This was the initial message I posted:

>To the multiple recipients of Biomch-l:
> I am part of the Mechanical Engineering Senior Design Seminar at
>Yale University. Our project is to design a Spine Tester for Dr. Panjabi
>of the Yale Medical School Orthopaedics Lab. The proposal that we are
>exploring involves a six degree of freedom robot called a Stewart Platform.
> In order to build this we are looking for a linear actuator (either
>pneumatic or electric ball screw) We are looking at controlling
>the actuator force, not displacements with a compluter. We want to apply
>about 25-50 lbs of thrust with a stroke length of about twelve inches.
> We were wondering if anybody has a suggestion of where we could
>find an actuator that would do this.
> If we can not find one, we will probably have to solve this problem
>by controlling displacements measuring with a force sensor.
> If anyone has experience with building or designing a Stewart
>Platform and can give us helpful advice we would appreciate that.
> Thank you for your time. We will post a summery or replies.
>Reply to:
>Andy Snow

I apologize in advance if I left anyone off of this. My mailbox has been a
mess. Thank you to all who helped. I believe that this is a great froum
and has helped me, a student recently becoming interested in biomechanics,

Here are the replies...
From: Steve Bresina

Dear Mr. Klein,

Look up a student in the business school there at Yale, Terry McIff.

He developed a six-degree-of-freedom testing machine here at our laboratory
before going to business school.

Good Luck,
Steve Bresina

From: Sean W.P. Batte


I have a suggestion.

We use bimba actuators in our lab very frequently.
They are pneumatic, come in various sizes capable of different
forces. Force control is simply a matter of adjusting the
voltage. We either use a simple 12V power supply to control
the voltage in constant force applications or a virtual
instrumentation software package Labview. Labview has the
advantage of offering variable cyclical or static loading,
ramp rates, variable waveform outputs, selectable amplitude &
frequency, and selectable # of cycles.

We have calibrated the Bimbas we use with an
Instron 8501, and they are in agreement with the theoretical
outputs. We supply them with standard 'lab' compressed air
or a compressor (for loads >700N with a 2" bore actuator).

I hope this helps.

Bimba Manufacturing Co. Labview:
Monee, Illinois National Instruments Corporate HQ
60449-0068 Austin, TX
(708) 534-8544 78730-5039
fax 534-5767 1-800-433-3488

Sean W.P. Batte
HULC Musculoskeletal Lab
St. Joseph's Hospital
London ON
(519) 646-6000 ex.5279

From: J. M. Stuart

I would think that pneumatic cylinders would be ideal for your
application. All you need to do is to provide a constant air pressure
to get the constant force that you are looking for.

Companies that make such air cylinders are Bimba, Aro, and others. The
air cylinders are available thru distrubutors or industrial supply
houses such as Mc Master Carr.

J. M. Stuart

Engineer / Inventor
From: Ranganathan Ramanathan

We have tried some ball screw actuators for other applications
and it is a linear actuator and has very good force capabilities.

You can call the company and get a catalog and check if it meet
you requirements for the stewart platform.

Motion Systems Corporation
600 Industrial Way West
EatonTown New Jersey 07724

Ph 908 222 1800
Fax 908 389 9191


================================================== ==================
\__ ____ \__ |Name : Ranganathan Ramanathan
/__ / / / \ /___/ / / / \ |S-Mail : ASEL, AI DuPont Institute
/ /__/ / / ___/ /__/ / / | PO Box 269, Wilmington | DE 19899 |Phone : 1-302-651-6869;1-215-895-2376
WWW Home Page :
================================================== ==================
From: Zoltan

I think your Stewart platform would work best if you used ballscrew actuators
with current sensing and optical encoders on the drive motors for force,
position and velocity feedback. We have a home page on the web and you can
buy from us force sensors, motor controllers, perhaps multiprocessor systems
and a ballstick for operator interface. We can deliver these components as an
integrated turn-key system with software that is customized to your
application. We can also recommend motor and ballscrew manufacturers. For
example Pittman makes a small motor that we use in the robot gripper that you
can see on the web, this has 39 ozi of torque. With a standard 8 turn per
inch ballscrew this will give you about 100lbs of force maximum. We have
files on the ballscrew manufacturers too.



From: David Knutsen

You might try contacting Giddings and Lewis or Ingersoll Rand. They are both
machine tool builders that have machines similar in concept to what you are
proposing. I don't have their phone #'s, but they should be listed in the Thomas

From: Jonathan R Barnes

The combination of a voltage controled pressure regulator and a
pnumatic cylinder could be an answer if you can live with the
friction dead zone.

I know norgren do such equipment, but don't know if you would
have a local supplier.


From: Mile Erlic

There is a potential source for your needs. The company
provides coreless dc motors. These have very high power/weight
ratios. They can be located at:

Portescap US Inc.
National Headquarters
36 Central Avenue
Hauppauge, NY 11788
Tel. 516-234-3900
Fax. 516-234-3986

They have a particular motor/gear model that has an rotary
to linear transmission. The standard motor/gearbox
combinations are:


The gear box model number is L10.

