Building a Robot Link with full metal Cycloidal Transmission and BLDC Motor - First Test 

Thanks! That’s cool the hear. It’s definitely quite a bit of work, but totaly worth it - I wish you a good build 💪🏼

Thank you! Nice to hear, keep it up. It’s quite a long way to get there, but the things you can learn by doing it are totally worth it!

I answered you as you might have noticed :) Thanks, I‘ll try upload some more ;)

A cascaded control scheme with an additional driver was used, since the focus of this test was more on the transmission. I‘m currently building a full metal 6DoF robot arm, I‘ll not yet use this transmission though. There will be a video about the driver and control of the robot in the next few month.

I wrote a script in MATLAB that generates my profile based on some parameters. Then I exported this to CAD a designed the rest. The profile really has to match and getting the right pressure angle is also important. Even when the designed profile is “perfect“ the manufacturing tolerances are the most difficult thing. With 3D printing, there will always be a bit of variation in output speed, due to the limitation in accuracy.

Danke! Das Spiel variiert leider über die Umdrehung. Größtenteils wenigen Winkelminuten, in manchen Bereichen der Umdrehung können es jedoch bis zu 0,5 Grad (30 Winkelminuten) sein. Das ist es auch was größtenteils den die hörbaren Vibrationen im Lauf ausmacht - unterschiedliche interne Kräfte.. Dies ist auf die begrenzt präzise Fertigung mit meiner Fräsmaschine oder gar auf das Fräsen selbst zurückzuführen, da das Verfahren und natürlich auch meine Maschine keine hinreichende Formgenauigkeit des Werkstücks herstellen kann. Bei einem „full metal“ Aufbau sind die Anforderungen daran jedoch sehr hoch, um eine gute Funktion des Getriebes zu erlauben.

Or should it be done on an EDM machine and then sanded/polished?

Yes, I’m going to build a new one for my six axis robot though. I’ll try to make a video about it.

Hey, the disc is made from aluminum bronze, due to the good friction coefficient paired with steel. The main body is made of aluminum, mainly because of weight and manufacturability. The pins that are in contact with the discs are actually hardened steel

@@EngineeringSpareTime Thanks!

Danke dir! Wenn die Toleranzen nur nicht so kritisch wären. Ich habe den Prototypen nur durch Läppen der gefrästen cyclo dics zum Laufen bekommen 😄 mit geschliffenen cyclo dics sollte das nicht mehr notwendig sein, dazu muss ich aber erstmal mein setup upgraden. Materialtechnisch gilt es dort auf meiner Seite auch noch etwas zu optimieren - hier sind die Cyclo Discs aus Beryllium Bronze

Thanks for your comment. Yes, they are definitely suitable and scalable. The cycloidal transmission is actually used in many industrial robot application for a long period of time. The great advantage of them is their large overload capacity. Building such a transmission with a reasonable amount of backlash or preload at its best, definitely takes some effort. With my approach back then, just milling the cycloidal discs, it was not precise enough. It is possible though

@@EngineeringSpareTime What manufacturing process do you think would be needed to achieve the necessary level of precision? Milling the cycloidal discs to almost tolerance and grinding them to final spec?

@@TheFabricator03 exactly, grinding is the way to go here. There are not to many videos on RU-vid of the actual manufacturing process between all of those diy approaches. Here is on example: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-8WFtYFmAVps.html In terms of precision: wire edm would be suitable as well, not optimal in terms of surface finish though. Ans with grinding you can manufacture, as shown in the vid, multiple almost identical parts in one setup. This is also beneficial, since in most cases multiple discs are used in on transmission And just btw: the counterpart “housing” is not easy too :)

@@EngineeringSpareTime What are your thoughts on grinding of the counterpart housing? Small drill like grinder? I would imagine the most important part would be the location of the dowels that rub against the cycloidal discs?

If dowel pins are used the important part is more the orientation and positioning - precise machine and process..

The precision of my CNC might be different to yours but your version seems to run very smooth and it would definitely be a good starting point

You‘re right, I should have balanced the volume of those segments better :D the calm vibe was my actual intention :) Thanks!

I could make a video about it - good idea :)

@@EngineeringSpareTime it would be a great video for many 👍, think about it

This comment is a bit late, but I’d like to see that as well! I’m currently trying to cnc mill a cycloidal gearbox, but I am having a bit of trouble finding good resources.

I subscribe under the request :)

Thanks! The Materials: Aluminium, bronze, dowl pins, bearings and screws cost about about 70 euro. The motor, encoder and PCB cost about 300 (the encoder is the most of it..)

@@EngineeringSpareTime waw 70 euros is a low price for a reducer.. What is the reduction ratio of this prototype? ... I wonder if it is possible the estimate the maximum torque it can handle...

@@bilelsk9000 It’s the price for the raw materials - without my time, machine time... that‘s what makes it “cheap“. The transmission ratio is 35:1. Peak torque of the motor is 0,8 Nm. With a approximated efficiency of 0,8 -> 35 * 0,8 * 0,8 = 22,4 Nm

Sure.. This was quite helpful to understand the geometrical dependencies: www.geogebra.org/m/Zh5MJAVU It’s a step by step description of a graphical solution for one position, this has to be redone around the circumference - interpolation. Other then that I used books as a source of knowledge, since in e.g. RU-vid there is quite some misunderstanding going on - there are good videos as well, you can’t be sure which one is correct without the understanding of it.. not great for learning. In mechanics it often relates to “motion study“ or “motion technology“ to name some searching dot points. Most of the stuff I have used is in German, probably not too helpful then sorry :)

@@EngineeringSpareTime thanks a lot mate... It would be great if you could point to some good content on youtube that you referred... Or if i could read the research paper you might have published after your research

@@parthpatel1530 Searching for it, there are actually not too many :D This one might be helpful: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-guvatctnjww.html I haven’t released a research paper since I just did this for my own seek of interest :)

It’s not problems, it‘s difficulties ;) To get a proper reliable working unit, the shape of the cycloidal discs has to be as precise as possible. Otherwise you‘ll have a significant amount of vibrations/ friction/ wear due to the inhomogeneous force vectors in the contact areas of the disc with the outer rolling/gliding elements (e.g. leads to very high load on the disc and eccentric shaft). An ideal/ perfect machined shape of the disc would provide contact to all elements at any time - different force directions obviously.. Getting it precise is just difficult to manufacture (this is one things that makes them expensive). The well known 3D printed versions running kinda smooth (some of them) due to the lower stiffness of the material, with the cost of overall weakness, stiffness, wear, capability... The 3D printed versions are still a good way for low cost robotics. Definitely not for applications where precise movements are needed ... at least not for now :)

@@EngineeringSpareTime thanks for the info! That makes sense. Kinda like a grain of sand in the precision gearing that would cause wear. I saw this comment after the skyentific, so I'll post this comment here for your viewers too: @Engineering Spare Time Wow, that is precision! Is the wearing out of non-uniform backlash actuators something you observed empirically, or is there info on this subject commonly available (I didn't study ME only EE)? Would adding SEA fix the wear? So spur gears naturally flex uniformly to reduce wear. Would a more flexible inner cycloidal plate suffice? It might end up with the same backlash of a spur gear system, but still be compact? Or maybe instead of a single cycloidal inner plate, maybe a stack of 3 thinner plates would wear evenly and take up the "slack" being able to flex a little more.