PDA

View Full Version : [Question(s)] AX12+ how to turn off overload detection



ratto1234
06-18-2010, 05:14 PM
Hello to everyone,

I'm an italian software developer, I live in Verona, I was born in 1967 and my name is Andrea.
My hobby is robotics.

I'm reporting here a question I've asked in private to lnxfergy: and his reply down below:

QUESTION:
Dear lnxfergy,
Please I'd like to ask you a question

I've seen your post:
http://forums.trossenrobotics.com/showthread.php?t=4192

I'm trying AX12+ and I see that they have a running torque of only 8Kg*cm, after that it shuts down.

I've seen you were talking about turning off overload detection.
Do you mean clearing bit 5 of 0x12 (Alarm Shutdown register) ?
Does this way release more torque?

I've seen that robotis doc says it has an holding torque of 15Kg*cm @12V 1.4A
I measured current, but it sinks no more than 1A and then it shuts down.

Could you suggest me other cheap motors or servo motors suitable to build a little robotic arm (arm lenght is 15 cm and forearm length is 15cm )?

Thank you in advance for your help
Andrea

REPLY FROM lnxfergy:
Holding torque does not equal moving torque. The general consensus is that the AX-12 moving torque is about half the rated holding torque, which is about what you are seeing.

I would also strongly recommend that you do not turn off overload shutdown. It's there for a reason -- to preserve the geartrain. If you turn off the shutdown, expect to strip gears out

Unfortunately, torque costs money. The AX-12s are one of the better torque to dollars ratio servos available.

I'd really suggest you post future questions into the forum -- that way these responses stay around for everyone to benefit from! Seriously, we don't bite, and you've done your research, so you'd get plenty of good responses.

-Fergs

ratto1234
06-18-2010, 05:20 PM
Thank you lnxfergy

Please another question, just for testing I'd like to turn off overload shutdown, the method is to clear bit 5 of 0x12 (Alarm Shutdown register) ?

Will this way AX12 release more torque or will it just stop as before but without shutting down so that it will shutdown later for temperature?

Thank you in advance
Andrea

JonHylands
06-19-2010, 07:08 AM
Thank you lnxfergy

Please another question, just for testing I'd like to turn off overload shutdown, the method is to clear bit 5 of 0x12 (Alarm Shutdown register) ?

Will this way AX12 release more torque or will it just stop as before but without shutting down so that it will shutdown later for temperature?

Thank you in advance
Andrea

If you clear bit 5 of entry 0x12, the AX-12 will not stop when it overloads. You will either end up stripping the gears, or you will burn out the motor driver. It might still get too hot and switch off torque, so if you're bound and determined to destroy a servo to see where the limits are, make sure you clear bit 2 of that entry as well.

- Jon

ratto1234
06-19-2010, 07:56 AM
thank you Jon, lnxfergy was right, you don't bite ;)

of course I don't want to burn out the AX12.

I'm testing it and powering it with a laboratory power supply which monitors current.
During the test I'm observing the current drawn and after a few seconds I'd switch it off.

The fact is that on the robotis web site (http://www.robotis.com/zbxe/dynamixel_en) they say that at max torque (15Kg*cm) it should draw 1.5A @ 12V.

The problem is that AX12 stops moving when it draws just 1A, then after about 5 seconds it shuts down.

To explain better my test:
- I've connected an arm with a weight to the AX12 which produces 10.2Kg*cm when the arm is horizontal.
- I start with the arm down (vertical) and I command to rise tha arm 90 degrees (to get the arm horizontal).
- Before reaching the horizontal position AX12+ is no longer able to move (but it only sinks 1A, it does not try to sink 1.5A).
- I've monitored the "Present Load (0x28)" when it's stopped and it reads 1023, which means maximum torque while it's drawing just 1A, not 1.5A.

For this reason I suspect there is something strange maybe in the calibration or in the AX12 F/W.

