I'm about to build a set (6) of linear actuators for my robot (AUV) and was
wondering if I could run a few design issues by the group?
I'm thinking hat I can either use a rack and pinion or rotate a piece of
threaded rod. Would I be right in thinking that if you were to push on the
R&P method you would therefore be pushing on the motor and subsequently
motor draw would increase proportional to force applied? I'm not thinking
about using it as force feedback, but don't want to use up battery life if I
have a constant force pushing on it. And therefore would the threaded rod
therefore be the preferred method as the force is dissipated through the rod
and teeth as opposed to the motor itself?
Sticking with the threaded rod method, I presume I could calculate the time
to extend (I presume there's a better term) by multiplying under load RPM by
thread pitch? I've always heard 'slow' and 'linear actuators' in the same
sentence and using a 5000rpm turning a coarse metric thread (pitch 1.25mm) I
get 3000(if we say this is load RPM)*1.25=3750mm/min which sounds pretty
fast to me....
Has anyone here constructed a linear actuator and got any
suggestions/comments/pictures etc that could help me design/make it?
And finally, in order to have level of position feedback, I was thinking
about using a hall effect sensor on the rod/motor (depending which one I
choose) and using that combined with knowledge of knowing which way the
motor's turning fed into a 4029 to provide a value as to the current
position that a uC could then read.
Am I heading in the right direction here?
Thanks in advance,
in message news:kRmBh.338332$ firstname.lastname@example.org...
I'm a bit confused as to if you want your linear actuator to drive something
or to feedback something.
A screw will hold its place once powered down. A rack/pinion will not. If
you put an LVDT in conjunction with either, you will have very accurate
feedback for a servo system like in an RC circuit.
What is it that you are trying to do?
I've just built something similar for a FIRST robot.
1) Run the screwed rod in tension if possible.
2) You can run the screwed rod fast, but you may well see a lot of vibration
etc - support at both ends if possible.
3) As a general principle I would always try to connect feedback sensors
directly to the item in question - you don't need much accuracy here, so a
simple pot would probably suffice.
4) It is kind of obvious - this kind of actuator can produce very large
forces - think about limit switches and end stops - experience suggests that
hitting a solid end stop at high motor speed can be spectacular.
On Fri, 16 Feb 2007 18:37:04 GMT, "Michael"
If you are spending serious $$$, you may want to look at
commercial actuator suppliers like below. Some have feedback pots
for position indication. If you want to go really cheap and DIY,
you probably can hack a standard servo to do most of what you
need using the threaded rod. I'm currently looking at a similar
simple DIY setup, but I need a roller skate wheel bearing to use
as a cheap thrust bearing for the threaded rod. WalMart no longer
carries replacement skate wheels, so I'll have to expand my range
of parts suppliers. ;)
.........but don't want to use up battery life if I
I used cable around a 1/2" diam' drum .
If you do it right , it has same loss as
expensive rack and pinion .
But the ultimate is linear motor
and quick clutches/ locks .
The ram has springs to recover
the lost energy .
Its very efficient , dont waste batteries .
Microswitches have an 'operating' travel and an 'overtravel'. My experience
is that some fail fairly rapidly if you use up all the 'overtravel' every
time. The very small ones seem particularly bad.
Another common mistake with limit switches is placing them so that they are
the end stop ! Your first control glitch will demonstrate the folly of this
arrangement. Mount the switch off to the side, operated by a cam or similar
Wow ! Thats REAL digital , using MicroSwitches !
I use lots of ARM 7 mcu's with 144 pins ,
ussually one can use 60 GPIO's , for stuff
like reading 1n4148 glass diodes , in a string .
But if you need more precise , the ADC can
read 8 channels and the Timer-Match can do
many more inputs , to transducers , your
robot has .
Ideally , a robot should have ARM7 mcu's
in many places of movement , legs , arms ,
and wheels .
Each arm,leg , can send
rate of its "position"
to another ARM7 , thats reading the
rate gyros and other "centralized" stuff .
So , if robot falls/crashes ,
the s/w can improve automatically .
Robots are exciting , but human replicas
are not optimal , wheels are much more
useful , beacuse wheels can climb and
go fast , you just gotta articulate the wheels
clever like .
I will make one , but now im perfecting
a free OpSys for ARM , that trashes all
the silly C++ , assemblers and Pythons
and other farm animals .
New bidirectional protocol for USB ,
Software is fun to simplify ..
Those are things that rotate. Why do you want to put a linear actuator
into the system? If you just need more torque and can accept slower speed,
add a reduction gear.
Read the catalogs of Berg and of Stock Drive Products to get some idea
of what's available.
SubMarine ! Water has too much force.
You must use a long lever inside the Sub
then when you want to turn sudden , you
lock the planes , and take another step
with the lever . When you want to return
to nuetral , you release the lock .
Its like a jack . Normal , is very easy and
the force of water cant cause problems
against such a long lever , but when you
need to make a sharp turn , you crank
the lever several times and after each
step ,the planes are prevented from returning
so they 'jack" to extreme position .
Then simply release the planes , and spring
force returns them to nuetral .
On an Airplane , one would use "servos"
But not electric , but aerodynamic servos
that help to move the much larger control
You can do this on a SUB , but sailors
dont like complicated .
You could use a 2 piece plane ....
Good luck .