Society of Robots - Robot Forum

Mechanics and Construction => Mechanics and Construction => Topic started by: toomanyvalves on January 14, 2011, 04:33:23 PM

Title: reliable servos - really reliable servos!
Post by: toomanyvalves on January 14, 2011, 04:33:23 PM
Hi, newb here.
I have a lab automation project that would be perfect for some of the newer robot servos (dynamixels, uptech cds5500), but I don't know if they will last long enough.
The motion itself won't stress the servos much - pretty much like turning a knob on an oven: (~1 kg cm breakaway torque (friction), 300 degree range). I won't need much speed or acceleration, and it will be a very light duty cycle.  But over the course of months to years some of the servos will be looking at 100,000 operations ...
From what I've read, even if the gears last, the film potentiometers on the dynamixels, uptech, and most hobby servos start to wear out after a while, probably in the 10's of thousands of turns range. Stepping up to a dynamixel with a magnetic encoder is $500, about 5x my budget per servo.
There are plenty of digital servos for the RC hobby market, some with magnetic encoders and plenty of reliable sounding components, but no lifetime information on them that I can find.

Steppers would work, but that means more boards, more programming ....

Anyone know of a source for cheap industrial-grade servos(<$100, 60+oz in torque, 270+ degree range), or should I plan on using steppers instead?

Thanks!


Title: Re: reliable servos - really reliable servos!
Post by: photomark on January 14, 2011, 06:10:49 PM
I have heard of the dynamixels and from what I can tell you will not find a cheap alternative to them , you want ultra reliability and to get that you will have to pay for it .

There maybe some used industrial servos available that will fit your need but you will have to do plenty of hunting to find them and being used they would hardly be ultra reliable would they .

Steppers may be the best for you and your budget
Title: Re: reliable servos - really reliable servos!
Post by: Admin on January 15, 2011, 10:25:54 AM
The Dynamixels should be fine.

Just buy one, and program it to do 100,000 operations with a typical load attached. The experiment will probably last a full day, so just check on it occasionally.

If it lasts, then you know it's good enough to spend time on for your actual project.


With the cheap hobby servos, you'd need to test a handful of the same servo type: some last just ~3 hours while others might last up to 30+ hours. Basically, poor quality control in manufacturing. I'm not sure if you might experience the same with the Dynamixels, but maybe someone has tested them already?
Title: Re: reliable servos - really reliable servos!
Post by: toomanyvalves on January 15, 2011, 02:35:46 PM
Thanks folks. I think I will pick up one servo and try it out.  The ability to detect the servo or its load beginning to go out of spec before they actually make the machine fail (expensive and corrosive chemicals going places they really shouldn't: do not want) without having to add more sensors myself is really appealing.

Has anyone had experience with the RX-10 ($99, metal gears) lasting longer than the ax-12+($45, plastic gears) or the cds5500 ($35, metal gears)? 
One other question: if I'm already running a light duty cycle and programming to limit the speed, current,  torque, etc, would there be any further advantage to running at a lower voltage? I'm hoping to use the 12V and 5V rails of a PC power supply for this project, but stepping down to 9V might be worthwhile.

Cheers!
Title: Re: reliable servos - really reliable servos!
Post by: Admin on January 15, 2011, 02:45:51 PM
Quote
would there be any further advantage to running at a lower voltage? I'm hoping to use the 12V and 5V rails of a PC power supply for this project, but stepping down to 9V might be worthwhile.
The three points of failure are:
- gears wearing out
- motor coil/brushes wearing out
- your control electronics wearing out

Voltage affects torque (of gears), heat and sparks (of coil/brushes), and limits of control electronics. So lower voltage is always a good idea if lower torque/speed isn't a concern.

But adding in a switching regulator to drop the voltage could also be a point of failure, so another thing you should test first . . . I'd say keep it simple and do your first test at 12V and see what happens.