"Fast" control system

[GJLR]

I would like to implement a control system using LABView and a DAQPad-6070E. Would it be feasible to try this, or is a real-time controller absolutely necessary?

I assume it should be possible to read a single value from a channel and write a calculated value to an output and repeat this loop at the highest possible rate depending on the computer. So far I have not seen rates above 1000Hz. Is it possible to give an indication of what rates could be obtained with the non-realtime SW and HW versions.

I was also thinking that using an external trigger source might help to get a constant sample period.

Thanks for any advice

Message Edited by GJLR on 12-04-2007 08:21 AM

[Ben]

I have used hardware triggered Timed Loops to run at 2KHz under Windows. Be aware that Windows reserves the right to act stupid.

10KHz or better is possible in LV Reat-Time. On high-end RT platforms I have measure loop rates approaching 200KHz but I have never "closed a loop" at that rate.

FPGA's can do about 20MHz (I believe).

Ben

[GJLR]

So it seems I would have to resort to another system. The trouble is that the DAQPad I'm using is not compatible with LABView RT as far as I know, requiring considerable investment in new hardware.

[Ben]

Those numbers are for "closing a loop" i.e. "Read, Think, and Write" operations.

If your readings do not have to be evaluated to control your writting then there are other options.

Ben

[GJLR]

Yes, closing a loop is what I intend to do and possibly even multiple loops.

[Kevin_Price]

What kinds of thingies do you want to close your control loop around?  Do they really have the bandwidth to benefit from a > 1kHz control loop?  Are there hazards if Windows messes up your loop timing and they go unstable?   The more it's crucial to maintain continuous control and the higher the control loop rate, the less you can hope for a Windows-based solution.

However, as Ben mentioned, with careful coding you can get Windows to run at 1 or 2 kHz.  I'd expect you could meet that timing for >95% of the loop cycles.  But some of the misses could easily miss 10's of control loop cycles.

-Kevin P.