Building a Linear actuator ?

[Shubham]

Hi all,
I want to build a linear actuator which can (with some precision) have a displacement of around 70~80 cm ,the load on the linear actuator will be around 10~15 Kg(the load will be a rod about 2 mtrs long, the other end of which will be again on a similar actuator). Some other features that i would need are: the actuator should not have back driving characteristics i.e. it should be self locking. Also suppose the actuator is moving in the vertical direction(i.e. up-down) then there if my linear actuator is not moving then the other end of rod should be able to move up-down by a small amount i.e. if vertical is Z-axis then the rod should be able to make an angle ~20 degrees with the Z-axis.

So i was thinking (after looking on net/forum) that using a screw-nut(acme screw) arrangement seems  easy to implement and control ?
Since i don't know an iota of mechanical design except common-sense   so easy of construction is a major requirement!!!

If we use an acme screw with efficiency < 35% we can prevent back-driving. but about the 2nd requirement i am not sure. will the second requirement introduce some backlash in the gear?
Also i was initially thinking about using a geared DC motor with encoders , then what kind of precision can i expect ? or using maybe a stepper would be a good idea?
How do i calculate the torque needed for a motor to move such a load ?

My preference of screw-nut gear was due to my thinking that i can get this build easily in a turning shop ( that has a lathe and some drills and some other things.)  , since i am not from mechanical background , so i need to know about things like what kind of material would be suitable for screw and nut in my case and what possible things might go wrong
Also this whole linear actuator assembly will be quite long , so how to minimize the effects like vibrations of robot movement,bending of screw ?
How should i best support the long screw , means i need to attach it to the robot chassis  , so i need to attach the screw at top as well as bottom.


thanks......

[Admin]

Holy monkies that's a long displacement!

How about rack and pinion?
http://www.societyofrobots.com/mechanics_gears.shtml#rackpinion

the load on the linear actuator will be around 10~15 Kg(the load will be a rod about 2 mtrs long, the other end of which will be again on a similar actuator)

I'm not sure which way the load is being applied. Can you draw a diagram of your concept?

the actuator should not have back driving characteristics i.e. it should be self locking

Use a worm gear, they can't be back driven.

with encoders , then what kind of precision can i expect ?

Check the encoder datasheet.

this whole linear actuator assembly will be quite long , so how to minimize the effects like vibrations of robot movement,bending of screw ?

You're biggest problem would be it bending under it's own weight. You will have to do a stress analysis calculation on it (I can't guess it for you ).


Whenever I do a mechanical design, I start off by using CAD. Then I start doing math on it, modifying thickness and length of the components based on my calculations.

This might help you too:
http://www.societyofrobots.com/mechanics_statics.shtml

[Shubham]

i have no prejudice against rack and pinion gears  , but the only thing is that they should fuilfill my requirement and should get easily fabricated(a rather deciding factor).

yeh the CAD modeling part I will try out with Solidworks and simulation using Visual Nastran , but i am new to both of them , it really a pain designing all the components from scratch

Attached is a rough drawing of what i am trying to achieve , hope i have made it clear.

[ArcMan]

Assuming that you want to drive both screws at exactly the same speed, I would not use 2 motors.  Instead, use one larger motor with a sprocket and chain or toothed belt.  This will guarantee synchronization between the screws.

[Shubham]

@ArcMan
do you mean to say that there will be a motor (say in between of both the screws) and it will be rotating both the screws at the same speed .

But the screws are located on different robots and it is possible that one screw is stationary and the other moving up or down.


One more question is that is given in the first picture is what my robots will be looking like , so while moving on the incline if I make the central tower vertical as shown in second figure , will the robot be stable on slope ?

[Admin]

Shubham, your last post makes much more sense. I was confused what you were doing earlier . . .

Just some questions . . .
Why does it need to be flat?
What happens if the robot goes up the hill at an angle? It won't be level if the hill isn't a perfect slant . . .
Is your real goal just to keep the robot stable?

[airman00]

Useful
http://content.answers.com/main/content/wp/en/thumb/a/aa/250px-Linear_actuator_basic.gif

[Shubham]

@Admin , I have to keep the rod/load straight(or horizontal) by constraint .
and to make the robot stable is a necessary requirement for doing anything else. The hill will be quite smooth , but it would be great if i could make my design more robust.

@airman00 : thanks for the link , i guess this arrangement of having a "movable cylinder" instead of a nut might help is improving the mechanical robustness of the actuator. but i doubt if i could implement the actuator of my requirement(70-80 cm displacement) without making the actuator too high and making the robot more and more unstable.

[Admin]

Ok so your goal is to keep a platform flat on rough terrain . . .

Here is an idea:
make a 4 wheeled robot
at the center of the robot, have one single pole
at the top of the pole is your platform
at the base of the pole, have two servos to rotate the pole on the x and y axis

an accelerometer can tell if the platform is tilting

then it just becomes a 2 degree of freedom inverted pendulum problem

[Shubham]

"then it just becomes a 2 degree of freedom inverted pendulum problem"
great , we will be soon doing some matlab/visual nastran simulation of inverted pendulum problem soon as part of linear control system course.

But what about the actual construction of the linear actuator , i really need some advice from people who have previous experience in building things like these.

[Admin]

oops I misspoke - its a 1 DoF pendulum problem. There are two DoF, but both can be treated entirely independent of each other.

My *strong* advice would be to CAD it and look for parts at like mcmaster.com, etc. - that's exactly where I'd start. You might end up going through several ideas before you find one that's doable. If you are lucky, you might be able to find close to what you want off the shelf, and you can just mod it a bit.