire/string to measure linear travel without the same issue of the radius changing that you have using that CF spring. A commercially available sensor product that does this is a string potentiometer.
Great job mate, I really love cable mechanisms. Just as a fun fact, engines use half-bearings with different thicknesses to compensate the manufacturing variations between parts, your solution really made me remember that, they're also diferenciated by color. Again, great job!
Haha yep, colour differentiation helps so much. Got the idea from the adapters that came with the encoders to fit differently sized shafts. Such a simple solution! Cheers for the kind words 😊
What if the spacers were eccentric so you could rotate them for a perfect fit? Might not work with a 3d printed part. Seems like your solution with a few sizes works great already though 🙂
That's a very interesting idea! One potential issue I can see with that is that it'd make straightening the extending rod a little tricky. I didn't explicitly mention this in the video but I've been keeping each "set" of spacers (ie. top three or bottom three) the same size so that the central rod is kept centred. Having the spacers be eccentric might make this a little difficult. Very cool idea though and I'll keep that in mind if I need to do anything similar in the future. Cheers!
Random ideas: Get a CF tube that the PTFE tube fits in, put it in there and now its stiff. Use as one kf the trio of stiffening buddies. take 3 of the stabilization bearings, and put them on the internal end of the big moving CF rod, instead of bearing in onto the central rod, bearing out onto the trio of kuter rods. Would increase the weight of the reciprocating mass, but perhaps will reduce wobble? IDK, thats all i got so far. Great work. I really hope to see this go somewhere awesome!
I like the idea of bearings on the main push-rod! I hadn't thought of that before! Might be tricky for a few reasons: 1) While the CF tubes are stiff, having the central tube pushing against the outer tubes (via the bearing) might cause the outer tubes to flex a little 2) I think you've hit the nail on the head with respect to excess weight: if the actuator's moving at > 2m/s, I want to keep that as light as possible. Cheers for the suggestions, and I hope to see some more of these ideas in future designs!
what a great design, thank you for sharing. i'd go for cheap brake cables with a teflon liner, that way you get rid of the backlash and its essentially friction free.
Bicycle brake cables have excellent compression strength while being somewhat flexible and light! Also, they're relatively cheap. They are usually lined with some softer plastic but they are designed to house steel cable under high tension. may be worth a check, but not sure if the diameter of your aramid string is small enough to encourage cutting into the plastic liner of those cable housings.
Really nice job inventing such a fast-moving linear actuator system! I learned a lot of cool techniques, especially your ingenious way to handle the critical alignment requirements of the 3D-printed end parts with the custom spacers!
This looks very nice and I can't wait to see this running. @11:57 - is exactly why kiting shops don't sell these kevlar lines over here any more. These lines cut fingers too so we are now allowed to only used kevlar with a braided cover. @James Petersen suggested the use of bicycle brake cables but this is the wrong type to use. Outer brake cables can be compressed. You could try to use the outer gear cables instead, a lot of them even already have a PTFE liner inside to reduce friction. But do go out to a bike shop, tell them why you need these and check yourself if these are flexible enough for your use. Original Shimano cables are most likely way too stiff.
Cheers for the suggestions! That's interesting re. kevlar strings. Thanks for sharing that! Thankfully I'm pretty sure I have both brake and shifter sheathes lying around that came with cables for my bike. I'm planning on making a testing rig to test a bunch of the ideas that everyone has suggested, and I'll absolutely be trying shifter sheathes. Cheers!
Ive just just read every suggestion and thought of the most mentioned one's. String in support legs, use of bike cable sheathing, different shaped bearings. But didn't see any suggestion of just running a rubber pully wheel at the top of the 3 CF legs that snugs up against the CF tube with a tension adjuster. So your motor would spin the rubber wheel instead Would save string going to top and bottom of main CF leg. Obviously there could be slippage, just a thought. Great idea thou. Im subscribing............. Now. Thanks for sharing 👍
For the bearing that has the string running under. you could use one that has an inside radius. I have seen a few for larger rods where the inner radius is designed to match a metal rods curvature. You could use one of these to allow the string to travel through the bearing without coming into contact with it - also it would allow two contact points instead of 1 theoretically increasing the surface area of the bearing against the CF rod reducing wear.
YES! I've been thinking about this but couldn't find any bearings that fit that description. Do you happen to know what they're called? I tried "bearing inside radius" and a few variants but couldn't find anything. I suppose I could also use a smaller bearing and print an adapter with a groove for the string, though I wonder if the string would cut through it over time 🤔Only one way to find out!
I think your comment was removed by UA-cam but I can still see it in my notifications. I'll check out U groove and V groove bearings. They seem like exactly what I've been looking for! Thanks heaps for the suggestion!!
I've been doing some digging and thought you might be interested to know about a cool bearing profile that I found: "gothic groove" profile. Could work to hold the CF tube more solidly *and* have the string between the tube and the bearing in the small space in the profile. I might have to pick some up to test 🤔
I'm a CS guy, not an electro one. The linear actuator I saw seemed difficult to use, so I made a simple "linear actuator" for a kinetic sculpture with a stepper motor, a string, a tube and a spring. Nothing compared to what you did, but this made me realise of what potential just a simple motor + a wire can have, you definitely extracted the full power of this mechanism :D
Haha the potential isn't exhausted yet! I've been working on a improved version that is better in many ways. Will hopefully finish collecting data and putting the video together soon 😁
Great build. Just found your channel so excited to see what is next. In addition to the hoover method for string, I have used a can of compressed air to blow the string through. This works well on small strings in tubes too small to use the suction method. (Saved our butts at a robotics competition when rubber tubing began to stretch- we blew string through and tied it tight and it got us through.)
The ptfe tube buckling could be remedied with something like bicycle shift cable. The bending looks pretty high energy! Thats wild! Impressive machine youve built. Odrive looks amazing
Wow, that's an amazing mechanism! It's a bit like a hydraulic actuator, in some sense, because they typically (?) apply force through high pressure to one side of the piston at a time, rather than using pressure and vacuum on one side.
Very nice solution! To make it more long-lasting, I strongly suggest to put an outer liner of hard PU on the bearings with a slight groove. Rolling friction will of course increase a little but the contact pressure and noise will decrease radically. Another alternative is to use really thin-walled stainless tubes with carbon inside for strength and stiffness and the bearings as is.
Id recommend looking into either spectra or dyneema line as its EXTREMELY strong, zero stretch, abrasion resistance is off the charts and its very affordable. As for the play in the bowden tube, you could probably route the ptfe tube through the carbon fiber tubes or... You could use some rigid brake/fuel line or copper condenser coil tubing to route the ptfe through which would remove any loss of motion there resulting in a more accurate and faster action.
For the cables, you can reduce buckling if you pair the string with a tube whose inner diameter more closely matches the string diameter. That could mean a thinner tube, but it could also just mean thicker string: for instance, type 95 paracord has an outer diameter of 1.75mm, which would be a pretty good match for the tube you're using.
Yup, I 100% agree. I actually have some very thin PTFE tubes on hand but I don't have any way to fix them to the printed parts on each end, so I just opted for the larger ones. Now that I think about it, I might be able to put the smaller tube INSIDE the bigger tube 🤔 Thicker string is a very interesting idea that hadn't crossed my mind! I'll pick some up and give it a test! Cheers for the suggestions!
