PDA

View Full Version : Mammoth Servo For A Haxapod



WinterT
01-11-2017, 04:10 PM
Hello, I'm new to this forum and Hexapods in general. I've been doing some research on the subject and I decided I'll go and build my own. I wanted something that could carry a bit of weight overall and in doing so I came across these servos http://www.topmodel.fr/en/product-detail-18825-topmodel-mammoth-38055-servo. They are so far the ones with the greatest amount of torque I've seen so far. Half the speed but it would be twice the size of a phantom so I guess that's ok. My question is are there and servos comparable or better in regards to specs/price?

On a side note my original idea was to build a machine for combat (similar to battle bots) so I was planning on using a mix of Grade 4 and Grade 5 Titanium as armor, hence the need to carry a bit of weight. But after really thinking about it I don't think any hexapod would stand a chance against some of the professionally made robots. I'm still gonna make it for me though so keeping the original idea seems like fun.

KevinO
01-11-2017, 07:34 PM
Wow that is a lot of torque and almost seven times the weight of a MX-106! My Golem Hex is twice the size of a phantomX and still only uses MX-64s. I'd love to see your calculations on how much your hex could carry.

WinterT
01-11-2017, 08:26 PM
Well if its 380kg-cm at 12V it should break down like this.
380kg(380,000g) at 1cm armature so...
38kg(38,000g) at 10cm armature
Presuming only 3 legs will be supporting the hexapod at any given time(worst case) with a 10cm armature I just multiply the result(38,000g) lets reduce that 3,000g to be safe so (35,000g) by 3
Giving me 105,000g(231lbs) that just the 3 servos connected to the chassis itself can lift. Lets subtract 10% to account for any deviance giving me 208lbs.

I think my math is correct but who knows. KevinO what is the armature length on your golem? Oh by the way your hexapod and youtube is one of the reasons I decided to make my own. Thanks for the inspiration!

jwatte
01-11-2017, 10:41 PM
First: Is that controlled with RC pulses? If so, you won't be getting any feedback back. That's annoying when you want to measure load, record poses, and such.

Second: It seems somewhat slow. It's about 100 degrees per second, which ends up at about 17 rpm. When walking, you may have joints that want more speed than that. Then again, most of the Dynamixel Pro servos have speed ranges in th 15-30 rpm.

Third: It's pretty heavy. The lowest-end Dynamixel Pro servo is 1/3 the weight, but has less rated torque too.
(When it comes to hobby-style servos, the rated torque, and the torque you can actually get out of it, are often very different. Meanwhile, for most professional motors, the recommendation is for continuous duty to be 1/4 to 1/5 of stall torque -- hobby servos are unlikely to do better than that, and many do worse.)

Fourth: 37 Nm of torque? That seems ... higher than I thought a gearbox could actually take when in the size of that servo. It's quite possible you'll break it before you stall it? And how much backlash is there in that gearbox? The specs don't tell.

Fifth: How far does it move? 90 degrees like a hobby servo? 180, 270, 300, 360 degrees? Some joints may need more than 90, for sure.

Sixth: It doesn't have dual horns, so it will have a single faying surfave towards brackets, which is significantly less table than designs with two horns (or idler horns.) You may be able to design an idler into a mounting solution of some sort to compensate.

Anyway, a Robotis servo with similar specs (but likely better communication and full 360 degree control) is the
http://www.trossenrobotics.com/dynamixel-pro-h54-100-s500-r-robot-actuator
Which is like $2,500 each ... If you get a machine part at 1/8 the cost, you'll have to ask yourself how that is possible.

My guess is that you can probably build a large-ish hexapod just fine with those servos, but that you should de-rate the rated torque by a large factor to avoid overheating the servo. Also, you have to design it with whatever the movement and protocol limits are for the servo, which are not specified -- to me, it's a warning sign that there are no specifications, no torque curve, no data sheet, not even mentinoing what kind of gearbox it's using.


Finally: I believe you're right that battle bots are very purpose-specific. The mechanics involved in walking add a lot of complexity and fragility and weight that's there just for locomotion, rather than for battle or armor. So I don't think it would be competitive. But it would likely be, as we say around the bay, "hella cool!"