go_away

Author Topic: Servo Speed  (Read 1722 times)

0 Members and 1 Guest are viewing this topic.

Offline Robotboy86Topic starter

  • Robot Overlord
  • ****
  • Posts: 139
  • Helpful? 0
Servo Speed
« on: November 11, 2007, 06:03:09 PM »
I have noticed that *most* hexapods/quadrapods are extremely slow to move and react.  Is this just because of the spped of servos?  Beacuse the seem somewhat fast to me..  spec wise at last.  60 degrees in .18 seconds or some such nonsense.  That would be about 180 in a bit over half a second.  But watching most movies its MUCH slower..

I assume because of the legs weight and such.

Is there any solution to this?  I was thinking maybe a constant magnet attached to a leg.. and then two strong electro magnets on either side of it.  So when one was hit on, it would throw the constant magnet away.. thus causing leg to move.

But I have no idea..

Offline HDL_CinC_Dragon

  • Supreme Robot
  • *****
  • Posts: 1,261
  • Helpful? 5
Re: Servo Speed
« Reply #1 on: November 11, 2007, 07:11:03 PM »
The reason they go slower is because of the weight of the legs and such that its under a higher load so they usually need to be geared slower to get more torque out of them so they can actually move. If they were used without any kind of gear ratio, they would simply just not move and probably die within a few minutes from the stress.

I dont think EMFs would work for moving legs like that. the power to weight ratio is just too far off I think. However, they do use giant electromagnets on certain trains to elevated the train off the track so its basically floating in the air. Im not sure how they propel the trains forward but im assuming they generate a higher same-pole EMF at the back of the train to push it away... not sure though, that part is only speculation on my part.
United States Marine Corps
Infantry
Returns to society: 2014JAN11

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,666
  • Helpful? 169
    • Society of Robots
Re: Servo Speed
« Reply #2 on: November 11, 2007, 07:21:20 PM »
So the servo speed you list is a servo without counter torque. A servos speed slows as more force is applied against it.

As a mechanical engineer, I will tell you there are three reasons that they are slow:
A) servos are used that are too weak (because hexapods require a lot of servos, people get the cheap kind)
B) poor mechanical design - position of the servos, length of the legs, addition of springs, etc could really help
C) unoptimized programming

Mechanical design is something even a beginner can guess with fairly accurate intuition . . . but you can make your design significantly better if you plan it out and crunch the numbers. I spend weeks in CAD design, run stress analysis simulations, calculate torque and applied forces, determine expected losses, dynamics, etc . . . perhaps overdesign I dunno . . . in basic diff drive robots you do not need to do this, but for hexapods and robot arms its almost required to get your robot to work as planned.

By calculating forces, you will know if your solution will work/fail without ever building it.

And to motivate you to study up - picture spending a wasted $100 for a motor that turns out not to have the torque you expected. Calculating can save you money! :P

One day I'll write a tutorial for all of this . . . for now you have this:
http://www.societyofrobots.com/mechanicsbasics.shtml

Offline Robotboy86Topic starter

  • Robot Overlord
  • ****
  • Posts: 139
  • Helpful? 0
Re: Servo Speed
« Reply #3 on: November 12, 2007, 08:49:36 AM »
so lets say the legs weighed about a quarter pound..  and were roughly a foot long from the base of the robot(from servo to tip) I would need at least a small amount of .25lbs/ft of torque to move it.

But is there a way to calculate how much speed it will loose?

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,666
  • Helpful? 169
    • Society of Robots
Re: Servo Speed
« Reply #4 on: November 12, 2007, 11:03:12 AM »
Quote
is there a way to calculate how much speed it will loose?
not really . . .
if you look at the servo datasheet it will list speeds for two different torques (or voltages). and of course a 3rd datapoint would be stall torque at zero speed. if you graph those three datapoints (torque vs speed), draw a trendline between them in excel, and interpolate to the torque you expect, then you can estimate what the reduced speed would be.

Quote
so lets say the legs weighed about a quarter pound..
what you really want is the body weight divided by the number of legs lifting (3?).

 


Get Your Ad Here

data_list