You could use car break lines. Basically the hydraulic equivalent to those pneumatic tubes and pass throughs. They can be bent easily and you should be able to find threaded connectors for the plastic parts
This is awesome! I recommend bike shifting cable lubricated with silicone spray, its thinner and has a lot less resistance than bike brake cable. The housing is reenforced to prevent buckling and you can easily refill it with silicone if it starts to stick
Mountain Bike gear change cable outers would be perfect to replace your string guide pipe. Brake cables are just wound steel outer, but the gear change cables are not spiral bound and have a teflon/ptfe liner - they are designed for both push and pull actions with low friction.
tldr: both ptfe tubes can probably terminate at the bottom part. thin brass/copper pipe for redirection. down-stroke the down-stroke string can pull the tube towards the bottom directly, instead of pulling on the topmost part. This allows more weight to be shifted to the less moving part and reduces ptfe tube length. It would require a redirection bushing or similar in the bottom part, but as that moves less i think it might be advantageous. up-stroke the up-stroke string can run alongside one of the structural carbon fiber tubes to the top part, where it gets redirected back down. this reduces ptfe tubing and probably tube compression issues. redirection solutions for redirecting the string, i would evaluate whether a small bent brass tube which you feed the string through wouldn't be enough. all this would require is a section on the printed part to glue this into (so that it doesnt fall out while assembling/maintaining, while in use the string tensions should keep it in place on its own) which might be easier for assembly and even lighter. another crazy idea: it might be possible to reduce this design to a pair of telescoping pipes, by using a ha:f open linear bearing (like the LM12UU OP) at the top and running the string between the 2 carbon fiber pipes. though this would likely require a larger pipe for the outer part. This would probably be overengineering for this usecase tho.
For a cable transmission system - maybe look into bicycle brake cables or even the twist-throttle mechanism of a motorcycle. These already do what you are trying to accomplish at forces and levels or reliability magnitudes greater.
You should use capricorn bowden tubes. The blue ones. I found they're much stiffer and less easy to bend than the normal ones. Perfect for this application.
As you said at the end the PTFE tubes will 'wear out' over time. Have you thought about using brake cable covers/sleeves from bicycles? If you can find something that has a proper inner diameter. Those are lined with either composite or metal weave which makes them much more wear and longitudinal compression resistant
Nice work! Inspiring! I had issues with PTFE tubing buckling (from compression) when I used them as bowden cables. I've wanted to try the small coils (smaller than bicycle kind) but I never got around to it (and project had already met its goals).
Hydraulics have uses, but as more all electric fast acting and accurate systems are built going forward. Iinear actuators and servos will become even more prominent in new designs. Nice job 👍
Really enjoyed this video. I think the youtube algorithm may have removed my last comment due to it having links in it so I'll try again, without links. A possible replacement for the PTFE tubes could be "armour spring" that is used to protect loop connections on fishing rigs. It is highly resistant to crush and compression forces and is very flexible and light. There are two potential issues. 1. It is usually only available in packets of 10 x 10cm lengths. 2. It is usually green. Regarding issue 1. I tried joining two lengths with a small section of heatshrink. This worked well. I set it up in a rig using 100lb dyneema and was able to lift 3Kg without any permanent deformation of the armour spring. Regarding issue 2. I see that there is 1.2mm ID armour spring in the Shogun brand that is red. You may have a tackle shop nearby where you can pick some up to play with. It may even be possible to wind your own using suitable coated stainless wire.
Thanks for the kind words 😊 I've never heard of armour springs! They seem to be difficult to find good information about... Can't even find them on Amazon! (I don't have a tackle shop super close by). Do they go by other names? As it happens, I actually have a few lengths of long, narrow springs on hand so I'll play around with them. Hopefully the "armour" part of "armour spring" isn't too critical 😉 Cheers for the idea!
Just gave this a quick test and it seems like a promising idea! Working prototype is a (~2.5 mm ID, 3.25 mm OD) spring that's about 50 cm long. The spring conveniently fits the PC4-M6 PTFE connectors that I already have, and it's working pretty well! I'll give this idea a proper go some time in the next few days. My biggest concerns right now are: 1) friction between the string and the spring; and 2) sourcing the springs if I were to do this at-scale. I have these ones as a leftover from a uni project a few years ago and I don't think they're very easy to source in Australia...
Great video! Can't wait for the next one! I don't have any experience with cabling systems either, but I have used pneumatic tubing like this for other projects (something like McMaster 9446K21) and it's considerably stiffer than bowden tubing. It won't be as slippery a ptfe, though, so it's possible the tubing could melt at high enough speeds.
Thanks for the kind words 😊 Interesting! By stiff, do you mean in compression, or bending? Jugglebot needs fairly flexible tubes (in bending) to not restrict its range of motion. I WISH we had McMaster Carr in Australia. Honestly such a pain to source parts that are so easily available in the US 😭
@@harrisonlow In compression - since the walls are thicker, it's much less likely to buckle. It might not be quite as bendable as a bowden tube, but I think it would bend just fine even at the speeds you're running. They should be pretty common, lots of home air compressors use similar tubing for their hose.
@@tategeiger5317 Thanks for the suggestion - I will start the hunt for some pneumatic tubing! I agree re. bending stiffness: it wouldn't need to be as flexible as the PTFE to be a non-issue for my purposes
@12:08 My friend did this with a $250k capstan driven robot, but the line started slicing through the aluminium extrusions. Really impressive what thin line can do under tension.
You could make one of the bearings be in a tab that flexes a bit. You flex it out to fit the rod then the bearing will have tension against the rod. It's like making the bearing be pushed by a spring. I would recommend PC filament for this kind of thing as it retains its shape well when tensioned and is strong
Also, with a cable, it is possible to push and pull, though a return spring directly on the opposite end of the shaft can have accurate repeatability, less parts, and so on. Simply switch your ptfe tubes to 1mmx4mm or omething with a thicker wall.
I think, that if you'll get a wire going through a ptfe tube less flexible and with exact diameter of inner diameter of a tube, you won't have trouble with tube bending and collapsing. Great project!
Hi!!! You can use a V profile bearing, perhaps using TPU as a bearing cover. I think that will provide more stability (at least two contact points per bearing). This V profile can allow you to run the thread close to the apex of the V.
You could use PA(nylon) tubing instead of a ptfe, they're way more ridgid and are meant to withstand +100bar of hydraulic pressure (they're used for grease in industrial machines). You can keep the fittings as well since they're available in 4&6mm outer diameter
Try an IGUS linear bearing/bushing instead of 6 bearings. You will need to preload the bushings perfectly to achieve the right tolerance that’s not too tight or too loose, but it should be a lot lighter for you.
Interesting! I previously tried a linear bearing instead of the 6 radial bearings, but the carbon fiber tubes that I'm using aren't specced for that level of tolerance; one manufacturers tubes had a loose fit, another didn't fit at all. I didn't know you could get bearings/bushings that you could preload. I'll look into them! Cheers!
How to get rid of the bowden tubes? Why not use a flex shaft to relocate the drum from the motor shaft to the base of the actuator? This would also save space by allowing you to attach the pull cable to the bottom of the actuator shaft. The pull side of the drum would be centered under the shaft. The push cable would still need to run up to the pully and back down to the bottom of the shaft. No more strings running to the top. How to remove the backlash? Why not just add a spring to your newly relocated drum mount and turn it into a tensioner similar to a serpentine belt tensioner? A TEMO flex shaft is like $15.
