Dual Linked Two stage Cycloidal Gearbox 3D Printed NEMA17

272:1 Reduction Ratio
First stage 16:1 and Second stage 17:1
www.thingiverse.com/thing:4709490
gouldpa00​
hackaday.io/paulgould
Based on Previous Designs
kzbin.info/www/bejne/mpW6lJ-qqt1mi68
kzbin.info/www/bejne/m6jShqaMgquKnLc
Music (KZbin)
Clover - Vibe Mountain

[@gordon6029]

Very cool Paul.

[@swannschilling474]

Awesome!! :)

[@PhG1961]

Nice !

[@BHARGAV_GAJJAR]

Curious when you push on the output arm does the motor shaft back drives at all ?

[@olafberend8837]

Great progress so far! Do you know the reason for unsteady movement of hypocycloid gears? Perhaps the design has some flaws or imprecise rules? Or is it due to imperfect 3D printing tolerances?

[@PaulGouldRobotics]

I don't know yet. Others have had this problem as well.

[@Keri-Kerigan]

It sounds like there's 2 in-phase tork ripples harmonizing causing a surge of positive and negative tork over baseline constant. Stepper motors, as well as servos, have a tork ripple in operation, ie, it's not a constant rotational force is a wave. Cyclodeoid drives can have one as well stemming from pin engagement disengage cycles also forming a wave. If they're not matched to negate each other, they could amply each other. Servo motors would be less prone to this over steppers. You could try, using a 5 phase stepper, or servo as well to help mitigate the ripple as they have much more even tork ripple due to their design. 2 pole would be the worst. Depending on your controller, you may be able to adjust the timing's on the servo or stepper to mitigate tork ripple somewhat. Additionally, if you were to find the harmonics of your drive, and ensure your cyclodeoid drive operates in an inverse phase, you could likely get rid of it. And I'm an Idiot.. I was half right. It is a tork ripple, but for a different reason. 1 rpm of motor = 1 tooth movement. looking at the footage, only 2/3 of 1 motor rotation are providing full engagement power. In whatever controller you are using for your stepper or servo, have it reduce power at the beginning of the disengagement stroke, and bring it back on the engagement. If you can do this as a sinewave ( some drivers let you do this ) so that the power is gradually increased and decreased all the better.

[@olafberend8837]

@@Keri-Kerigan If ripple in torque of the driving motor is the cause of the non uniform output (rotational speed) then this should be proportional to the gear ratio? That doesn't seem to me in this video. Your theory can be consistant if you think about an amplification of non uniform movement of a first stage due to engagement / disengagement of teeth that are superposed of the second stages non uniformity? Anyhow, there must be some non linear effects in the law of speed reduction? Or have I misunderstud your apprach of explaination? When I regard a cyclodeoid drive, isn't there allways a contact zone (in a single stage) with full engagement that continously is moving forward in radial direction?

[@i-make-robots]

You used the circlips!

[@PaulGouldRobotics]

I will use any good idea that improves the design. Cheers

[@DerSolinski]

What's your stepper driver? It doesn`t sound cheap 🤣 I like this "reduced gearbox". It is a good option for a lot of torque not bound to "just robotics".

[@PaulGouldRobotics]

kl-4030 but any stepper driver for a 3D printer will do.

[@aritopoko3325]

great design. good job. what kind of material the 3D print? PLA?

[@PaulGouldRobotics]

Thanks. It is PLA but it would be better with fancy filament

[@aritopoko3325]

any suggestion for a stronger and tougher material other than PLA?

[@PaulGouldRobotics]

I’ve used nylon PA66 at work but mainly pla and abs for robotics. Materials is not really my skill area.

[@chenhan91]

do you have any idea about noliniear motion? i had tried 2 two stage cycloid gearbox and it is also noliniear. but one stage is fine.

[@PaulGouldRobotics]

I'm not sure where it is coming from. I will have to do some research.

[@martylawson1638]

My best guess is that the non-linear motion comes from small printing errors. Since the final ratio is the difference between the first and second stage ratios, the output moves much slower while any non-linearity in the gears stay the same size. Recently experimented with elliptical bearings and strain-wave gearing and my 2015:1 model suffered badly from non-linear motion while a 31:1 variation was much smoother. www.thingiverse.com/thing:4855353

[@marvingriener3533]

I applied your design for my own gearbox with first stage 15:1 and second stage 15:1. But somehow the output just moves back and forth a little. The tolerances for the 3d printed parts are quite good and everything fits nicely. Do you know if there is somehow a problem with two identical reductions?

[@PaulGouldRobotics]

The two stages need to have a different ratio.

[@marvingriener3533]

@@PaulGouldRobotics Thanks, I realized my mistake...

[@PaulGouldRobotics]

@@marvingriener3533 I don't know where to find the information you need. I've never seen an actual equation for a dual stage. You can chat to me on Instagram, Hackaday or Discord. I would like to see your design.

[@zinojoe8974]

This structure seems wrong! It must have shown 1stage power itself, could be confused

[@konstsh2240]

this type of kinematic scheme of planetary drive has a very poor efficiency for large ratios.

[@PaulGouldRobotics]

I would be happy if it could achieve 50%. It is plastic with many sliding contacts.

[@konstsh2240]

@@PaulGouldRobotics the problen is not in material or tooth profile itself, it's in the first place related to this kinematic sheme of planetary gear

[@8797lilly]

@@konstsh2240 Can you elaborate on this?

[@heartminer5487]

yeah please elaborate

[@konstsh2240]

@@heartminer5487 In short - in layouts like this with "differential" reduction principle the tooth relative sliding velocity becomes comparable to its absolute velocity in contact point and thus frictional power loss rises significantly. This happens because when ratio increses the eccentricity decreases and the average distance between contact point and instant velocity center decreases also. So in terms of efficiency on high ratios the classical multistage planetary transmission is more preferable. Though it is not so "cool".

[@urielalbertodiazreynoso6309]

There are simpler way to make a two stage cycloidal gearbox.

[@PaulGouldRobotics]

Please link to other simpler designs. I would like to try them.

[@urielalbertodiazreynoso6309]

@@PaulGouldRobotics ok let me see.....

[@urielalbertodiazreynoso6309]

@@PaulGouldRobotics This guy has the idea kzbin.info/www/bejne/r3y3c4KcpqqMf5Y . But is awfully constructed. I've made some of them. You only need 5 pieces. The case, the double gear, the excentric bolt and the second cap attached to the output shaft. The case has the N1+1 teeth. The double gear has N1 and N2 teeth. The second case has N2+1 teeth. Think about it and try it. If I have some time I would upload a video talking about it.

[@PaulGouldRobotics]

@@urielalbertodiazreynoso6309 That is the basic of the dual stage gearbox. It has many problems which I am trying to solved with this design. My design is statically balanced and the forces are balanced around the housing. There is very little vibration at high speed.

[@urielalbertodiazreynoso6309]

@@PaulGouldRobotics You can balance it with some metallic load in the opposite side. Also you don't need bearings. For a 3D printer project I think simpler is better. If you are designing some industrial/commercial prototype maybe you are right. There are too many pieces for my taste :)

[@funtilltheend4607]

Hi dad I do watch you video