Anyone use the Dynamixel AX-12A with a Propeller? Tips?

WBA Consulting

Has anyone used Dynamixel AX-12A Servos (Robotic Actuators) with a Propeller? I just bartered into a set of 5 AX-12A servos and a handful of Biolid pieces that should enable making a rather nice robotic arm. These things are really nice but are $45 a piece! However, from what I have gathered so far from the data page, it looks as though I need to be willing to learn quite a bit to make use of them as they aren't just a servo that works with a single PWM signal.

I am only after a simple yet sturdy arm for my underhouse robot project that I started a year ago. My camera will be on it's own pan/tilt setup so the arm will only hold a paint marking pen, a spring loaded probe (to check for dry rot), and the bug spray tube/nozzle. So, I am thinking I don't really need the Hulk version of an arm that these Dynamixels would create. So...........

Should I dive in to how to use these things or save time and hassle and just sell these off to go with a basic arm with standard servos?

Dynamixel AX-12A web page and AX-12A data page

Roy Eltham

There was a thread a long time ago where someone did it. Can't find it right now. Found some: http://forums.parallax.com/discussion/99068/interfacing-the-propeller-to-a-bioloid-dynamixel-ax-12 The second post in this one has links to more threads.
The links in that post are broken, so here's another thread: http://forums.parallax.com/discussion/122369/dynamixel-ax12
And another: http://forums.parallax.com/discussion/153927/anyone-using-the-dynamixel-ax12a-servo

There is this in the OBEX: http://obex.parallax.com/object/152

Ken Gracey

I'm interested in knowing if anybody has used their continuous rotation servos. I'm not clear exactly how they work but I'd be curious if they have some internal feedback which allows the RPM to be set precisely. While they may not work for dead-reckoning, perhaps they can be matched from a control signal / speed perspective.

Anybody have any experience here?

Ken Gracey

GordonMcComb

Check out the posts by Xanadu, and maybe follow up directly with him (he's still on the forums when he's not flying his plane or quad). He had some Dynamixels he was finally able to get to work using some revised code Duane provided.

xanadu

No airplanes or quads today. Yeah right haha.

The issue I had were with MX servos running on the AX code. The AX worked perfectly with JM's code in this thread - http://forums.parallax.com/discussion/153927/anyone-using-the-dynamixel-ax12a-servo

If you are using the MX, the notes toward the bottom of that thread are the only modifications you would need.

I have an AX-12A, MX-28, the USB programmer, and would be happy to help anyway possible.

To answer the original question, the AX-12A would be a vast improvement over regular servos. Whether or not the cost is worth it, probably depends on what the arm is doing. What did you have in mind?

xanadu

Ken Gracey wrote: »

I'm interested in knowing if anybody has used their continuous rotation servos. I'm not clear exactly how they work but I'd be curious if they have some internal feedback which allows the RPM to be set precisely. While they may not work for dead-reckoning, perhaps they can be matched from a control signal / speed perspective.

Anybody have any experience here?

Ken Gracey

The AX12x supports continuous mode. You can set a goal RPM and it will do its best to maintain it. I would go with AX-12W, they make better wheel motors vs. joint actuators - http://www.trossenrobotics.com/p/ax-12w-dynamixel-robot-servo.aspx

I can measure the precision of the RPM with an AX-12A if you're interested, I don't own any AX-12W, though.

GordonMcComb

xanadu wrote: »

The issue I had were with MX servos running on the AX code. The AX worked perfectly with JM's code in this thread - http://forums.parallax.com/discussion/153927/anyone-using-the-dynamixel-ax12a-servo

Ah. I misremembered, then. I recall not being able to get either servo to work. But that was probably because you asked me for help. You'll know better next time!

xanadu

You're right, we couldn't get the AX or MX working that day. It had to be a loose connection. I remember not wanting to hack the included cable. I left everything right where it was, powered it up the next day and the AX worked fine.

It's worth mentioning, the connectors on the Dynamixels are not what you'd find on servos. This is the connector that plugs into the Dynamixel, http://www.mouser.com/ProductDetail/Molex/50-37-5033/?qs=AplfTeSvkkCfnVdKv8UuEg==

Paul K.

WBA yes they are worth it. Little smart servos that stop if you push them to hard. They don't strip gears. My old go to servo for robotics is the 645-mg by hitec. Theses still wear out over time. I don't use standard servos anymore just AX-12's with the exception of micro servos.

My rover5 with ax-12 and standard micro servos for the gripper. Went with standard as they are lighter and smaller and it was an old lynxmotion gripper I had on the shelf. Working software archive attached. Its for my controller which is XBee based. Build Link for the bot.

The project that I just started is an ax-12 prop based hex using a prop board i made. I recently got a 3d printer so I'm printing alot of it. My progress in spare time is the bottom plate and battery shroud. Started 2 weeks ago plan to move the legs this weekend and design a top plate. Im keeping the the bioloid brackets just printing my own tibias.

Take a look at the code it's an IK-Arm code for the configuration in the first pic. This work is 7 years old, updated for the pwm portion of the code. It works i'll get it working as I need to drop an xbee into and I'll post a video. Agave robotics started working prop ax-12 but they gone. I use Dan's code and have John's version but never tried it.