Interesting! I'm not sure I 100% understand what you're suggesting, but I have a few thoughts related to this: 1) I'm interested by the idea of flex shafts / push-pull cables (are these the same things?). I've seen people suggest them on the Wintergatan project but haven't had the time to look into them much. Are they particularly flexible? 2) For my application (Stewart Platform) it's somewhat useful to have the motor be physically separate from the actuator itself, as the actuator will be moving around and the motor would get in the way and just be more mass to move around. 3) An unintended boon to the way I have the strings routed now is that the end caps on the central tube are only ever pulled in towards the tube. This is rather nice as it means that they'll never get pulled off if the string pulls too fast/hard, since they're only connected via a not-super-strong interference fit. I could of course make that fit better, but that would be more effort! Re. backlash removal: Yes! I agree with your idea of the spring, and this is something that I will absolutely be adding if backlash becomes an issue for whatever reason. Cheers for the ideas!
Another suggestion for your string issue - try having a chat with Otis Elevators, they use a flat rubberised material in their elevator shafts instead of cables, they may have a few ideas that you can experiment with.
You may remove tubes completly, just move the motor to one of the end of actuator. Also , for brushless motor, you can use the body of motor as the base for spool - that way you have ability to center the motor relative to the plane along the actuator. Event more, you can use closed loop string, or even chain.
You only need the extension string. The retraction string can be taken off and you can use the constant force spring to retract the actuator. That way there is no string visible on the extended actuator which is the weak link in terms of ability to snap whilst being pinched by external factors. Cool project!
It's funny you mention that; that's actually how the last design worked! Check out this video if you're interested: ua-cam.com/video/h-th1lcGFq0/v-deo.html
Bike shifter cable could be good, maybe try reinforcing the ptfe tubes with a metal spring on the outside, you could probably turn one yourself. Nice video.
Actually I was thinking along the bike cable line myself. Maybe he could run bike cables thru the ptfe that's in the actuator but tie the string to the cable at the motor end. So kind of a string and cable hybrid. Also for the cable routing issue, he should change the angle of the one fitting location so that it goes all the way up to the top and sleeve ptfe thru ,having the cable exit on the top edge instead of exiting down inside next to the pully center.
The fact that this kevlar-string version of actuation is much more effective than pneumatic or stepper is a strong defense of how evolution selected muscles+tendons for biological actuation
thats a neat actuator! one problem i can think of is the off axis load created by the cables pulling on one side of the rod. this might not be an issue with light loads, but with higher forces along its extending & retracting axis it would require increasingly high lateral stiffness between the bearing assembly and the rod. as for the ptfe lines. maybe running them through the existing carbon fiber rods could help. it shouldnt interfere too much with the movement of the base hinge as long as you run it directly through the pivot axis of each hinge. if that is too complicated / would require too much precision, you could still shorten the length of the ptfe tubes significantly by only running them from the base of the carbon rods to the motors.
Awesome work, love the progression! With the bearings, you're running the hardened steel outers against (relatively) soft carbon fibre tube, which will have a bit of diameter variation along its length. The contact area between the bearing and tube is pretty small, so will exert a very high pressure. You'd be better to increase this contact area by using smaller bearings and adding a plastic sleeve that fits the tube diameter. You can buy these as track rollers, but you might not be able to find one for the exact diameter of tube you have. The sleeves could be 3D printed then turned true on a lathe. If you haven't got access to a lathe, you could use a drill with a mandrel and a piece of sand paper wrapped around a rod. The plastic sleeve will give a bit of compliance too so you can preload the bearings without risking damaging the CF tubes. Look forward to seeing the progress!
You should consider a capstan style of transmission to the rope. Also, if you want to ba able to desassemble the tubes from the plastic, you can make a full cylinder, drill it in the center and screw in a slightly oversized screw to make it press fit
Very nicely structured and well explained video! Keep up the good work! Edit: i really like the simplicity of the design with the range of motion it offers. Thinking outside the box work here!
Check out IGUS linear bearings made from plastic to reduce weight. Or shaft guides. They also make square ones. And also square carbon fiber tubes exist. Maybe some telescopic thing can be built with these to get the absolute lowest weight.
Haha it's funny you mention that - the earlier design (the third one in the "Design Iterations" chapter) had space for not just two, but *three* pulleys for the exact reason you're describing. Long story short, the string is close enough to the central tube and the bearings do a good enough job of containing the bend that it isn't an issue (that I've observed, at least), and routing extra strings + making sure they're the *exact* same length is very tricky/tedious. Thanks!
Brilliant job. The string which pulls down could be come from the bottom (part standing on the table). This saves half the boden strings connected to the slider part. Also no string going up along the center rod when it extends. So at the same spot were you connect the string pulling up you hook the string going down. As this is a straight line to the bottom part there is no cover needed (plain string should do).
Try printing TPU sleeves for the bearings with a concave shape to exactly match the piston cylinder. Maybe you can get rid of more wobble by squishing the rollers in since they can now flex a little, but that would add a ton of friction so there's a tradeoff.
Hey everyone! I hope you enjoyed this deep dive into the design process of the high-performance linear actuators for my juggling robot. I'd love to hear your thoughts on the design! What improvements would you suggest? Have you ever built or used linear actuators in your own projects? Share your experiences and tips in the comments below! 😃🤖🔧
this is slick! have you considered routing the cables through the actuator arm itself?
As others have mentioned, one or two PTFE or Rulon J linear sleeve bearings would be lighter and probably stiffer than 6 6001RS steel bearings. You could also swap to acetal plastic ball bearings instead of steel to lighten them up and reduce wear.
Other than that you could replace the push and pull strings with a single string and have a tensionning spring at one end to remove backlash. You could also have the pulleys be tensioned against each other to acheve the same result but that sounds clunkier.
14:20 use a ceramic eye from a fishing rode for strung position 😉
u can use the bike/bicycle brake tubes those are quite tough instead of the ptfe
I would put bearings on the 3 rods, back end of the extending arm, just for a bit more stabilization.
For the string you might also look into braided uhmw, commonly used for high performance kites. It is not degraded by UV (if your robot has to perform outdoors) doesn't absorb water, and has far better abrasion resistance and has less friction than kevlar, also has nearly zero stretch for low backlash. Pretty cheap too. Excellent vid. Thanks for posting!
I was thinking the same but with Braided monofilament as used in fishing, for the same reasons.
UHMWPE does have problems with creep, though. So you'd want to be sure that it's not stored under tension when not in use. Otherwise slack will creep in, and you'd need to re-tighten everything before you restart
Use SK99 for the lowest stretch and creep.
You could probably fit the ptfe tubes inside the brake cable housing for a bicycle. Alternatively, you could design an print a modular reinforcement part that surrounds the tube and can carry the compressive load. This is a great design, very cool project!
Thanks for the kind words!
Re. PTFE ideas: a big consideration for these actuators in this project is that they need to be able to move around fairly freely, and I'm worried that bike brake sheaths would be too stiff, and I can't visualise how a printed support structure would be both flexible (in bending) and stiff (in compression). I would love to know if you have any suggestions on that front!
FWIW, a few people have suggested trying brake cables and I'll be testing that (and a few other) ideas soon. I'm sure we can eventually find the perfect tube! (or completely get rid of them, as Eric McGraw suggested in another comment. So many options!)
@@harrisonlow Maybe Jagwire Elite Link can do the job.
I was thinking about this while falling asleep last night, and I realized that simple beads-on-a-string will provide resistance against compression, but still allow flexibility while not under compression. Anything that allows the tube to freely pass through it, with a rounded edge to allow flexibility, would probably work. However, if it needs to remain flexible _while under compression_ it may be more of a challenge.
