I have a grievance with this question (and non-practical certification exams in general). I set up this VI using a simulated device, and it appears that this is what happens (time in seconds):
Time 0: Elapsed Time executes once, sending "0" to its shift register and "FALSE" to the Loop Condition.
Time 0+: DAQmx Read executes for 5 seconds, blocking the loop.
Time 5: DAQmx Read finishes reading one sample per second for five seconds, sending a 5-sample row to its shift register. VI is unblocked, loop immediately iterates.
Time 5+: Second iteration - Elapsed Time sends 5 to its shift register (since the first iteration took 5 seconds), and sends "Time has elapsed" to the stop sign. However, the loop needs to wait until all outputs are present to end. So it waits for DAQmx Read to execute before stopping.
Time 5++: DAQmx executes for 5 seconds, blocking the loop from exiting.
Time 10: Loop stops when stop sign is read and all outputs are available. The loop took ten seconds to execute, even though the second "Elapsed Time" output was 5 seconds - because the loop had to wait until DAQmx Read completed the second time.
My main grievance is that this is a devious question. The VI does not execute intuitively, and nobody would set up a VI like this on purpose unless they're trying to trick someone. If I asked someone "what if I wanted to read five samples at one Hz and stop after four seconds?" They would say, "either no samples are read or five samples are read, depending on the implementation." It is highly unintuitive that ten samples are read. If I wanted ten samples (i.e. Read to read twice), I would have put Read in a For loop, setting N = 2. If I wanted the loop to stop after a certain number of seconds, I would have set up Read to read a single sample so it reads one sample per loop iteration. Why would anyone want to end up with a 2x2 array where each row represents time-series data for the same channel over a different time period? Typically you want all data for a single channel in a single row. So my real grievance is, "this VI does not represent a solution to a typical use case." Why would you test people based on what they shouldn't do if they would never have thought of the wrong way to do it anyway? Don't you want people to automatically think of the correct solution, instead of 17 ways to do it wrong?
Secondary grievance: Why does Elapsed Time execute before DAQmx Read every iteration? Nothing in the code indicates the order of execution. Is there some general principle that "timing events occur before IO events?" Granted the answer to the CLAD exam prep question doesn't depend on execution order, but the value of Elapsed Time would be different for each iteration. Of course, you don't see this unless you wire Elapsed Time to a shift register. But you cleverly don't do this, since it would indicate the order of execution, which users aren't supposed to care about. But if you want to figure out what this VI does, you have to think about which VI executes first. The conclusion "it doesn't matter" only comes after the student has thought way too much about a VI that isn't a solution that one would normally think of. In other words, after about an hour of otherwise free-time is wasted, during which I could have been building radio controlled trucks.
This is the problem with multiple-choice certification exams - since it's not a practical exam, the authors are forced to ask either simple questions (e.g. "which VI would you use to read from a DAQmx device?") or clever "fool-people" questions like this one, that represent a wrong solution that nobody would ever come up with on their own unless asked "how might you do this incorrectly, but in a way that looks correct, so as to confuse people on a certification exam?"
Rant over.
