FIRST Tech Challenge Discussions
Feedback Node
You can use a shift register on a while loop (see below). After you drop a while loop, right click the outside and choose add shift register. Another comment if you look at the bottom left corner of the picture I uploaded notice it is orange and says NXT. If you have this as your target you shouldn't get things on your functions pallete that aren't available on the NXT.
-Steve
Mentor
Team 4150
Steve,
Thanks for the feedback. I have been using shift registers. I want to create a set of sub VI's for things like filters. The only way that I have been able to do this is by sending the feedback signal out of each VI in the hierarchy back up to the top while lloop, connecting it to a shift register, and then wiring the old value as an input down the VI hierarchy to my low-level VI. This works for one or two filters. It gets very old when I need to wire many more than this.
I have tried adding while loops around the filter Sub-VI to store my state. This ends up calling my filter either 0 times or two times. I can not figure out how to get the shift register to give me a past value and to run the filter only one time. Here is a picture of the test Sub-VI. This one does not pass the old value through when I call it from within a parent VI while loop:
I tried using a local variable. This works until I try to wire the output of a summer through the local variable and back to the input of the summer. Labview adds a feedback node automatically for me. Isn't that special.
This Z-1 works great. The problem is when I wire it up in my parent VI I get the following:
What good is a Z-1 block when Labview keeps adding a block to break an algebraic loop that isn't really an algebraic loop?
Here is what my navigation program looks like when I have to wire all of the Sub-VI Z-1s back to the parent shif register. I only need one or two shif registers if I find a way to do a feedback node in the Sub-VI's.
Help. 
The first approach you took is the recommended way to accomplish what you are attempting. Just to make sure, I just ran a test with this as my caller.vi:
and this as my callee:
I targeted my NXT and ran the Caller on it. As I would expect, the NXT display counted up from 1 to 10. I did not see any double calls or skipped calls as you experienced. Does the simple case that I have shown here work for you?
Tony
PS: This is my first attempt at using LabVIEW's code snippets in a document like this. It if works, you should be able to drag these images to a block diagram and it will copy the code in there. Good luck!
This problem seems to occur when I try to create my own feedback node (Z-1). If I build the entire filter, integrator, PID controller inside of the while loop, things appear to work. Here is a test parent and its child sub-VI accumulator. These seem to work.
The image on the left is the test block. I run the while loop 5 times. The image on the right is the accumulator. The sub-VI accumulates the input signal. The frustrating thing is when the students wire an output from a summer back into one of the input pins, they get a feedback node by default. This causes an error. I don't understand why the Labview target environmen won't support a default behavior of the tool. Very frustrating.
Thanks very much for the help. The comments were very useful. Yay