@@James02876 "beads-on-a-string" is basically the idea behind Elite Link and similar products. They allow smaller radius than conventional bowden tubes and at least one of such (of which I just can not remember the name) claims to be flexible under load… so further investigation may be worth it. I never used such myself because on all of my vehicles, classic bowden works just fine.
Of you go down the bicycle cable option. Try with gear cable and gear cable housing rather than brake cable (thinner diameter cable and come with far more housing and connector options)
Nice project! One little trick for getting strings (and wires) through tubes is to use a hoover to simply suck the string through! Works incredibly well ime 🙂
Damn, that's a good trick
You can also use pressurized air and it will blow it right though. My wife has an embroidery machine and that's how she loads the threads.
Lots of great comments and ideas on this project here in the comments. So impressed by the ingenuity of this design. Can’t wait to see future versions of this as you keep developing this.
I agree! My notepad is rapidly filling up with suggestions. So good!
Really nice. I wonder if it would be possible to route the Bowden string directly inside two of the carbon rod... This would make it looks like it's moving magically.
If two of the bearings that holds the shaft straight had a channel in it to guide the string this could be done easily.
exactly what i was thinking
@@nemesis1588 Maybe turn the whole system inside out, where the actuator is a large tube on the outside, and the bearings are on the inside mounted on a smaller rod. Hide everything. Meh ... may make it too heavy.
Wow this project presentation was EXCELENT!
Besides the design being impressive, the presentation was beyond what you can expect from the average engineering video. It was very easy to follow, well paced, and most crucially it covered all* your design constraints which produced the end design. That piece of context is so incredibly important to make the listener understand. It demonstrated a deep understanding of the issue, it's multiple solutions, and that your design really *is* near the ideal design for your particular use case. I felt like I actually understood the design, I didn't just observe it.
That level of maturity in design presentation just blew me away. Personally I think you outdid "Stuff Made Here", which in my eyes is the gold standard of engineering videos on youtube. Although the comparison is nto entirely fair to "Stuff Made Here", as he obviously targets a more casual viewer.
Wow! Thanks for the great feedback!
This is so interesting! People like you are the future of technology, not big corporations. Thank you for sharing this! Subscribed!
Thanks for the support!
This is absolutely brilliant work. I came up with a similar mechanism for my vacuum former. I needed to plunge the part piston rapidly to create a sort of surrounding bubble of soft plastic, before activating the vacuum. I'm surprised there aren't commercial products using this technique.
Keep up the good work, I love your projects!
Cheers for the kind words! I agree that it's remarkable that this mechanism doesn't seem to exist on the internet anywhere else (at least not that I could find). Hopefully anyone else in the future who needs to build themselves a speedy actuator can now save themselves the ~1.5 yrs of development that I had to go through to get these working!
Wow man, it's so rare to find a channel this small with your level of production, presentation and co tent quality. Really amazing stuff. Seriously, not many channels with 4k subs who as articulatly and succinctly get complex stuff across like you. Keep going and you won't be small much longer. :):)
Cheers!
Consider mounting one bearing of each triplet on a small flexure preload spring. That way you don't need to adjust each guide. Stiffness up-to the preload will be determined by the remaining four fixed bearings. :)
Interesting idea! I think this might have the same issue as what Charlie Wynn was suggesting in another comment - the central tube should be perfectly centred in the actuator for what I'm using it for. Precision will be very important for Jugglebot and I want to try and keep everything as concentric as possible. Thankfully the spacers/guides are really easy to swap out and once you find a size that works, there's no need to change them again
@@harrisonlow if you’re worried about accuracy, I would adjust your attachment points on the rod for your strings. They look like they are at an angle in your video. Combined with the fact you’re only measuring distance on your motor, you’ve introduced potential backlash in the system, AND you’re measuring the change of the hypotenuse of the triangle you create with the string instead of the actual linear, parallel distance. It is probably not a lot at all, but in belt driven systems that need accuracy, the small angle can definitely throw off measurements.
@@harrisonlow as long as the system is stiff enough, can't you compensate for the lack of concentricity through software?
@@JustTryGambling Good thinking! Something I didn't explicitly say in this video is that these actuators (there will be 6 of them) are being used to power a Stewart Platform, so the actuators need to move around quite a bit. The PTFE/string attachment points are angled to (hopefully) lessen the degree to which the PTFE inhibits the platform from moving around
@@LimitedWard Yeah, probably! Though I want to limit the amount of hard-coding of this sort of physical "defect" that I need to do
Bicycle shifter cable and housing is super cheap and cand stand up to A LOT of tension. You could even include a barrel adjust in the system and easily adjust the tension of the cable on the fly.
Great concept! An easy fix for the string getting pinched would be a small screw eyelet. Have you considered using Bowden cables? That would likely need you to change from spools to cams to drive it though.
Hello, one of the string could go through the center of the other red printed part (the one attached to the base of the printer at the bottom of the 3 small carbon tubes), and go through the big carbon tube to the top. It shoud fix your issue with the string under the bearing, and use a more direct path :)
Good thinking! The main reason I haven't done that is that these actuators (there will be 6 of them) will be powering a Stewart Platform, so the actuators need to move around quite a bit, meaning it's tricky to route the string through the central tube since the central tube moves so much. I love the idea though, and it'd certainly make the design a lot cleaner. Cheers!
For the string running between the bearing and the slide, I would suggest 'v-groove' bearings, which as the name implies, have a groove along the face of the bearing. Maybe someone already suggested it, but just my 2 cents (or pence) :)
Super awesome! Another alternative to the bearings are igus drylin type r bearings. They are a split sleeve that rides very nice on carbon tubes and are reasonably affordable at around 5 dollars. We use them all the time at work for that purpose. If you put a slot and adjusting bolt in the 3d printed part you can set the preload/slop. You could probably even make a 3d printed spring feature to auto set it
Hello, try wrapping your ptfe tubes with a metal spring. will add some support about the length of the tubes in sections that demand a turn for fitment if you want to keep the cost low. otherwise you could consider a hybrid setup using metal cable bowden sections and then connecting the strings in the “interface” at the pushrod side and motor side. this should reduce stretch and keep the weight where it needs to be and where it shouldnt. keep in mind the material compatibility here, kevlar is strong and the ptfe will handle the abrasion for a little while, so periodic change outs may be required. nice job!
It can also be equipped to shake a quart or gallon of paint. It properly programmed and dressed, it can do the MJ Moon Walk!
Thanks kindly for sharing!
The possibilities are endless! 😁
Loving the progress! You can get rid of the bowden tubes and also stiffen your transmission coupling by guiding the cable around pulleys which rotate coaxially with your gimbal joints. The two cables need to wrap in opposite directions so that the total cable length remains constant. Normally this would slightly kinematically couple your linear motion with the rotation, but you can cancel this geometrically by offsetting the linear axis from the gimbal axis by the same radius as the pulleys. You can duplicate this concept rotated 90 degrees to get the other DOF of rotation. If you need help visualizing/designing the pulley layout, let me know, but I think it should be a feasible way of getting well-constrained cable motion to the actuator with almost zero flex/compliance or motion coupling. The pulleys could actually be a single, dual groove pulley since they will rotate in tandem, but you can also just use 2 parallel pulleys or one on each side of the joint itself.
Thanks!
Hmm, I'm having difficulty visualising what you're describing. Do you know of any real-world systems that use something similar? I do quite like the sound of a tighter transmission 😁
@@harrisonlow Skyentific has shown a mechanism using pulleys co-axial with arm rotation, but for the purpose of actuating the arm itself.
ua-cam.com/video/utDagouxM5U/v-deo.html
The idea I'm suggesting simply uses such pulleys to guide the cable onto and back off of the rotating body. I will try drawing the idea a little later and sending a picture if you would find that helpful.