The maximum holding force is 45 lb. The linear speed range
is from 0.1 to 3.9 ft/min.

Mile Erlic
Precision MicroDynamics Inc.
We provide board level products for robotics and motion control.

From: Roger Barry Hertz

Use pneumatic with an accurate pressure regulator if you need force control.
Neglecting friction the F=P/a will be the most direct way to control force
on your Stewart Platform [BTW, ckeck out Jean-Pierre Merlet's html refs in
his posting to comp.robotics.research for more papers on Stewart Platforms].

You'll be able to find a good list of suppliers and manufacturers if you check
out the site (free membership, good searches). A company
named PhD comes to mind that has a lot of pneumatic products.

I recommend staying away from ball screws and electric motors if you want to
do force control. Current is a very poor estimate of force, and force feedback
methods are overly complicated, IMHEO (in my honest engineering opinion). I
would go with the most direct method, by controlling pressure, since most
pneumatic control valves control that variable directly. It seems that
"across" variables are easier to control then "though" variables.

Just my 2 cents.


University of Toronto
Institute for Aerospace Studies

From: bill billotte From personal experience on the Instron machine we have it is very touchy
when you are doing load control. For instance, when you tell the machine to
put 500Lbs on the test specimen it tries to do this even after the specimen
breaks. Luckily there are safe guards to prevent the actuator arm from slamming
the machine. If you do build such a system do not forget safe quards such as
this as your whole machine could be quickly ruined. Displacement control is
by far the safest and easiest to control. hope this helps. please feel
free to email me at the below address as i check it more regularly.


From: Peter Kinsella

Warner Electric of Toledo, Ohio, can help. You need an LA 10 model.
They should be able to advise further but let me know if you require any
further help.
Peter Kinsella

From: Jean-Pierre Merlet

We have some experience designing and building what
you called a "Stewart platform" (BTW it is the wrong name..).
But the data you have given are a little bit insufficient..
If I understand well you want to apply a force of 100-220N in
the actuator but what is the velocity you want?
For our first prototype we have used electric linear actuators
used in airplane with a force of about 1000N (but you can find
some with less force) but they are slow. For our second prototype
we have developed our own linear actuator (this is not so difficult)
with ball screws and rack and pinion. In our first prototype we have
also force sensor in each link and that work pretty well. The only
problem is that the force measurement are not isotropic: the sensitivity
is very good for forces in the x-y direction but not so good in the
z direction (normal to the moving platform).
As for hydraulic actuator V. Hayward at McGill University in
Montreal use such an actuator with sensor
for measuring the displacement together with force measurement. But I am
not sure of the stroke, better to ask him directly. I am interested
in your application and we have a lot of experience for designing parallel
robot, especially for finding optimal design : for example if you give
the desired workspace, the desired accuracy (or any other criterion to
optimise) we are able to find what will be the best position of the
join centers.

If your are looking for references I maintain a bibliography data base which
can be found on:

From: Juan Domingo
Subject: On your question about actuators

Dear Andy:
We purchased recently a 3DOF platform (even there are 6DOF, too)
manufactured by the English company:
Denne Development Ltd.
Unit 4, Cedar Park
Cobham Road
Ferndown Industrial State
Winborne, Dorset BH21 7SB
Tel. 44 (0) 202 861 661
Fax. 44 (0) 202 861 233

The platform incoporates electroneumatic actuators (so it is said,
most of the force is exerted by the neumatic system, but fine
positioning is done by electric means). They are equipped with
linear encoders to get feedback of the position. I send you
this because they also sell these actuators sepparately. Please,
if you post a summary, make clear that we have no commertial
relationship with them, we are just customers.
Yours sincerely
Juan Domingo

From: Joel M. Bach

I developed a six degree of freedom testing device for knee testing
applications. This apparatus was pneumatically actuated and a computer
controlled the force application (not displacement). Your problem is not
in finding an actuator but rather an appropriate controller. I used a
servo-pneumatic valve/controller combination from a company called DYVAL
located in the Palo Alto area of California. Using this combination I was
able to utilize easily available pneumatic actuators. Force feedback was
provided to the controller board by a simple 1 DOF load cell. If you would
like more info please contact me, either by e-mail or by phone.


*****************LIFE_IS_NOT_A_REHEARSAL********** *******
* *
* Joel M. Bach *
* *
* UC SanFrancisco Ergonomics Program *
* 1301 S. 46th St Bldg. 112 *
* Richmond, CA 94804 *
* *
* Phone510)231-9448 Fax510)231-9500 *
* *
* *
* *
************************************************** *******
From: Rich! Walker

The Shadow Project makes and markets a selection of pneumatic muscle
actuators. These come in various sizes; one about 30cm long will provide a
pull of 300lbs at 3 bar. The force is more easily controlled than the
contraction. However, if your application is restricted to pushing, move the
actuators to the other sides of the joints :->

For more information, feel free...


Rich! Walker System hacker @ Shadow Project, working to build the Domestic Android. Robotics researcher for the
GNU code hacked University of East London. Acorn-compatible.