2d raster scan photon counting optimizati​on

Question

I have written a VI for a 2-D raster scan and counting TTL pulses from a photon counter at every pixel.
It seems to be working fine and is serving its purpose for the time being. However there are a couple of thing I want to improve
Currently, I have 2 loops, on within the other for each of the axes (x and y) for the scan. Then within the inner loop, I am counting the pulses. Now, the problem I am facing is that I have not figured out how to start the counter from 0 at every iteration. Instead, what I am doing is running the counter twice in a for loop and outputting the difference between these two iterations using shift registers.
What I would like to do is avoid having to count twice for each iteration. I tried the start task and stop task vi but they seem to be doing nothing. 
The other thing that would be nice is to avoid having the loops. I have heard that it is "expensive" to use the loops and they cause some overhead time in the program. I don't really think it is bothering me at this point but I think if I have to do faster scanning rates then it might. Currently I am scanning at >100ms per pixel. 
(And also, just FYI, my raster wavefrom is slightly different in that in is a "triangle"instead of sawtooth. i.e. it goes to the end of the fast axis, then starts the next line from the end rather than the beginning to avoid sudden jerks to my hardware.)
I have attached my VI. thanks
Aditi
Attachments:
Galvo_Scan_Image 5.vi ‏92 KB

Answer 1

I cannot help with your counter issues.
The loops do have some overhead. It is on the order of a few nanoseconds. While loops are probably slightly slower than for loops because of the extra testing. Compared to your DAQ timing the loop time is completely negligible. Look at the test program below.
Since you know (or can calculate) the number of steps, for loops are probably the better choice. I think the conditional for loop is available in LV 2011.
You should probably move the AO Create channels VIs and associated Clear Task VIs outside the loops. Connect the task ID wires via shift register so that the value will be passed in the event someone enters start and end values which result in zero iterations of the loops.
Generally to speed things up you want to move any calculations and any displays (indicators) outside the loops when it can be done without adverse impact on functionality.  For example the divide by two can be moved to the outer loop; the inputs do not change within the inner loop.  If you do not need to see every update immediately, reducing the number of writes to the Intensity graph amy speed things up a bit.
Avoid right to left wires and wires behind other objects. These have no effect on program performance but they make it much harder to understand what is going on and to fine problems. I have attached a cleaned up version of your program.
Lynn
Attachments:
Loop times.vi ‏12 KB
Galvo_Scan_Image 5.2.vi ‏74 KB