@@ericcmcgraw Very interesting! I just watched that video and I think I have a better idea of what you're describing. I have a few thoughts:
1) I'm quite averse to shifting the linear axis away from the centre of rotation as that makes the inverse kinematics for the stewart platform (what these actuators will be used for) significantly more difficult. I don't know exactly *how* much more difficult, but I started working through it a while ago and after a while decided to just make the joint axes coincident with the linear axis 😅
2) This design looks like it'd require quite a few parts, and at the scale I'm working at, they'd have to be fairly small. I'm trying to cut down on complexity as much as possible (especially where it concerns manufacturing).
3) As a slight (maybe) spoiler, I'm actually thinking I'll change the lower universal joint for a magnetic joint, in a similar fashion to the upper joints (this video if you haven't seen it already: ua-cam.com/video/ACidat_EQ3Q/v-deo.html ). I'm not sure if this would affect the design you're describing, but it seems relevant to mention.
Again, thanks for the idea! I'm interested to know what you think about the above points 😊
@@ericcmcgraw Grr, I think UA-cam may have removed your last comment. I can see some of it in my notification, but nothing past a "small error" re. shifting the linear axis. Did you include any links? UA-cam seems to dislike links to external sites 😡
@@harrisonlow Yep, I tried obfuscating the link by spelling it out, but still no luck. Here's the comment with the link removed:
1) I thought of that too, so I played around in Geogebra and found that you could still assume the linear axis rotates through a single point and you'd have relatively small error compared to reality. Assuming ±45 degree range of rotation, and 10mm diameter pulleys, the actual length error would only be ±0.1 mm throughout the range of motion. Here's the Geogebra file if you want to play around with it: (redacted) There do exist solutions which would cancel this error exactly, but the complexity is higher still. Let me know if you're interested in them.
2) I've found some design possibilities which add only a few additional pulleys to the part count you already have. Complexity is a little higher, yes, but the benefits are: no sliding friction in bowden, no risk of bowden buckling, higher tension and higher stiffness capability. It's up to you whether it's worth it.
3) Yes I saw those ball joints, it's a pretty elegant solution and I like how it makes the actuators detachable and modular. Of course the only drawback, as you mention, is the pulling force capability. As you start to push the limits of acceleration (how many balls can it juggle at a time, etc.) you may want to be able to apply high forces in both directions without worrying about the joints falling apart. Again, it's up to your own judgement (and level of ambition 😁)
Skyentific did a video a few months ago exploring different bowden tube types for cables. I found it very helpful.
Heat the CF tubes up to the glass transition temperature of the filament you used (e.g. PETG = ~85°C), then push them in slow and steady. That should make things less violent! ;-)
Instead of the 6 axial bearings, maybe look into using just one linear ball bearing (like LM8UU, but with larger inner diameter)? That should save a lot of weight, as you'll only need one. Centering it could be done via 3 bolts then.
Use PTFE tubes with a smaller inner diameter and more wall thickness, they should get fairly rigid.
I wouldn’t recommend standard bearing or bushings on CF. You could go with a plastic bushing like the IGUS ones though and they seem to work okay, but there’s a bit of friction that needs to be overcome to get them moving. They also have to be preloaded quite heavy for tolerances to be good unless you go for their bushings that are already preloaded with a metal housing, which will add a bit more weight.
Linear bearing is a top idea, I would recommend a Teflon bushing though as it will work well with CF
Sure, igus are great, i used them on one of my 3D printers (Prusa, got rid of that damn bed slinger). I didn't notice the friction, though - but maybe that was because they were on steel rods, not CF tubes, and had only a small contact surface at the top.
But yeah, anything other than those axial bearings, or: any kind of a linear bearing, would be the perfect application for this - because that's what they're made for! 🥳
@@jangrewe Great ideas! Some thoughts:
1) I love the idea of heating up the CF tubes. I wish I saw this comment before I built the remaining actuators; some of them developed cracks from my... enthusiastic... hammering 😅Doesn't seem like anything structural was damaged, but heating would certainly be better. Cheers!
2) I previously tried a linear bearing instead of the 6 radial bearings, but the carbon fiber tubes that I'm using aren't specced for that level of tolerance; one manufacturers tubes had a loose fit, another didn't fit at all. Linear bearings would be awesome (smaller, lighter, more elegant) but I don't have the cash to get high tolerance, light tubes... yet 😉
3) Good idea for PTFE with larger wall thickness. Very obvious but never came to mind! I'll see if I can find any.
Cheers for the ideas!
@@harrisonlow cheers, 2 out of 3! ;-) I'll let you know if i can think of anything else...
This video was such a joy to watch. Persistence has paid off. What a clever idea.
Thanks for the kind words 😊
Harrison, great job! Suggestion; instead of using the PTFE tubes for cable guides, why not incorporate the cables into the carbon fiber structure tubes?
You'd still want cable guides (shifter cable housing seems like the best bet) into the bottom block but yes, using the structural tubing as part of the string guide is a great idea.
Cheers!
Routing the strings through the CF tubes would probably work well for many applications, but I'm using these actuators (there will be 6 of them) to power a Stewart Platform and they need to move around quite a bit. Having the strings routed through the PTFE tubes means that the strings don't get in the way of the actuator moving around
@@harrisonlow i had the same idea, why not routing the strings through the CF tubes. I don't exactly undestand why would the strings get in way of the actuator?
This is just a simple idea, but you could keep the PTFE tubes and make some sort of either hard plastic or TPU sleeve to brace that goes around the tube to prevent buckling, just the parts that curve up to where the string meets the PTFE adapters/nuts so that the only part of the tube that has not bracing is just the straight-away section to the motor. Cool project nevertheless, some pretty cool innovation here! I always thought regular linear actuators were slow so this is the perfect design for some practical applications!
Seeing that new design in action made me realize that this would also work well for a different kind of machine...
If you do decide on another iteration, you could use the three linear actuators to make a very large and potentially fast delta printer. Edited to add: Think Bike cables for the cable issues. The tubes have a twisted wire inside to ensure flex and length of housing unlike PTFE tubes that can ' crush ' under stress. You may be able to find similar gauge wires as the string or at least close enough for your purposes.
With the encoder for leg length measurement: it's going to be slightly non-linear due to the changing diameter of the bobbin/spindle, do you think it would need correction? If so, a simple lookup table could work well.
Haha I remember a question in high school math on this exact topic! Something about a toilet roll being unwound and dealing with the changing radius. I would be very interested to know how big of an effect that would have; my intuition says not much, but I've been wrong before 🤔
Ben you are a smart thinker.
You can use a badge reel that winds and unwinds really small wire/string to measure linear travel without the same issue of the radius changing that you have using that CF spring.
A commercially available sensor product that does this is a string potentiometer.
Great job mate, I really love cable mechanisms. Just as a fun fact, engines use half-bearings with different thicknesses to compensate the manufacturing variations between parts, your solution really made me remember that, they're also diferenciated by color. Again, great job!
Haha yep, colour differentiation helps so much. Got the idea from the adapters that came with the encoders to fit differently sized shafts. Such a simple solution!
