How do you autoconfigure a linear actuator?

[Cody Herring]

I would like to use Labview to automatically configure a linear actuator to measure the voltage, displacement, and speed as a function of time using an autocalibration procedure.  I would like to do this by taking a measure of the actuator fully in (high voltage) then 1 inch out (lower voltage).  I'm not sure how to do this or if it is possible, but if possible I would like to make it simple to use via the front panel.

Please let me know if this request needs to be clarified, as it is somewhat of a difficult question.

[Henrik_Volkers]

Doesn't sound impossible. I have done thinks like this in the past. What kind of resources can you use?  What are the core specifications?  

I would say LabVIEW is ideal to do that job, however it`s software and only your spec and hardware($$) will set the limit 😉  

BTW: If your actuator is a electro dynamic one, (like a speaker or shaker) , you might go better in measuring the current, since the coil will heat up, change resistance and off you are with a voltage control.(depending on your spec). So, if you need an amplifier for your analog out card, it's worth of thinking about a voltage controlled current source. (Current sources are tricky, where goes the current if the actuator is offline?)  

Message Edited by Henrik Volkers on 10-07-2008 04:03 PM

[Cody Herring]

I'm using Labview 7.1 with a Penny+Giles D45887/N Linear actuator going through a NI USB-6009 A/D converter.

I'm basically trying to make it so anyone can switch between different linear actuators and possibly string potentiometers without having to do anything more complex than pressing a few buttons to measure max voltage, moving the actuator/potentiometer a set distance using calipers, then measuring the voltage again in order to get the distance/voltage ratio.  Then I'll implement that into a displacement/time graph indicator to get the speed of the actuator and string potentiometers.

Thanks for the quick reply,

Cody H 

[Conseils]

Just a suggestion but you could possibly identify the devices with resistors in the connectors and use a specific analogue channel to identify the external devices.

The external device would effectivly be identifiable by a resistor, you could use a preselected analoge channel to measure the resistance presented by the external device.

Thus the system would not require any user intervention. Of course this depends on the connectors and things.

This is a simpler form of TEDS of course before TEDS got all clever  😉

[Cody Herring]

How exactly would I implement this? Would it be possible to implement it inside of labview with little or no hardware besides the A/D card and the transducer itself?

(Please excuse the ignorance, I'm still learning the program! 🙂 )

Thanks for the help, I'll read up on how to do this; looking forward to any future ideas.

Cody H

[Henrik_Volkers]

I tried to find information about your actuator... without success 😞   is it a actuator or a linear potentiometer sensor?

Is it a limited number of known sensors, or unknown (new) ones?  I like the idea of an automatic identification (analog or digital, whatever inputs are unused) , the basic princible is to use unused inputs of your 6009 to identify the sensor.

One common way: You define one type of connector (and pinout) that fits to all sensors to your DAQ (6009) ( say a DB25). Use one (or two)  free analog input to measure a resistor, that is mounted on the sensor connector. Now the (your) Software measure that resistor and can identify the sensor, as long as you always use unique values for R with a secure difference of 5% to 10% of your input range (assuming 1% resistors 😉 ).  Good for up to 20 different sensors/actors

If you don't use the digital DIO lines, you can use 8 of them to identify 255 sensors ( and the xFF (to get to 256) for no spec), by adding pull down resistors (the 6009 has internal pull ups) to code your connector.

Coffee break over ... more later

Message Edited by Henrik Volkers on 10-08-2008 09:36 AM

[Conseils]

That's an excellent explanation of what I had in mind....

It deserves some Kudos

[Cody Herring]

Henrik,

I have to admit I didn't entirely understand what you were talking about :), but I'll read up a bit more on it and see what I can come up with.

(I'm a first year comp eng student just learning this program and these methods for the first time, so I'm still a bit behind as far as circuits and that sort of thing go)

 Could you clarify as to what kind of resistor I should use/how it should be connected?

Thanks again,

Cody H

[Cody Herring]

As for the linear actuator, it is basically an older version of the following:

http://www.pennyandgiles.com/products/products.asp?strAreaNo=402_10&intelement=1208

It has identical I/O to the one I am using.  I believe the only applicable difference is the voltage output.

[Henrik_Volkers]

So we are talking about a linear displacement SENSOR (a linear potentiometer sensor in this case).  Linear actuators sometimes have sensors incorporated , but are something different 😉

Your linear potentiometer output voltage is ratiometric to input voltage and position of the 'piston'.  

Your mechanical stroke length M is higher than your electrical E, so your approach for autocalibration will fail, because of the unknown dead length (length where a change in movement will give no change in electrical output) in the full in position.

Best approach for an two point 'auto' calibration are the boundaries of your real working range and position. I assume you (or someone else) don't want to use the full in position in your (his) application since it is outside the sensors working range.

Assuming your sensor is linear (as linear as needed) and your voltage source is stable during calibration and measurement, you don't need to know the absolute value of U, since all you need (as far as explained here) is dU/ds.

All the rest about auto identification with a special connector is useful if you have a limited number of known sensors and limited time and/or knowledge of the operator is a factor (standard in industrial use 😉 )

An other hint: If you make a three point calibration (needs a second caliber) you can proof the sensors linearity and validate your calibration result.