Is it possible to replace Robotis actuators by cheaper ones?

Technology
21

Yep we just have to finish the new mechanics of Poppy so we can fork it for a smaller one compatible with AX-18/12.
If we fork now, we will have to redo all the work a second time to update it. With CAD design, there is not yet versioning and merge tools as efficient as the ones we have for software development.

I will release the new structure ASAP so external contributors can help to create a cheaper version.

2 Likes
22

One Super Like form me :slight_smile: Please let us know when it is available !!!

Regards
NicoX

23

The main differences between AX-18 and MX-28 are :

  • The price :smile:
  • the size and weight
  • the torque 20kg.cm for AX-18 and 30kg.cm for MX-28
  • and most of all AX-18 cannot be compliant (there is always a resistive torque)

Then it depends on how you want to play with the robot.

I work with AX-18 (I already have lots of these) to see if it is possible to have a good gait and gesture management.

24

Hello there!

@Matthieu About the replacement of the two MX-64 motorsā€¦ Is it necessary -maybe- for the control algorithm to be these two motors that ā€œstrongā€? If their replacement would have stall torque of around 40Kg-cm (instead of 60Kg-cm) would this affect the control that much?

Regards,
Kimi

25

What kind of motors do you have in mind ?

26

High torque servos with metal gears. I am looking to the known RC webstores but I canā€™t find any similar with that high stall torque.

27

I did a quick search for servos:

The RC servo are indeed rarely exceeding 30kg/cm, the strongest I found is here.

During my search I found that Turnigy did a rought copy of the dynamixel motors (see: 1, 2, and 3). We know nothing about these servos yet: what is the communication protocol? is the mounting frame the same as the dynamixel? And they also look pretty weak, but for sure they are cheaper.

Finally just for fun, if you want to go heavy duty: 92kg/cm and 380kg/cm!

I recommend none of these.

28

If you just want to have a standing or sitting Poppy, you do not need to have high torque. You can freely ajust the design to your needs.

They look to have the same frame so it should be compatible with the current Poppyā€™s frames. For the control, it would require to write another pypot IO controller but with pypot 2 it will be quite straightforward :wink:

Personnaly, I would be interested if someone tries.

29

I should have read more, the Turnigy motors I mentioned are compatible with Robotis Dynamixel communication protocols.

We are highly interested by any feedback on these servos.

The full spec for this motor that cost 45$:

Features:

  • High torque
  • Double end installation, suitable for installation in robot joints
  • High precision double bearing
  • Good heat dissipation
  • Rotation scope: 0~300Ā°
  • Can be 360Ā° continuous rotation
  • Bus connection, theoretically, 254 units can be in series connection
  • 0.25Khz servo update rate
  • Compatible with Robotis Dynamixel communication protocols
  • Feedback of position, temperature, speed and voltage.

Specs:

  • Model: TGY-S518
  • Output: Single
  • Operating Voltage: 6.0 / 7.4V
  • Operating Speed: 0.25 Sec/60Ā° / 0.21 Sec/60Ā°
  • Stall Torque: 14.4kg.cm / 17.3kg.cm
  • Rotation scope: 0~300Ā°
  • Size: 35.5X36X51mm
  • Weight: 75g
  • Motor: Brushed
  • Ball Bearing: 2BB
  • Gear: Metal
  • Wire Length: 20cm
30

@jgrizou For these motors keep in mind these three factors:

  • Much less stall torque than MX-28
  • Different power supply (7.4V instead of 12V)
  • MX-28 resolution is 0.088 deg. TGY-S518 probably (on that price most likely ā€œdefinitelyā€) donā€™t use magnetic encoders which means much worse resolution.
31

@Thot
I do agree with your point. We can not simply compare MX with AX.
But if you look at Kondo it uses a 14kg.cm servo (even less than AX 18 18kg.cm) but still can produce good walking gaits. A tons of that listed in youtube :wink:

So I was wondering can we use AX 18 to have a smaller poppy? I was actually trying make it in budget :smile:
NicoX

32

Yes I know this robot. Here is an hypnotic video which shows it is possible with a condensated humanoĆÆd robot. It is then possible with a greater robot but smaller than Poppy. This is what I want to do too.

33

Hi @Thot
That is my inspiration too. Let me get into a bit deeperā€¦
I have a BIOLOID GP and seeing Kondo walking ( the vid you referred ) I thought if Kondo servos can then we also can with AX 18. May be the skeleton of joints we need to consider most. As in GP and in Kondo I can see there is a difference in servo placement in (mainly leg section). Then I saw Poppy and the its skeleton design is interesting, more humanly. So can we not design a smaller poppy that can do all the gaits that the full grown poppy can do and keep it in budget?

See AX 18 has lot of issue making gaits ā€¦ (problem comes basically from Knee/Ankle/ Heap-y servos). Again CPU bat will put more load on top so the bot will become more heavyā€¦more problems to come to AX 18. May be a bit expensive RX can helpā€¦
NicoX

34

Here is a quick summary of the alternative motors, we discussed so far:

Already available to buy:

In development:

I will try to update this post as new information come in.

35

http://www.thingiverse.com/thing:476030/#files is a link to a thingiverse design creating a dynamixel case for standard servos. Iā€™m not as interested in the complete control of the poppy robot as of yet but am very appreciative of the design. My goal is to learn about programming and potion control of the robot and teach my kids who are 2 and 4 how things work. I stumbled accross the design on thingiverse from wanting to just print components that will mock up to make a statue of the poppy until i can afford the properish motors. Now however if this works my cheap hobby servos will make a hopefully funtional robot to play with and teach my kids about.

where there is a will there is a way hope this helps anyone else in my spot to bring this great project to more people who are setback by the costs associated with building one.

3 Likes
36

Thereā€™s also the MX-12W. Same strength as the AX, but probably a lot smoother:
http://www.trossenrobotics.com/dynamixel-mx-12w-robot-actuator

1 Like
37

Someone has already played with this servo ?

38

I did not find the Max Torque for MX-12W. We can get finer control with respect to AX 18A/12 but not sure about the torque.

39

I tested them today. These servo are totally compliant when torque is off :slight_smile:

1 Like
40

@Thot
Do you find the Stall or working torque for MX -12W ?
Actually I did not find it in Robotis site. I might miss though :smile: :smile:

One quick note: I think MX-12W is made basically for wheel mode, may not be for joint mode.
We can consider it is a direct replacement of AX-12W.
See Robotics video https://www.youtube.com/watch?v=Tw4GfqUpzNA
and a thread http://forums.trossenrobotics.com/showthread.php?6515-MX-12W

ā€” NicoX