Cheers for the kind words 😊
What if the spacers were eccentric so you could rotate them for a perfect fit? Might not work with a 3d printed part. Seems like your solution with a few sizes works great already though 🙂
That's a very interesting idea! One potential issue I can see with that is that it'd make straightening the extending rod a little tricky. I didn't explicitly mention this in the video but I've been keeping each "set" of spacers (ie. top three or bottom three) the same size so that the central rod is kept centred. Having the spacers be eccentric might make this a little difficult.
Very cool idea though and I'll keep that in mind if I need to do anything similar in the future. Cheers!
@@harrisonlow ooh, very good point about wanting to put the same spacers around so it stays centered!
Random ideas:
Get a CF tube that the PTFE tube fits in, put it in there and now its stiff. Use as one kf the trio of stiffening buddies.
take 3 of the stabilization bearings, and put them on the internal end of the big moving CF rod, instead of bearing in onto the central rod, bearing out onto the trio of kuter rods. Would increase the weight of the reciprocating mass, but perhaps will reduce wobble?
IDK, thats all i got so far. Great work. I really hope to see this go somewhere awesome!
I like the idea of bearings on the main push-rod! I hadn't thought of that before! Might be tricky for a few reasons:
1) While the CF tubes are stiff, having the central tube pushing against the outer tubes (via the bearing) might cause the outer tubes to flex a little
2) I think you've hit the nail on the head with respect to excess weight: if the actuator's moving at > 2m/s, I want to keep that as light as possible.
Cheers for the suggestions, and I hope to see some more of these ideas in future designs!
what a great design, thank you for sharing. i'd go for cheap brake cables with a teflon liner, that way you get rid of the backlash and its essentially friction free.
Bicycle brake cables have excellent compression strength while being somewhat flexible and light! Also, they're relatively cheap. They are usually lined with some softer plastic but they are designed to house steel cable under high tension. may be worth a check, but not sure if the diameter of your aramid string is small enough to encourage cutting into the plastic liner of those cable housings.
Came here to say this.
Wanted to give a Hard Congratulations for hitting this milestone through your Project. 👏👏👏
Looking forward to see where it goes
Cheers!
Really nice job inventing such a fast-moving linear actuator system! I learned a lot of cool techniques, especially your ingenious way to handle the critical alignment requirements of the 3D-printed end parts with the custom spacers!
This looks very nice and I can't wait to see this running.
@11:57 - is exactly why kiting shops don't sell these kevlar lines over here any more. These lines cut fingers too so we are now allowed to only used kevlar with a braided cover.
@James Petersen suggested the use of bicycle brake cables but this is the wrong type to use. Outer brake cables can be compressed. You could try to use the outer gear cables instead, a lot of them even already have a PTFE liner inside to reduce friction. But do go out to a bike shop, tell them why you need these and check yourself if these are flexible enough for your use. Original Shimano cables are most likely way too stiff.
Cheers for the suggestions! That's interesting re. kevlar strings. Thanks for sharing that!
Thankfully I'm pretty sure I have both brake and shifter sheathes lying around that came with cables for my bike. I'm planning on making a testing rig to test a bunch of the ideas that everyone has suggested, and I'll absolutely be trying shifter sheathes.
Cheers!
Ive just just read every suggestion and thought of the most mentioned one's. String in support legs, use of bike cable sheathing, different shaped bearings. But didn't see any suggestion of just running a rubber pully wheel at the top of the 3 CF legs that snugs up against the CF tube with a tension adjuster.
So your motor would spin the rubber wheel instead
Would save string going to top and bottom of main CF leg.
Obviously there could be slippage, just a thought.
Great idea thou.
Im subscribing.............
Now.
Thanks for sharing 👍
For the bearing that has the string running under. you could use one that has an inside radius. I have seen a few for larger rods where the inner radius is designed to match a metal rods curvature. You could use one of these to allow the string to travel through the bearing without coming into contact with it - also it would allow two contact points instead of 1 theoretically increasing the surface area of the bearing against the CF rod reducing wear.
YES! I've been thinking about this but couldn't find any bearings that fit that description. Do you happen to know what they're called? I tried "bearing inside radius" and a few variants but couldn't find anything. I suppose I could also use a smaller bearing and print an adapter with a groove for the string, though I wonder if the string would cut through it over time 🤔Only one way to find out!
I think your comment was removed by UA-cam but I can still see it in my notifications. I'll check out U groove and V groove bearings. They seem like exactly what I've been looking for! Thanks heaps for the suggestion!!
I've been doing some digging and thought you might be interested to know about a cool bearing profile that I found: "gothic groove" profile. Could work to hold the CF tube more solidly *and* have the string between the tube and the bearing in the small space in the profile. I might have to pick some up to test 🤔
@@harrisonlow Just did a quick google. They definitely look good. I also found some luck with searching "track roller cam follower"
I'm a CS guy, not an electro one. The linear actuator I saw seemed difficult to use, so I made a simple "linear actuator" for a kinetic sculpture with a stepper motor, a string, a tube and a spring. Nothing compared to what you did, but this made me realise of what potential just a simple motor + a wire can have, you definitely extracted the full power of this mechanism :D
Haha the potential isn't exhausted yet! I've been working on a improved version that is better in many ways. Will hopefully finish collecting data and putting the video together soon 😁
Great build. Just found your channel so excited to see what is next.
In addition to the hoover method for string, I have used a can of compressed air to blow the string through. This works well on small strings in tubes too small to use the suction method. (Saved our butts at a robotics competition when rubber tubing began to stretch- we blew string through and tied it tight and it got us through.)
Great idea! Never heard of that and I've got a can of compressed air on hand. Cheers!
The ptfe tube buckling could be remedied with something like bicycle shift cable.
The bending looks pretty high energy! Thats wild! Impressive machine youve built. Odrive looks amazing
Great video and walkthrough! Well done. I'm so impressed by your design. I have thing for string pulley systems.
.. so you like puppets ?
@@boonjabby 😁 only those that have pulley systems... Not just strings!
Wow, that's an amazing mechanism! It's a bit like a hydraulic actuator, in some sense, because they typically (?) apply force through high pressure to one side of the piston at a time, rather than using pressure and vacuum on one side.
Very nice solution! To make it more long-lasting, I strongly suggest to put an outer liner of hard PU on the bearings with a slight groove. Rolling friction will of course increase a little but the contact pressure and noise will decrease radically. Another alternative is to use really thin-walled stainless tubes with carbon inside for strength and stiffness and the bearings as is.
Sweet man. I watch Skyentific all the time and just now I get your video recommended.
These would work perfectly in my world-domination robots! Subscribed!
Really impressive performance! The bearings might be larger than needed.
Id recommend looking into either spectra or dyneema line as its EXTREMELY strong, zero stretch, abrasion resistance is off the charts and its very affordable. As for the play in the bowden tube, you could probably route the ptfe tube through the carbon fiber tubes or... You could use some rigid brake/fuel line or copper condenser coil tubing to route the ptfe through which would remove any loss of motion there resulting in a more accurate and faster action.
For the cables, you can reduce buckling if you pair the string with a tube whose inner diameter more closely matches the string diameter. That could mean a thinner tube, but it could also just mean thicker string: for instance, type 95 paracord has an outer diameter of 1.75mm, which would be a pretty good match for the tube you're using.
Yup, I 100% agree. I actually have some very thin PTFE tubes on hand but I don't have any way to fix them to the printed parts on each end, so I just opted for the larger ones. Now that I think about it, I might be able to put the smaller tube INSIDE the bigger tube 🤔
Thicker string is a very interesting idea that hadn't crossed my mind! I'll pick some up and give it a test!
Cheers for the suggestions!
