Our Hackerspace is starting a new robot arm project. Have any improvements been made in the last 3 years that we should be aware of before we begin? Thanks!
I noticed after placing the top gear onto the shaft, that it wobbles and doesn't stay flat. Once I add the output housing, it eventually pops off. Have you experienced this before?
Cool, thanks for sharing and making it open source! What's the anti backlash part here? Is it just down to getting a good and slight interference fit? I remember one of your older videos where the outer gear was split into 3 parts that were tensioned in opposite directions with a cable to remove backlash
thanks for sharing your work. I looked at the thingiverse list but I cannot find the size and length of the dowel pin and the size of the small bearing for installed inside the gear. Do you mind providing that? thanks!
Hello Paul ! Why did you replace the multi star brushless motor by a nema 17 ? This design is still compatible with a brushless motor ? :) I'm following your work since many years now and I love your designs ! thank you so much for sharing!
I think timing/capstan drives for NEMA17 are better in the following ways:- Lighter, a lot cheaper, zero backlash, higher efficiency, higher ratios and linear/smooth. Tensioning is a bit hard for both the timing belt and capstan cables. I've only been playing with capstans for a few months but I feel they are overall better for NEMA17. I'm still doing tests on Brushless motors.
Hi there ! thanks a lot for your work and those involved in this design, but I would like to ask why you didn't post the version you are showing in this video on thingiverse, I would like to have it constrained with nut and screw like in the video, but the file like that is not on thingiverse, and I would not want to drill a hole to make a mess, I can see that you replaced one of the steel rods with a long screw. Do you think the design on thingiverse is better ? thanks again.
If I want a 10:1 reduction instead of a 25:1 what more is it than a different number of "teeth" of the eccentrically driven yellow gears and the black housing as well as a different eccentricity?
Once the design stabilizes, its may be possible to create a parametric CAD model with the required design rules to generate a model with a specific ratio. But this would be specific to a particular CAD modeler.
@@PaulGouldRobotics Great! Do you have it optimized for different ratios, with all geometry associated to the number of teeth? Chances are formulas will need to be embedded in the model and proper associative rules would need to drive resultant geometry such as the eccentricity of the gears and housing. These days its much easier to apply these rules after the design is finalized. Modern parametric CAD systems make this a lot easier today than when I first became involved with it in the 1990's (ComputerVision's CADDS5 and PTC's Pro-Engineer). Back then, once the final design was created, we would design a completely new solid bases on the geometry of the final, being sure to create all the construction geometry first. This is because the early parametric CAD systems were 'history' based. But these are things that shouldn't be addressed until later. You are doing a remarkable job implementing design improvements. Building in (and testing) a CAD model to generate a "family of parts" (ie kick out different ratio designs reliably) is a whole different domain.
Cycloidal gearboxes can achieve a higher reduction ratio. The motor can be inline with the gearbox. But I'll be moving to timing belt and capstan for most of my future work.
@@PaulGouldRobotics I havent actually printed any reducers. But I can imagine that the fluctuations of how thick the printed line is makes printing accurate walls for the cycloidal gear very frustrating. And by walls I mean the functional "gear part" walls of it. Is that the case or am I overthinking? If it is, then moving forward with using belts or capstan is indeed the better choice. Do you plan on ever working on a harmonic drive that uses belts? I think that the combination of belt on the flexure and 3d printed teeth on the outer ring would be affected much less by any printing discrepancies compared to how it affects the smoothness of a 3d pritned cyclodial gearbox. Because the slight elasticity of the belt teeth would allow it to conform much easier to printed teeth. Especially if there was sufficient pressure behind it. It could result in some very rigid and low backlash gearbox. What do you think? On another question, do you plan on integrating any simple metal parts on your capstan drives to increase their longevity? I am mostly talking about the shafts. Because 3d printed plastic and tensioned steel seems destined to cause problems in the long run. Cheers!
@@Mr30friends You are correct about the 3d printed thicknesses. My friend Simon make these harmonic drives. I'll continue developing capstans for a while. I don't want to use metal because I would like the design to be easily reproducible with a simple 3D printer. I will strength the capstan input drive with M3 bolts. Capstan's only have rolling resistance so if I spread out the load it should be fine long term.
It is an Absolute magnetic encoder. There is a diametric magnet connected to the Steppers shaft. The next step is to run SimpleFOC or some smart controller so you cannot use steps.
Cycloidal has a higher ratio so it is stronger. Dual belt has low backlash. I like this one better. th-cam.com/video/ENMZsPwCUcA/w-d-xo.html It has zero backlash, higher ratio and is stronger. It is also cheaper and lighter.
Why did you make this channel just for kids???? I can’t get alerts for your videos. You videos are full of such interesting information and contribution to makers/and hobby engineers. This is not child content!
