Closing the loop: proportional valve w/load cells

[_orko_]

Hello,

I have a system with the following currently:

1) 3k hyd pump/reservoir with pendant control of 3-way directional selenoid valve

2) Vertically-mounted cylinder with no positional feedback

3) Wheatstone load cells

This is currently providing a bang-bang up/down control of the cylinder, which is hooked onto devices to destructively test (pulling them apart in the upward direction).  When the breaking point is achieved however, the load is not constant, since it relies on human reaction to bump the selenoid valve to the proper load.

What I am trying to achieve is a closed loop, where the proportional valve is controlled directly by the feedback of the load cells to achieve the load setpoint at all times during a ramping or static load test.

I am currently looking at a proportional valve from Continental (VED03M) with on-board electronics that will accept +-10VDC as input to control the direction and flow rate from the pump.

I only have experience in the motor control side of the house, so what I am looking for is NI hardware recommendations.  I know that I'm going to get a lot of responses to go with a real-time solution, but what modules would I need in a cRIO or cFP to provide the interface with the load cells as well as the motion control analog output?

Also, if money becomes a huge issue here, could this solution be done with a NI motion controller card with a UMI interface?  I would imagine that I would have to set the feedback to operate on an A/D line and use some sort of external (to the motion controller) interface to convert the low-voltage load cell input to a 0-5/0-10VDC level?

I would *really* prefer to stick with solutions that would provide NI Motion Assistant support, since that cut down programming time tremendously i the past for me with electromagnetic motor solutions.  Is this even possible using the cRIO/cFP solution?

Tuning... I would imagine that tuning the system would be a nightmare in non-PC (RT) solution.  It would take a lot longer since the PID/gains would need to be re-sent to the RT controller each time they are changed, right?  Am I wrong in this assumption?

Thank you for your time,

Joe "orko" Sines

[risheeb]

Hi Joe,

Thank you for posting on the National Instruments discussion forums.

Your application should be possible using either a motion control card or a cRIO.  Our motion control cards have analog feedback inputs which would allow you measure the load cells.  With the cRIO, the NI-9505 servo drive module is available, and you could use an analog input module for the load cells.

If you do decide to use a real-time system, keep in mind that you will need to program your application using SoftMotion.  This is somewhat different from using the Motion Assistant, so there may be a learning curve that you won't have to deal with if you decide to use a motion card.

Regarding the load cell signals, you will most likely need to externally amplify these signals before reading them, since these tend to be very small in magnitude.

Finally, if you do use a real-time solution, your PID gains can be shared variables, which will allow you to change them without re-compiling the real-time target.

I hope this helps!  Please reply if you have any follow up questions.  Thanks.


Rishee B.
Applications Engineer
National Instruments