You could use car break lines. Basically the hydraulic equivalent to those pneumatic tubes and pass throughs. They can be bent easily and you should be able to find threaded connectors for the plastic parts
Yeah definitely I think other peoples of said bicycle cable and maybe some Kevlar string would work really well. very neat.
Silicone socks for the OD of the bearing, you can put a radius on it increasing the contact area, and you can put a recess for the string
This is awesome! I recommend bike shifting cable lubricated with silicone spray, its thinner and has a lot less resistance than bike brake cable. The housing is reenforced to prevent buckling and you can easily refill it with silicone if it starts to stick
How would it hold up long term? When used in bikes those cables don’t need to move an awful lot compared to what they would in this application.
Mountain Bike gear change cable outers would be perfect to replace your string guide pipe. Brake cables are just wound steel outer, but the gear change cables are not spiral bound and have a teflon/ptfe liner - they are designed for both push and pull actions with low friction.
Awesome design! I keep thinking of all the possibilities.
Awesome stuff! Your robot is mesmerizing to watch!
You had my subscription at "I'm building a juggling robot".
tldr: both ptfe tubes can probably terminate at the bottom part. thin brass/copper pipe for redirection.
down-stroke
the down-stroke string can pull the tube towards the bottom directly, instead of pulling on the topmost part. This allows more weight to be shifted to the less moving part and reduces ptfe tube length. It would require a redirection bushing or similar in the bottom part, but as that moves less i think it might be advantageous.
up-stroke
the up-stroke string can run alongside one of the structural carbon fiber tubes to the top part, where it gets redirected back down. this reduces ptfe tubing and probably tube compression issues.
redirection solutions
for redirecting the string, i would evaluate whether a small bent brass tube which you feed the string through wouldn't be enough. all this would require is a section on the printed part to glue this into (so that it doesnt fall out while assembling/maintaining, while in use the string tensions should keep it in place on its own) which might be easier for assembly and even lighter.
another crazy idea: it might be possible to reduce this design to a pair of telescoping pipes, by using a ha:f open linear bearing (like the LM12UU OP) at the top and running the string between the 2 carbon fiber pipes. though this would likely require a larger pipe for the outer part. This would probably be overengineering for this usecase tho.
For a cable transmission system - maybe look into bicycle brake cables or even the twist-throttle mechanism of a motorcycle. These already do what you are trying to accomplish at forces and levels or reliability magnitudes greater.
You should use capricorn bowden tubes. The blue ones. I found they're much stiffer and less easy to bend than the normal ones. Perfect for this application.
As you said at the end the PTFE tubes will 'wear out' over time. Have you thought about using brake cable covers/sleeves from bicycles? If you can find something that has a proper inner diameter. Those are lined with either composite or metal weave which makes them much more wear and longitudinal compression resistant
Totally forgot to say, this is a super cool project! Subscribed and exited to see more
Nice work! Inspiring!
I had issues with PTFE tubing buckling (from compression) when I used them as bowden cables. I've wanted to try the small coils (smaller than bicycle kind) but I never got around to it (and project had already met its goals).
Excellent work! Your solutions to inherently slightly sloppy FDM parts is very smart! Somebody get this guy a SLS printer ;)
Oh man I've been dreaming of having some better tools; SLS printer being pretty darn high on that list 🤤
Hydraulics have uses, but as more all electric fast acting and accurate systems are built going forward. Iinear actuators and servos will become even more prominent in new designs.
Nice job 👍
Really enjoyed this video.
I think the youtube algorithm may have removed my last comment due to it having links in it so I'll try again, without links.
A possible replacement for the PTFE tubes could be "armour spring" that is used to protect loop connections on fishing rigs. It is highly resistant to crush and compression forces and is very flexible and light.
There are two potential issues.
1. It is usually only available in packets of 10 x 10cm lengths.
2. It is usually green.
Regarding issue 1. I tried joining two lengths with a small section of heatshrink. This worked well. I set it up in a rig using 100lb dyneema and was able to lift 3Kg without any permanent deformation of the armour spring.
Regarding issue 2. I see that there is 1.2mm ID armour spring in the Shogun brand that is red.
You may have a tackle shop nearby where you can pick some up to play with.
It may even be possible to wind your own using suitable coated stainless wire.
Thanks for the kind words 😊
I've never heard of armour springs! They seem to be difficult to find good information about... Can't even find them on Amazon! (I don't have a tackle shop super close by). Do they go by other names?
As it happens, I actually have a few lengths of long, narrow springs on hand so I'll play around with them. Hopefully the "armour" part of "armour spring" isn't too critical 😉
Cheers for the idea!
@@harrisonlow maybe try googling "Wellsys ARMOUR GUARD".
I'd put in a link but UA-cam clobbered me last time.
Just gave this a quick test and it seems like a promising idea! Working prototype is a (~2.5 mm ID, 3.25 mm OD) spring that's about 50 cm long. The spring conveniently fits the PC4-M6 PTFE connectors that I already have, and it's working pretty well!
I'll give this idea a proper go some time in the next few days. My biggest concerns right now are: 1) friction between the string and the spring; and 2) sourcing the springs if I were to do this at-scale. I have these ones as a leftover from a uni project a few years ago and I don't think they're very easy to source in Australia...
Wow...brialliant!!! Thank you for sharing your inspirational work!!
That’s really amazing thank you so much for the video!
Have a look at synchromesh cables if you need more precision
This was very enjoyable, I really liked seeing the evolution of your designs, thanks for sharing.
Great video! Can't wait for the next one! I don't have any experience with cabling systems either, but I have used pneumatic tubing like this for other projects (something like McMaster 9446K21) and it's considerably stiffer than bowden tubing. It won't be as slippery a ptfe, though, so it's possible the tubing could melt at high enough speeds.
Thanks for the kind words 😊
Interesting! By stiff, do you mean in compression, or bending? Jugglebot needs fairly flexible tubes (in bending) to not restrict its range of motion. I WISH we had McMaster Carr in Australia. Honestly such a pain to source parts that are so easily available in the US 😭
@@harrisonlow In compression - since the walls are thicker, it's much less likely to buckle. It might not be quite as bendable as a bowden tube, but I think it would bend just fine even at the speeds you're running. They should be pretty common, lots of home air compressors use similar tubing for their hose.
@@tategeiger5317 Thanks for the suggestion - I will start the hunt for some pneumatic tubing!
I agree re. bending stiffness: it wouldn't need to be as flexible as the PTFE to be a non-issue for my purposes
Very cool reduction in moving mass! Fine Work
HarrisonLow, this is amazing education!!! Love it. Wish you were in the USA, so we can do an STARTUP... Wish you the best!!!
Had a similar problem at work, use 1/4 in copper tube and a pipe flare tube. Decently slippery and much much more robust
@12:08 My friend did this with a $250k capstan driven robot, but the line started slicing through the aluminium extrusions. Really impressive what thin line can do under tension.
You could make one of the bearings be in a tab that flexes a bit. You flex it out to fit the rod then the bearing will have tension against the rod. It's like making the bearing be pushed by a spring. I would recommend PC filament for this kind of thing as it retains its shape well when tensioned and is strong
I like the coil spring idea. You can integrate it into the bearing assembly.
Also, with a cable, it is possible to push and pull, though a return spring directly on the opposite end of the shaft can have accurate repeatability, less parts, and so on. Simply switch your ptfe tubes to 1mmx4mm or omething with a thicker wall.
I think, that if you'll get a wire going through a ptfe tube less flexible and with exact diameter of inner diameter of a tube, you won't have trouble with tube bending and collapsing. Great project!