Put the Ax arm on your bot. Super strong and worth the effort. I have to dig thru some old back-ups to find the cad files on the plates for the arms.

WBA Consulting

xanadu wrote: »

..............The issue I had were with MX servos running on the AX code. The AX worked perfectly with JM's code in this thread - http://forums.parallax.com/discussion/153927/anyone-using-the-dynamixel-ax12a-servo
........................ Whether or not the cost is worth it, probably depends on what the arm is doing. What did you have in mind?

Xanadu, thanks, that was the thread that I settled upon as having the most information. I noticed the code there is updated from the OBEX release.

As for what the arm is doing, it will be holding the spray nozzle for a tube feed of bug spray, paint marking pen, and a spring loaded probe for checking for dry rot. The idea is something that looks like the "Inspection Robot with 24” Arm" picture as shown half way down on this page at InspectorBots. I am thinking maybe a 16" arm would be fine for my needs. This post on the thread I mentioned before has pictures of the bots that are the inspiration for the total setup.

Ray/Gordon, thanks for the other links and info. I do believe I need to look into them more before I make a decision. There isn't a lot of info on the forums.

Paul, the arm on your Rover5 looks along the lines of what I am after. If you can find the files for the arm links, I would like to see them.

xanadu

I see you mentioned that in your original post, not sure how I missed it.

I had something similar. I remember getting around 0.5" of repeat accuracy with the cheapest servos available. I think 90% of the wear and tear on my standard servo arm came from powering it on. The AX don't flop around when you power them up, another bonus for an arm.

If you took an average servo at $30, added the current protection, feedback and a power up circuit to stop it from flopping around you're at the $45 price for an AX. If you go with $10 servos it would work fine, just not as well, and less reliably.

The 6 pack for $200 puts the AX-12A at $33.33 each. I don't see how that can be beat with standard servos.

If you're getting them less than that on a barter I say go for it, you're in for a real treat if you're used to working with standard servos.

One weird thing about the AX, and it could be PEBKAC, is that when I send it to a goal position it will stop near that goal but not on it. For instance;

repeat until goal position = 500
 goal position = 500

The above will get stuck inside of the loop, even with an unloaded servo. However, I found this to work fine;

repeat until goal position > 498 and goal position < 502
 goal position = 500

Those couple of ticks translate to a tiny little bit of travel, nowhere near the slop in a standard servo.

If you ever plan on putting the camera on the arm, the AX would be the way to go. I think the standard servos would vibrate the camera. The AX can hold their position well without using a lot of current which will add some run time.

If you decide to go Robotis, and need the programmer, let me know I'll ship it over. I only needed it until I was familiar with everything, and especially for the MX programming troubleshooting. Anyway, I don't think I would use it again.

Here's Mr Floppy arm -

Roy Eltham

xanadu,
Regarding the issue you mention with the goal position. This is due to the compliance margin setting in the servos. The default is 1 in either direction, so the servo will stop motion when it is within 1 of the goal.
The "Punch" setting may also impact the stopping position relative to the goal position, depending on load.

xanadu

Roy Eltham wrote: »

xanadu,
Regarding the issue you mention with the goal position. This is due to the compliance margin setting in the servos. The default is 1 in either direction, so the servo will stop motion when it is within 1 of the goal.
The "Punch" setting may also impact the stopping position relative to the goal position, depending on load.

Thanks, makes sense now. I will check that out.

Duane Degn

I like the AX-12 servos but be aware these servos still use pots. It has been my experience all pots eventually wear out.
I don't know if they use different pots in models intended for use with wheels but I avoid using pots on joints getting a lot of movement.

BTW, I am using my phone to post this. My internet is down until probably Thursday. I am not likely to post to forum often (if at all) untill then. And yes it is driving me crazy!

Duane Degn

Sorry I don't have a link but I used AX-12 servos in my 3 wheeled Omni not. Speed control was OK but not perfect.
I think there is an Omni bot video in my "Fun with Spin" playlist.
IIRC I had to tweak the speed settings to get the robot to "spin straight".

WBA Consulting

Xanadu, thanks for the extra insight. Once I dig through the code a bit more, I will decide on what to do. Since I bartered for something that I had for years and no longer had a use for, I consider that they were free, so I am in the black no matter what. My thought for going with a basic arm was that I could use the money from selling these AX-12s to get an arm and some extras (and keep my project budget in the black). However, if I can digest the code well enough where I would feel comfortable implementing them, I think a stronger arm would be worth it.

Since I have 5 servos, a thought is that I could use 3 for the arm and 2 for the pan/tilt for the camera. I may have to make some mounts or brackets to make it completely work, but I have access to two different laser cutters and a CNC. (I'll be dangerous once I get up to speed in Fusion360)

Thanks for offering the programmer. The code I looked at appeared to have commands for reprogramming the IDs of the servos but not much else, so a programmer would be a major plus. I'll keep you posted....

Duane: is the 3 wheeled omni bot you mention this one?

WBA Consulting

Well time got the best of me and my belated spring cleaning has put my Dynamixel robot arm in the eBay pile. Figured I would update this thread with the auction link in case anyone was interested.

https://www.ebay.com/itm/123296775306