I have selected the "No, its not made for kids" options box. I think you are not allowed to comment on kids videos. A lot of creators have had issues with notifications not being sent out. Subscribing will not get you alerts. Can you please check if the bell icon is selected on your end? If all is correct on your end, I'll ask TH-cam for help.
It is currently not good. I need to reprint some parts to reduce the tolerance. 3D printed cycloidals are never going to great for backlash. I have created an Anti-backlash cycloidal gearbox in a separate video.
@@muthukumarankothandaraman2371 Bearings make everything smoother. They are just expensive, increase part count and heavy. I have a few bearing only designs.
This exemplifies the great bits of open source hardware. Some share, some improve, many benefit.
Dann great, mate!
Cheers! Thanks to everyone for tributing!
these designs are getting really good, definitely going to give them a go when i get the printer up and running again.
Thanks for synthesizing all of the input and sharing the files!
As many times as I have watched the assembly, it still looks like magic when it works! Great design!
How much do all the bearings cost?
@@TheRainHarvester About US$1 for the small ones and US$5 for the big ones. Total ~$30
I printed your design some time ago and learned so much for doing so!! Thank you so much for sharing!!
I'm a simple man. I see cycloidal. I like.
so satisfying to watch the assembling part
thank you so much for your work on this
Our Hackerspace is starting a new robot arm project. Have any improvements been made in the last 3 years that we should be aware of before we begin? Thanks!
Just beautiful.
Great! work. It would be great if you share some details about the board you are using for the close loop setup.
My board will be months away. You can use S42B or check out SimpleFOC and hackaday.io/project/11224-mechaduino
cool!! congratulaitons paul! what is the backslash?
It is currently not good. I need to reprint some parts to reduce the tolerances.
I noticed after placing the top gear onto the shaft, that it wobbles and doesn't stay flat. Once I add the output housing, it eventually pops off. Have you experienced this before?
what type of screws did you use and is that a heat set piece that holds the outer cover together, thanks
Cool, thanks for sharing and making it open source! What's the anti backlash part here? Is it just down to getting a good and slight interference fit? I remember one of your older videos where the outer gear was split into 3 parts that were tensioned in opposite directions with a cable to remove backlash
There is no anti-backlash on this design. I have found that if I print the part with good tolerances the backlash is acceptable.
thanks for sharing your work. I looked at the thingiverse list but I cannot find the size and length of the dowel pin and the size of the small bearing for installed inside the gear. Do you mind providing that? thanks!
Thingiverse is update with the bearing (8x5x2.5) and rod (32x3)
@@PaulGouldRobotics thank you so much!
Had same question. Thanks Paul and Scott!
What is the approx weight of the assembly? NEMA17 motor + drive
Hello Paul !
Why did you replace the multi star brushless motor by a nema 17 ? This design is still compatible with a brushless motor ? :)
I'm following your work since many years now and I love your designs ! thank you so much for sharing!
Great Work Paul. I have seen that you have worked with both Cycloidal Gears and Capstan Drives. Which has given you a better performance?
I think timing/capstan drives for NEMA17 are better in the following ways:- Lighter, a lot cheaper, zero backlash, higher efficiency, higher ratios and linear/smooth. Tensioning is a bit hard for both the timing belt and capstan cables. I've only been playing with capstans for a few months but I feel they are overall better for NEMA17. I'm still doing tests on Brushless motors.
Hi there ! thanks a lot for your work and those involved in this design, but I would like to ask why you didn't post the version you are showing in this video on thingiverse, I would like to have it constrained with nut and screw like in the video, but the file like that is not on thingiverse, and I would not want to drill a hole to make a mess, I can see that you replaced one of the steel rods with a long screw. Do you think the design on thingiverse is better ? thanks again.
www.thingiverse.com/thing:4781349
If I want a 10:1 reduction instead of a 25:1 what more is it than a different number of "teeth" of the eccentrically driven yellow gears and the black housing as well as a different eccentricity?
Everything has to change. Smaller ratios require a larger eccentricity and larger outer pins also. 25:1 is almost idea for this diameter.
Once the design stabilizes, its may be possible to create a parametric CAD model with the required design rules to generate a model with a specific ratio. But this would be specific to a particular CAD modeler.
I use a parametric model for all cycloidal gearboxes. It is much quicker to determine the correct dimensions.
@@PaulGouldRobotics Great! Do you have it optimized for different ratios, with all geometry associated to the number of teeth? Chances are formulas will need to be embedded in the model and proper associative rules would need to drive resultant geometry such as the eccentricity of the gears and housing.
These days its much easier to apply these rules after the design is finalized.
Modern parametric CAD systems make this a lot easier today than when I first became involved with it in the 1990's (ComputerVision's CADDS5 and PTC's Pro-Engineer). Back then, once the final design was created, we would design a completely new solid bases on the geometry of the final, being sure to create all the construction geometry first. This is because the early parametric CAD systems were 'history' based.
