I am thinking of using the Dynamixel AX12 to build a phoenix-ish robot!
It seems to be far superior to the Hitec HS645-MG with way higher torque.
The AX12 also has temperature- and load feedback enabling it to sense obstacles directly.
Am i wrong?
The AX12 seems to good to be true…
With a linux EEEPC for brains it would make for a great distraction … study and job will suffer.
[size=75]ough the 1-day post prohibition after registering was killing me …[/size]
Lynxmotion has been primarily a Hitec supporter. I have committed to making some AX-12 brackets, but it’s going to be next year.
they are both good products however you need to consider the whole system in choosing which to apply. the ax12 runs at a higher voltage than the 645 and were you to run it at the same voltage the torque numbers would be similar but its speed would be somewhat slower. the ax12 uses a multipoint serial interface which requires a specific controller interface while the 645 uses a standard r/c servo pulse which has a much broader support base. the serial interface allows more information to be pulled from the ax12 but the 645 is simpler to use. again looking at plus and minus to each involves examining what you are going to use for a controller, what your programming proficiency level is, are you an off-the-shelf consumer or a deep-digging DIY type person, and how badly do you subscribe to the immediate gratification need impulse. it is not impossible to combine both products in the same project but it requires a degree of thinking it out before breaking out the credit card if you don’t want to waste money and time. 
Before spending significant $$$ on the AX12 servos, it might be wise to purchase one and experiment with it to verify it will meet your needs. Using the AX12 servos appear to be somewhat more involved than using standard servos.
Hi,
I have the bioloid comp kit, and the first I did was to test its torque. And I must say I was very disapointed. The holding torque was great, around 16 kg. But the running/load torque was not good at all, I measured between 6 - 7,5 kg. The Hitec 645MG was even better! 
A picture of my test bench:
This picture demonstrate almost the maximum of running torque. I also used a digital fish scale to test the torque.
I found this article that confirmed my findings about the AX-12.
After this test I didn’t find the AX-12 that fantastic anymore.
I’m still interested in the AX12 despite it having bad running torque. Eventually will order a bioloid starters kit to experiment with. In the meantime i shall experiment on its locomotion solution, more on that later.
I’ve spend two days on google … anyone know what the HS 645MG running torque is? Some experimental setup like Zenta offered would be fine.
To calculate a running torque: (knowing the weight is 0,5KG)
Zenta, how would you guestimate the length of servo-axis to counterweight-axis?
A fishscale comparison with equal arm length would be fine as well.
Agree with you there, you just have to be aware of the running vs. hold torque.
I think I remember doing similar test with the 645 MG, about 8 kg/cm I think… 
Not 100% sure if I remember correctly though.
12 cm.
I calculate in Metric, international notation. 13,887 ozf/in = ~1 kgf/cm
For USA/UK: also note the comma.
Advertised torque:Hitec 645 MG[list]7,7Kgf/cm at 4.8V
9,6Kgf/cm at 6VDynamixel AX-1216,5kg/cm at 10V
12kg/cm at 7V[/list:u]Running torque:
I intended to calculate by hand but this calculator could save me a screwups.
I leave above as a note to myself, its bedtime.
The max torque for my AX-12 testbench with a .5 kg counterweight at 12 cm is 0.5 kg x 12 cm = 6 kg.cm
Might i say, thanks to everyone (& in special Zenta) on the thread! 
Well TBH that’s what i also reckoned, but i found the difference on advertised torque so big i didn’t believe the outcome…
so:
Running torque guestimates:
Hitec 645 MG[list]~ 8kg/cmDynamixel AX-12~ 6kg/cm[/list:u]Findings:
Oh my … servo’s are weak considering their advertised torque. 
AX12: 25% less torque than HS 645MG.
I guess -for me- it all depends on
]the AX12’s sensor ability now, and whether my software can use it wisely./:m]]the servo being capable of higher resolution (1024 per 120degrees perhaps?)/:m]Seeing above, my outcome could be using a Phoenix setup with sensor feet instead.
Hi,
I believe the resolution for AX-12 is 1024 (10 bit) per 300 deg.
Good luck with your choise for servos!
heh, well marketing types always round up… sometimes more than just the fractions even. Don’t take the Hitec specs as verbatim either if you are in a position to test them… not saying they are wrong, just pointing out that they have marketing types too. Seriously though the Hitec stuff gets pretty beat on around here in many different assemblies so if they were really far off from their rated specs we would probably know by now, but if you are setup in a position to test them it would be cool to see the results anyway.
Personally I still love the bioloid ankle assembly they are able to create using AX-12s and their brackets. I have not figured out a way to replicate this, especially the low profile aspect, using standard servos and ots brackets from lynxmotion (yet).
For future reference:
Only after one week of delving into the tech of my idea, i found out someone already made a shot at it: AX12 hexapod by evilMachinaIt seems to be strong enough to support the longer Phoenix legs. drooldance
Glad to see my idea’s hold up: By coincidence (genius yadayada) i came up with a similar setup: C#, some x86 platform, AX12 and Phoenix legs.
I will focus on the software first, seeing the rest will work out anyway.
Will have to learn C# first, then make a cheap prototype of the locomotion model i have in my head using XNA.
AX-12 legs may soon be available:
crustcrawler.com/motors/AX12 … hp?prod=68
they are called “AX-12+ Smart Bracket System (ASBS)”
Alan KM6VV
Hmmm, a servo bracket system. Very inventive, wish I would have thought of that.
Sue 'em! 
I knew you’d be impressed. heh heh heh.
I thought you’d be interested. Don’t worry too much. They’ve only got 3 or 4 parts, and they are all for the AX-12 servos. Expensive!
Robot Savvy
Has servo brackets for it’s Bioloid (KHR-1 kit), that look nice, but that’s an entire biped kit.
Alan KM6VV
I’ve never worried about them…
now thats funny. LOL.