Hi!!! You can use a V profile bearing, perhaps using TPU as a bearing cover. I think that will provide more stability (at least two contact points per bearing). This V profile can allow you to run the thread close to the apex of the V.
You could use PA(nylon) tubing instead of a ptfe, they're way more ridgid and are meant to withstand +100bar of hydraulic pressure (they're used for grease in industrial machines). You can keep the fittings as well since they're available in 4&6mm outer diameter
Man what a great design! gonna print one to test some ideas I have
Impressive. I need linear actuators that are both fast and strong. That means very expensive. 😢 I think I’ll try yours! 😊
Try an IGUS linear bearing/bushing instead of 6 bearings. You will need to preload the bushings perfectly to achieve the right tolerance that’s not too tight or too loose, but it should be a lot lighter for you.
Interesting! I previously tried a linear bearing instead of the 6 radial bearings, but the carbon fiber tubes that I'm using aren't specced for that level of tolerance; one manufacturers tubes had a loose fit, another didn't fit at all. I didn't know you could get bearings/bushings that you could preload. I'll look into them! Cheers!
How to get rid of the bowden tubes? Why not use a flex shaft to relocate the drum from the motor shaft to the base of the actuator? This would also save space by allowing you to attach the pull cable to the bottom of the actuator shaft. The pull side of the drum would be centered under the shaft. The push cable would still need to run up to the pully and back down to the bottom of the shaft. No more strings running to the top.
How to remove the backlash? Why not just add a spring to your newly relocated drum mount and turn it into a tensioner similar to a serpentine belt tensioner?
A TEMO flex shaft is like $15.
Interesting! I'm not sure I 100% understand what you're suggesting, but I have a few thoughts related to this:
1) I'm interested by the idea of flex shafts / push-pull cables (are these the same things?). I've seen people suggest them on the Wintergatan project but haven't had the time to look into them much. Are they particularly flexible?
2) For my application (Stewart Platform) it's somewhat useful to have the motor be physically separate from the actuator itself, as the actuator will be moving around and the motor would get in the way and just be more mass to move around.
3) An unintended boon to the way I have the strings routed now is that the end caps on the central tube are only ever pulled in towards the tube. This is rather nice as it means that they'll never get pulled off if the string pulls too fast/hard, since they're only connected via a not-super-strong interference fit. I could of course make that fit better, but that would be more effort!
Re. backlash removal: Yes! I agree with your idea of the spring, and this is something that I will absolutely be adding if backlash becomes an issue for whatever reason.
Cheers for the ideas!
Another suggestion for your string issue - try having a chat with Otis Elevators, they use a flat rubberised material in their elevator shafts instead of cables, they may have a few ideas that you can experiment with.
You may remove tubes completly, just move the motor to one of the end of actuator. Also , for brushless motor, you can use the body of motor as the base for spool - that way you have ability to center the motor relative to the plane along the actuator. Event more, you can use closed loop string, or even chain.
You only need the extension string. The retraction string can be taken off and you can use the constant force spring to retract the actuator. That way there is no string visible on the extended actuator which is the weak link in terms of ability to snap whilst being pinched by external factors. Cool project!
It's funny you mention that; that's actually how the last design worked! Check out this video if you're interested: ua-cam.com/video/h-th1lcGFq0/v-deo.html
@@harrisonlow oh I am gonna check it right now!
Bike shifter cable could be good, maybe try reinforcing the ptfe tubes with a metal spring on the outside, you could probably turn one yourself. Nice video.
Actually I was thinking along the bike cable line myself. Maybe he could run bike cables thru the ptfe that's in the actuator but tie the string to the cable at the motor end. So kind of a string and cable hybrid. Also for the cable routing issue, he should change the angle of the one fitting location so that it goes all the way up to the top and sleeve ptfe thru ,having the cable exit on the top edge instead of exiting down inside next to the pully center.
The fact that this kevlar-string version of actuation is much more effective than pneumatic or stepper is a strong defense of how evolution selected muscles+tendons for biological actuation
thats a neat actuator!
one problem i can think of is the off axis load created by the cables pulling on one side of the rod. this might not be an issue with light loads, but with higher forces along its extending & retracting axis it would require increasingly high lateral stiffness between the bearing assembly and the rod.
as for the ptfe lines. maybe running them through the existing carbon fiber rods could help. it shouldnt interfere too much with the movement of the base hinge as long as you run it directly through the pivot axis of each hinge. if that is too complicated / would require too much precision, you could still shorten the length of the ptfe tubes significantly by only running them from the base of the carbon rods to the motors.
btw if you really want to lighten all the moving parts as much as possible, theres always the option of a cable driven parallel mechanism!
Awesome work, love the progression!
With the bearings, you're running the hardened steel outers against (relatively) soft carbon fibre tube, which will have a bit of diameter variation along its length. The contact area between the bearing and tube is pretty small, so will exert a very high pressure. You'd be better to increase this contact area by using smaller bearings and adding a plastic sleeve that fits the tube diameter. You can buy these as track rollers, but you might not be able to find one for the exact diameter of tube you have. The sleeves could be 3D printed then turned true on a lathe. If you haven't got access to a lathe, you could use a drill with a mandrel and a piece of sand paper wrapped around a rod. The plastic sleeve will give a bit of compliance too so you can preload the bearings without risking damaging the CF tubes.
Look forward to seeing the progress!
Big thanks! The "turning true" with a drill and sandpaper is a great idea!
You can feed a thread through a tube with compressed air. That method is often used in the textile industry.
You should consider a capstan style of transmission to the rope.
Also, if you want to ba able to desassemble the tubes from the plastic, you can make a full cylinder, drill it in the center and screw in a slightly oversized screw to make it press fit
Very nicely structured and well explained video! Keep up the good work!
Edit: i really like the simplicity of the design with the range of motion it offers. Thinking outside the box work here!
Cheers!
Check out IGUS linear bearings made from plastic to reduce weight. Or shaft guides. They also make square ones. And also square carbon fiber tubes exist. Maybe some telescopic thing can be built with these to get the absolute lowest weight.
putting 2 pully strings and parallel to the actuator piston would decrease bending.. 14:16 very sweet design!!!
Haha it's funny you mention that - the earlier design (the third one in the "Design Iterations" chapter) had space for not just two, but *three* pulleys for the exact reason you're describing. Long story short, the string is close enough to the central tube and the bearings do a good enough job of containing the bend that it isn't an issue (that I've observed, at least), and routing extra strings + making sure they're the *exact* same length is very tricky/tedious.
Thanks!
fair enough bra
You could use a bike break housing as tube. They can bend, but they will resist compression very well.
Excellent design. Thanks very much for sharing!
the almighty algorithm has blessed this man
Brilliant job. The string which pulls down could be come from the bottom (part standing on the table). This saves half the boden strings connected to the slider part. Also no string going up along the center rod when it extends. So at the same spot were you connect the string pulling up you hook the string going down. As this is a straight line to the bottom part there is no cover needed (plain string should do).
Wow, amazing project! Thanks for share it
Try printing TPU sleeves for the bearings with a concave shape to exactly match the piston cylinder. Maybe you can get rid of more wobble by squishing the rollers in since they can now flex a little, but that would add a ton of friction so there's a tradeoff.
I like the idea of TPU sleeves! I hadn't thought of that! I'll see if I have enough smaller bearings to give it a shot. Cheers!
There is a vacuum tube version of your tubes with concentric ridges along the length to prevent collapse, still very flexible and lightweight