I've tryed to power it with 13.9V (register 0xD Highest limit voltage is factory set to 14V), it releases a little more torque.

Do you know if there is the possibilit to change calibration parameters or AX12 F/W?

Thank you in advance
Andrea

lnxfergy
06-19-2010, 11:05 AM
Again, the issue here is moving vs. holding torque. The 15kg-cm rating is for holding torque, the moving torque has typically been seen to be less. There really isn't more torque "to release". Turning off the alarm shutdowns is a sure way to cook/break the servo.

-Fergs

ratto1234
06-20-2010, 11:11 AM
Thank you Fergs,

I understand that, but I was expecting to see the AX12 drawing 1.5A as rated before reaching "Present Load" = 1023 which is the maximum torque perceived by the electronics inside the AX12.
But I understand that it's not possible.

Actually indidicating the holding torque alone without telling the real running torque is just a trap for mechanics beginners like me. Robotis is not transparent regarding this.

I'll try other Servo or Motors or other ways, or may be I'll give up the arm project.
I'll let you know.

Thank you very much for your help.
Andrea

lnxfergy
06-20-2010, 11:16 AM
Actually indidicating the holding torque alone without telling the real running torque is just a trap for mechanics beginners like me. Robotis is not transparent regarding this.

This isn't just a Robotis practice -- it's basically all servo manufacturers. Torque ratings in the servo industry are all fairly wishy-washy, and are typically a bit trumped up denoting the holding torque just before the servo explodes....

EDIT: I doubt you'll find a $40-50 servo that outputs as much torque as the AX-12 and has similar reliability (yes, we can buy really cheap high torque servos from China, but they tend to die spontaneously or explode into flames). If you need more torque than a $40 servo can do, it quickly jumps into the $100 per servo, there aren't a lot of high torque servos in between....

-Fergs

ratto1234
06-20-2010, 11:55 AM
I'm looking at this now:
http://www.servocity.com/html/hs-805bb_mega_power.html
price is 40$
Stall Torque (6.0V): 343.01 oz/in. (24.7kg.cm)
But it's just analog, and so it does not have all the advantages of AX12.
Maybe I could substitute the electronics of the Servo with the open servo project one.
What do you think?

I'll look deeper into Servo offered on the market.

Andrea

billyzelsnack
06-20-2010, 12:25 PM
That's a mega servo and it is quite a bit larger and heavier than an AX-12.

http://farm4.static.flickr.com/3179/3014349059_401cf67d76.jpg

I also have a handy list of the more powerful servos here..
http://billyzelsnack-robotics.blogspot.com/2010/03/servo-torque-to-weight-ratios.html

ratto1234
06-20-2010, 01:03 PM
You're right, it's really big!

It's very interesting the list of servos in the link.

I'll study it.

Thank you again
Andrea

DresnerRobotics
06-20-2010, 09:04 PM
Might look into double back-to-back AX-12s to double up on torque and reduce backlash.

darkback2
06-21-2010, 01:21 AM
Ratto,

I hope this doesn't come across as catty or preachy, it is definately not my intent to sound that way, but I think maybe some advice that was given to me when I tried to build an arm.

Double up on the shoulder lift servos and/or use springs to counter the weight of the arm itself.

To that I would add, consider using a levered or geared system to maximize the amount of torque at the shoulder.

Check out the link below. You'll see they use 2 AX-12s for the shoulder lift joint, and one for the elbow. Probably can't lift a soda with it...For that I would consider using a minimum of RX-28s, and at that point RX-64s aren't that much more expensive :) Basically this can get really expensive really fast if you want it to.

http://www.crustcrawler.com/products/smartarm/index.php?prod=12

Hope this helps.

DB

ratto1234
06-22-2010, 12:54 PM
Thank you for the suggestions Tyberius and darkback2,

I've seen Lynx motion and crustcrawler SmartArm and both use two Servos for shoulder and SmartArm uses two Servos for the elbow too.
I've seen Lynx motion uses springs.

