Controlling four Linak actuators.......

[kitarusapien]

The units are Linak medical bed actuators, run on 12 or 28V, usually run by a hand-controller and usually without installed potentiometers. What I would like to do is operate four of these units with some sort of controller/power board and a standard PC control-stick so that they mimic the control-sticks' inputs as accurately as they can operate under normal operating conditions to basically computer-mix the actuators(stick back, all actuators extend, stick forward, all retract, any position in between, all actuators assume their proper position to reflect the stick's position.) If if this can be done, please let me know.

Thanks so much!!!!

 

[Skogsgurra]

It is not easy to do what you want to do with standard linear actuators without position feedback.

One way of doing it would be to use "sensor bearings" in the units and use their output signals to keep track of linear motion. The math from actuator position to target position can be complicated, but if the four Linaks are not dependent of each other (each Linak does its own thing) then the math is simple.

You could also, depending on the resolution you need, put little sensors in the actuators (I think that Linak has versions with sensors built-in) and count output from them in the same way as said above.

A few questions that will have an impact on what solution you should chose: Is this a low or high volume application? Does it need to be non-volatile (be able to resume operation without sync operation after a power outage)?. Is it a cost sensitive application? Is the environment tough or hostile - or is it a normal in-door climat?



Gunnar Englund

http://www.gke.org

 

[kitarusapien]

I can order the Linaks with pots built in, but I'm not sure that is what is needed for accurate feedback. The application is relatively high-volume with numerous continuous position changes, but very little structural load. Not sure what you mean by non-volatile load; my only concern is that each time the system is fired up on it's 12 or 28V power, the stick's position is emulated in the actuators' positions(ie: if the stick is spring-centered, all the actuators reflect those postions.) Relatively cost-sensitive, normal envir, not hazardous to electronic components. THANKS!!!!

 

[UKpete]

kitarusapien, if you don't have a sensor acting as a feedback device then you have an open loop system and you will get cumulative error with this type of actuator (motor driven, rack and pinion I believe).

One type of linear actuator which isn't particularly accurate though doesn't suffer from cumulative error is a voice coil type permanent magnet actuator that is spring loaded. The voice coil actuator is basically the same type as used in a loudspeaker (hence the name). Because it is spring loaded, the position is a function of the current hence approximate positioning can be achieved BUT the actuator must be energized continuously. For a given holding force requirement it is relatively large and expensive (in terms of both first cost and energy use), though it can be fast. One application is active valve control in a piston engine.

Better to get a position sensor I would say.

 

[kitarusapien]

I'll go back to the Linak site, and see what they recommend for accurate feedback control...mayhaps they have units with such controls already installed. The unit I have for testing does not. I'll report back with the info they give me to bounce it off of you folks.

 

[kitarusapien]

Here's their description of an optical encoder that can be rigged for the actuators:

"The optical encoder is used for positioning and servo control of an actuator."

 

[kitarusapien]

As opposed to their description of the potentiometer they offer:

"The potentiometer is used where the absolute position of the actuator is needed."

So, which do you folks think would work best? They also offer a controller board which can work two actuators at once......is it possible to link these actuators by two of these controller boards? If it's possible, I can get the specifics, and post them here.

 

[itsmoked]

A potentiometer should be more than adequate for most applications. (Don't know what yours is.)

As you mentioned the "bed" control pendants, they usually run one axis at a time so the power supply can be capable of only running one axis at a time. Being able to run multi axi will require a hefty supply. (less cheap)

 

[kitarusapien]

Can you elaborate on "hefty supply"? Are we talking power, physical size, both?

 

[itsmoked]

hmmm. The supply that it takes to run one unit will certainly be smaller than the one to run 4 at once. If you run an unregulated supply, typical for this type application, it would probably be two or three times larger and maybe 4 times heavier. If you run switchers they would probably be a third the size? But switchers cost maybe twice as much as unregulated supplies. Can't really recommend knowing so little.

 

[UKpete]

Regarding the choice of potentiometer or optical encoder, I would think that the former is cheaper, and is probably fine.

Optical encoders may be used where there is high speed motion. They generally have very high accuracy as well.

 

[itsmoked]

Yeah UKpete I agree. Anything run by a joystick can't possibly need extreme accuracy.

 

[Skogsgurra]

As I said, a Sensor Bearing seems to be the best choice if you do not need non-volatile storage of position. I gather that few of you heard about such a device. It is an encoder built into a standard DGBB (Deep Grove Ball Bearing). It has a much lower installed cost than a normal encoder and it has a resolution that makes it ideal for low precision applications. Google "sensor bearing" for more information. As it happens, the second item found has some personal interest. ;-)

Gunnar Englund

http://www.gke.org

 

[UKpete]

Sorry Gunnar, I didn't recognize the term "sensor bearing". I have heard of them but I haven't seen them used but I guess they are straight forward enough and reliable (especially if SKF is supplying them).

I have used Linak actuators with integral potentiometers in a valve actuation application. The coupling was via a Bowden cable as they couldn't be located next to the valve; no problems to report. It's probably better to buy the ready assembled item rather than strip down the actuator and fit sensor bearings.

 

[kitarusapien]

OK...I'll get the Linak units with the installed pots, but now, this power supply issue.........LOL!

 

[kitarusapien]

I'm thinking one power supply for each actuator mounted closely to each unit would be best, but.......what is the overall controller to which the stick inputs to? Do I run a standard PC board with a subroutine that simply maps the stick's patterning? Should I switch over to a computer forum to figure out the controller-to-power supplies-to-actuator rigging?! LOL! Thanks all, especially for being patient with an obvious newb!

 

[itsmoked]

If your're talking cheap, one supply per actuator ISN'T!

If you are talking easy, clean, and painless, one PS/actuator IS!

Make your choice.

You should look into a regular PC joystick to "output" card. This will allow a cheap replacable joystick and different input devices like game controllers etc.

 

[kitarusapien]

I'll take the more expensive, but easier route of one power supply per. The electronics hooking up to the actual power supply still boggles me, though. Ah well, a searching and a studyin' I go!

 

[Skogsgurra]

I think it is about time to discuss power consumption. How many amperes do your actuators consume? Or, if that is easier; what force do they have to produce and at what speed?

It is not automatically so that a single power supply has disadvantages over the one PS/actuator approach. It all depends and it depends on a lot of factors.

Regarding the positioning accuracy, it is sometimes possible to have a +/- 2 or 3 percent accuracy (=repeatability when moving from fully retracted position to setpoint) without position sensor. The trick is to measure motor EMF (voltage induced in armature winding). That can be done rather cheaply by measuring load current and compensate for it (well, there are a few more things to it than that). It will always be cheaper to do that sort of control using a central (common to all actuators) control unit than having the intelligence spread out over four different units.

I would recommend One PSU and One Control Unit. Definitely more economic and also much easier if you need to have the different actuators communicate or interlock.

Gunnar Englund

http://www.gke.org

 

[kitarusapien]

Here's the basic specs on the Linak actuators:

Max. thrust: 750 N
Max. Speed: 50 mm/s
12V or 28VDC

I only need these units to have a maximum 'throw' of two inches fully retracted to fully extended.

Does this info help, or do I need to find out more? Thanks!