Servos Built Specifically for Robotics

Martin_H

I've been using RC servos for several years now and built many robots out of them, If you accept the limitations they're a fun and cheap solution. But I've recently noticed servos that are purpose built for robotics and I have used any of them.

There's the Dynamixel servos which have monstrously high torque, on-board controllers, and a $40 price tag. I suspect these have been around a while, but I haven't seen to many projects using them as I suspect the costs add up quickly. Has anyone tried a project with these? Are they less twitchy than RC servos? How is their accuracy?

There are also servos that look like regular servos missing the standard mounting tabs, but have pan tilt style brackets built-in. They look like this:

http://www.ebay.com/itm/F04072-HDKJ-S3309D-60G-Torque-9kg-Metal-gear-Digital-servo-holder-mount-Bracket-/121101642066?pt=LH_DefaultDomain_0&hash=item1c32386952

Has anyone built anything with that sort of servo?

Duane Degn

I haven't used them and they're currently out of stock but HobbyKing has a robot servo.

It includes a swivel point for the bottom of the servo. They also sell one without the mounting holes.

I've used the Dynamixel servos some. They are nice and less twitchy (if you send non-twitchy commands).

I'm in the middle of programming a Prop to use a magnetic encoder as feedback for a continuous rotation HXT12K servo. The magnetic encoder has already been safely installed in the servo. Now it's software time.

I'm hoping the magnetic encoders will turn the cheap metal gear servos into something resembling (in control) of a Dyanamixel servo (Dynamixel servos have more going for them than just their position feedback though).

Edit: If I desire an additional pivot point on my servos, I use Lego turntables as the pivot point. I purchased a bunch from BrinkLink a while back.

Mike G

Hi Martin_H, I have a fair amount of experience with Dynamixel servos. Even wrote a 1/2 duplex driver for the prop a while back. Anyway, I like the servos.

I also had the opportunity to work with Robotis, the Dynamixel manufacturer. They're a great team and very customer focused.

Martin_H

Duane, I saw your thread on a new position encoder. It will be interesting to see where it goes.

Mike G, thanks for posting the video as that gives me a good idea as to their repeatability. The lines look much better than what I've done with analog servos.

erco

Cool video, Mike G! I made a simple BS2 printer last year using standard servos giving a unique polar coordinate perspective:

Martin_H

erco wrote: »

Cool video, Mike G! I made a simple BS2 printer last year using standard servos giving a unique polar coordinate perspective:

Erco, something I find interesting about your bot is that you attached the servo further up on the arm rather than at the joint. This sacrifices some of the arm's range of motion, but increases mechanical advantage. On most robot arms you rarely if ever need a full 180 degrees to shoulder and elbow servos, and these joints are precisely the ones under the greatest load. So it's interesting that all servo based robot arms generally mount at the joint.

While throw is decreased shouldn't that also decrease positional errors and vibration introduced by servo twitchyness*?

Here's the problem. As soon as I build one robot arm, I suddenly realize all the limitations it possesses. For example industrial robot arms have two wrist rotation axis, one before the wrist angle, and one after it. This allows for complex compound wrist movements useful for rotating a nut onto a bolt for example. Except for ST Robotics rather expensive arm, I haven't seen to many hobby robot arms with this feature. Also, while servo repeatability is pretty good, it's only accurate to a few millimeters and there's a nasty oscillation that can happen when a servo overshoots, pulls back, and does it again.

A stepper based robot arms like the ST Robotics model solves many of these problems at the expensive of significantly increased complexity.So in some ways I'm hoping that better servos or better mounting of servos might address some of these problems in a future robot arm.

* Dead band coupled with position feedback drift due to temperature, phase of the moon, age of Aquarius, etc.