In fram 0, you have made a classic mistake and caused a race condition. This is why people argue against using local variables. In your vi, you have created a local variable of Nb Samples and wired the local into the divide funtion. This is what is happening:
When you start the vi, Nb Samples is 0 (default), so the local variable is also 0. The local variable does not wait for the Nb Samples indicator to be set. This is how Labview works. An output doesn't happen until all of its inputs are satisfied. The Nb Samples vi won't execute until its inputs get there, then it will execute and only then will Nb Samples be written to. The local variable on the other hand, has no input requirements (no wires going in), so its value (0, default value) gets sent out immediately. This 0 value gets fed into the divide funtion vi right away. Doesn't matter what else happens, you have a divide by 0 condition, which is always a no-no. To fix the problem, delete the local variable. Wire the Nb Samples indicator to the divide input. Now the divide won't take place until the Nb Samples is defined by the Nb Samples vi outptut. Now everything works like a charm. Very interesting sounds with FM.
You other local variables in the following frames are OK because they are written to in previous frames. Locals in following frames don't get populated with anything until those frames are opened. So when those following frames are executed, the local variable data is already there.
Race condition is defined as a race between writing to a local and reading from a local. Are you reading the last data written? Or are you reading data before the write? In your case, the local was reading data before the write.
