What to use as "Central Repository" during acquisition

[dblk22vball]

I am working on a program that will be conducting a manufactoring test on some of our products. It involves monitoring pressure, temperature and flow rate, and specific ranges during each test (each part has to pass 4-5 different tests).

I was thinking of using the QMH architecture for this, with 3 consumer loops, hardware control, a test control, and data logging/ui update loop.

The test control loop will be the state machine that handles what test needs to be run next, and tell hardware loop if it needs to change an output.

The hardware control will obviously control which digital output is activated, and getting the data from the instruments. It will continually acquire data (when a test is started), and if it needs to change an output, it will be told so by the test control loop.

The data logging loop will log the data and update the controls on the FP (at least this is my thought now). I was hoping to log to a TDMS file (which I can do), which is a waveform, but all of my "data checks" will need to just be the raw data (no timestamp, etc). I am using gauges with 3 needles, so it is a cluster of (3) doubles.

So I was looking to create some sort of central repository that I could put the raw data and then update the front panel, and do the analysis for whether or not to pass the tests.

Do I create a shared varible, or a local varible, and try to control access with semaphors or some other method? Or do I write a temporary file and then everything reads/writes (this seems inefficent).

Or is there a better way to go about this (architecture or central repository)?

Thanks.

[crelf]

QUOTE (dblk22vball @ May 28 2009, 04:01 PM)

So I was looking to create some sort of central repository that I could put the raw data and then update the front panel, and do the analysis for whether or not to pass the tests.

You could use a Functional Global - then you could do functional stuff too like formatting (if you needed to).