I'd like to build an arm with arm length 15cm and forearm length 15cm, wrist/gripper length about 10cm, with 6 degrees of freedom (i.e. full degrees of freedom) plus gripper motor and able to lift a soda can (which is about 330g), if it would lift 500g I'd like it more.

I know it's a challenging project.
I'll let you know my progress.

Andrea

darkback2
06-22-2010, 01:56 PM
I think your going to have to look at beefier servos than AX-12s then. Even with the springs. You have to factor in the weight of the servos as well as the arm. Doubling the number of servos at each joint increases the amount of weight at each joint. I know 15 CM isn't that long, but fully extended your talking about 40 cm. even at holding torque of 12000 g/cm your only talking about 600 g of holding torque at the base using 2 AX-12s. Given that moving torque is about half that your looking at an arm that is going to struggle to lift itself. (I never was very good at math, and only think for a second that I understand any of this...My math may be way off!)

The servos weigh 55g. So double that at the elbow = 110 * 15 cm...Couple that with two at the wrist, and one in the gripper at 30 cm and you see how it all adds up fast.

If you go the dynamixel route things get expensive pretty quick given the kind of strength your looking for...There is a handy Strength chart on the dynamixel servo page that might be of use for this project. You might also have better luck using standard hobby servos, but then you loose a lot of important feed back, like when you are about to loose a servo to heat/overload.

Good Luck.

DB

JonHylands
06-22-2010, 05:31 PM
Here's a video of an AX-12/RX-64 hybrid arm I designed for Brainbot:

YouTube- RX-64 Shoulder Test

The shoulder servo is an RX-64, and it handles the heavy lifting (full soda can) with ease. The other servos are all AX-12, but you will notice that their rotation axes are all vertical, so they contribute nothing to the ability of the RX-64 to lift the load, at least when everything is straight.

- Jon

darkback2
06-22-2010, 05:55 PM
Yeah...but thats a diet coke...Try lifting a coke classic...then we'll talk. :p

ratto1234
06-26-2010, 07:33 AM
I think your going to have to look at beefier servos than AX-12s ...
Doubling the number of servos at each joint ...
If you go the dynamixel route things get expensive...
You might also have better luck using standard hobby servos, but then you loose a lot of important feed back...


Thank you darkback2,
sorry for the delay in my reply, I was very busy with my job,
I'm looking at beefier servos, but not robotis ones, they cost too much for me, RX64 costs $280 for example.
Probably I'll give up with the project, I'll let you know.

-Andrea

ratto1234
06-26-2010, 07:45 AM
Here's a video of an AX-12/RX-64 hybrid arm I designed for Brainbot...


Thank you Jon,
sorry for the delay in my reply, I was very busy with my job,

the torque of RX-64 is amazing (along with its price ;))

I've seen in your pictures gallery you build a lot of robots, is it your main job or just an hobby?
Is brainbot one of your robots?

-Andrea

ratto1234
06-26-2010, 07:58 AM
I googled...

I've seen brainbot voodoo: YouTube- BrainBot - Voodoo
It's very impressive.

One question please, if I'm not wrong, I've seen that AX12 have only "proportional parameter":

28(0X1C)CW Compliance Slope
They does NOT have "integral parameter".
So I'm wondering how can they reach the target?
Are they accurate?

Does RX64 has the "integral parameter"?
How can brainbot move so fluid in the voodoo video?

Thank you in advance
-Andrea

zoomkat
06-26-2010, 10:27 AM
How can brainbot move so fluid in the voodoo video?


If one updates a standard servo quickly with small incremental changes, then the servo movement is fairly smooth.

YouTube- Robot Arm Remote Control Interface
YouTube- Controlling a robot arm naturally
YouTube- Servo motor modified as input device

jes1510
06-27-2010, 12:00 AM
Off topic but, Check out the tutorials section. I wrote a tutorial on how to modify a standard hobby servo to get positional feedback.