Ye to put this into context:
This DAQ runs concurrently twice in two separate Exe's controlling two custom designed boards which run an experimental real time fast feedback position correction system for a particle accelerator. There is also a separate executable which runs a code which controls a mover system for the apparatus which measures the beam inside the accelerator and another which runs a system used to control the phase of some input signals.
Although these systems are complex they are more traditionally what I understand labview to be used for, passing information from a serial connection into data files and setting firmware on the boards.
The calibration Exe (a new idea and very much a work in progress) then sits on top of these other exes and implements a suite of control and calibration algorithms (written in C++ and Python) designed to perform various tasks. The problem I am having is a bunch of race conditions. We could of coarse implement all these systems into one large Exe, this however puts us into a situation where if one thing crashed (and if you know anything about experimental physics this isn’t infrequent) everything fails. It also means that changing one part and in particular the control algorithms could cause everything to fail, something highly undesirable as beam time is precious and at the very least you want your DAQ to always work.
Originally all this code was “linked together” using Global Variables and although there were some race conditions these were all carefully ironed out. When the decision was made to compile the individual VI’s into separate executables we ran into problems which I’ve been unable to fix. I’ve had a number of ideas, thinking maybe I could use the realtime module for example but this seems to be designed more for controlling discrete systems on different machines rather than many systems on the same machine.
The main problem I have as I say is to lock the loop between the DAQ and the calibration VI together. Every time a bunch travels through the accelerator (about 1.5Hz) the DAQ triggers. I have set up the system so that the DAQ then processes the data and afterwards passes the settings it has used (how many tiggers it has saved to a file for example) to the calibration VI along with a notification. This is setup to wait for this notification and when it receives it looks at the information given and makes decisions based on this information about what the DAQ or other VIs should the do (ie stop saving data, move the movers and then start saving data again with a different filename).
The problem seems to occur because as I can’t work out how to send a notification I can’t make the DAQ and the calibration VI lock together. This means they end up in a situation where both will be reading and writing to the same variables at the same time which obviously means the whole thing gets screwed up.
I get the impression that this system isn’t strictly what labview was designed to do and I may just be pushing it to it’s limits. I also realise however that I come from perhaps a more traditional programming background (Java, C++ etc) and I’m new to it and don’t understand it’s full scope so if anyone could offer me a way of doing this or a better way to structure the system as a whole I would be very grateful.
Thanks.