But these are things that shouldn't be addressed until later. You are doing a remarkable job implementing design improvements. Building in (and testing) a CAD model to generate a "family of parts" (ie kick out different ratio designs reliably) is a whole different domain.
What is the size of the large bearings? It looks like 65x50x7 mm.
Correct size, they are 6810-2RS
Good day paul, I have clarification with the bearing 10x15x4 , the available here is 10x15x3, can I use it?
i mean, just change the design right?
Are there any benefits to this compared to the much simpler capstan drive you had shown?
Cycloidal gearboxes can achieve a higher reduction ratio. The motor can be inline with the gearbox. But I'll be moving to timing belt and capstan for most of my future work.
@@PaulGouldRobotics I havent actually printed any reducers. But I can imagine that the fluctuations of how thick the printed line is makes printing accurate walls for the cycloidal gear very frustrating. And by walls I mean the functional "gear part" walls of it.
Is that the case or am I overthinking?
If it is, then moving forward with using belts or capstan is indeed the better choice.
Do you plan on ever working on a harmonic drive that uses belts?
I think that the combination of belt on the flexure and 3d printed teeth on the outer ring would be affected much less by any printing discrepancies compared to how it affects the smoothness of a 3d pritned cyclodial gearbox. Because the slight elasticity of the belt teeth would allow it to conform much easier to printed teeth. Especially if there was sufficient pressure behind it. It could result in some very rigid and low backlash gearbox. What do you think?
On another question, do you plan on integrating any simple metal parts on your capstan drives to increase their longevity? I am mostly talking about the shafts. Because 3d printed plastic and tensioned steel seems destined to cause problems in the long run.
Cheers!
@@Mr30friends Once it is "dialed in", it is very repeatable.
@@Mr30friends You are correct about the 3d printed thicknesses. My friend Simon make these harmonic drives. I'll continue developing capstans for a while. I don't want to use metal because I would like the design to be easily reproducible with a simple 3D printer. I will strength the capstan input drive with M3 bolts. Capstan's only have rolling resistance so if I spread out the load it should be fine long term.
any chance of getting the bores for the ball bearings thanks
Check out the thingiverse page
thanks man its excelent
THANKS BRO!! There's no need to change in the 3d files? Just print it and asemmble then it works fine?
THANK YOU BRO , PLS REPLY.
It should be fine to just print. Every printer is slightly different. Tolerances are important to get low backlash.
@@PaulGouldRobotics Thank you so much for the reply
@@PaulGouldRobotics btw, I also want to print this th-cam.com/video/pizl7i-uB68/w-d-xo.html, its one of your video, there' no need to change?
@@creativebuilder577 The files on Thingiverse are the actual ones I use for printing my own prototypes.
@@PaulGouldRobotics Thank you! So much appreciated your video and your files
What is the PCB for the top? Some kind of position sensor?
It is an Absolute magnetic encoder. There is a diametric magnet connected to the Steppers shaft. The next step is to run SimpleFOC or some smart controller so you cannot use steps.
@@PaulGouldRobotics Where do you get your diamtric magnets from? They seem rather rare.
Sorry to be a nusance. This is all new and super interesting and I'm very grateful for all of your exploratory work!
@@nateashby5613 What country are you in?
@@PaulGouldRobotics The US
is this actuator better than the dual belt one you uploaded a few days ago?
Cycloidal has a higher ratio so it is stronger. Dual belt has low backlash.
I like this one better. th-cam.com/video/ENMZsPwCUcA/w-d-xo.html It has zero backlash, higher ratio and is stronger. It is also cheaper and lighter.
Why did you make this channel just for kids???? I can’t get alerts for your videos. You videos are full of such interesting information and contribution to makers/and hobby engineers. This is not child content!
I have selected the "No, its not made for kids" options box. I think you are not allowed to comment on kids videos. A lot of creators have had issues with notifications not being sent out. Subscribing will not get you alerts. Can you please check if the bell icon is selected on your end? If all is correct on your end, I'll ask TH-cam for help.
先顶再看
Do you have any link for cad file?
Thingiverse link in the description
@@PaulGouldRobotics thank you so much!
TEST BACKLASH, TEST TEST TEST...!!
It is currently not good. I need to reprint some parts to reduce the tolerance. 3D printed cycloidals are never going to great for backlash. I have created an Anti-backlash cycloidal gearbox in a separate video.
@@PaulGouldRobotics would, using bearings for the internal profile (black part in this assembly) smoothen it better?
@@muthukumarankothandaraman2371 Bearings make everything smoother. They are just expensive, increase part count and heavy. I have a few bearing only designs.
Which encoder is this?
AS5147 but I've also used AS5047 or MA702