LabVIEW
Unable to save data every 5 ms
There's quite a few things wrong with your code, mostly style issues. Wire bends, right to left wiring, etc.
But the simples part might be what's causing you trouble:
Are you sure that times is called every ms? If it returns 6, 11, 14, 18, etc, the output might never become 0. You need to keep the previous save time, and compare the current time with the last time + interval time.
Are you sure that output ever gets 0? The integer is converted to a double, and that causes rounding. The output might be 0.00000000001. That 5 needs to be an integer. You probably be OK in this case, but you're walking on thin ice. And without good reason.
Search LabVIEW like a graph! 
wiebe@CARYA wrote:
There's quite a few things wrong with your code, mostly style issues. Wire bends, right to left wiring, etc..
I could not agree more!
- Blatant overuse of local variables causing potential race conditions (example: you read "ON" from a local variable while the control terminal sits right there next to it disconnected. Just wire from the terminal and delete the local)
- No, determination of "first" does not require an indicator and three(!!) local variables.
- Avoid overlapping objects on the front panel, they can cause extra redraw penalties (e.g. your plot legends)
- No, converting an 8 byte string into two DBL numerics does not require a postcard full of code! Correctly done, that will fit on a postage stamp!
- "Index array" is resizable. You don't need to wire indices if you want the elements in order.
- "Insert into array" is not the correct function to append a few arrays. "built array (concatenate mode)" would be correct.
- Creating/destroying file refs is NOT the correct way to save/not save and a 20ms wait won't protect you from anything.
- Byte order is meaningless if you are writing strings. Why all these long enum diagram constants cluttering the code?
- Where do you ensure that "stop" is false next time the program starts? (Why not use latch action with the terminal wired as is and a stop event in the lower loop?)
- Why do you enable/disable visa events if you never wait on one?
- Your U8 array diagram constants should either show all elements or a scrollbar. It is confusing if some elements are not visible.
- Two sequential sequence structures before the toplevel loop is one two many. You can put everything inside one.
- etc. etc. etc!
- This is just the tip of the iceberg!
@GerdW wrote:
Hi Dave,
@dave97 wrote:
How can I store the old value of my timer until the new value is 5 ms higher than the old value? I can't figure out a method.You should carefully read message#2 again and look at the images in that message!
Ever heard of shift registers?
In retrospect, that code wouldn't work.
The ms count is a U32 so 0 + 5 could take a long time to get the first trigger. The compare won't work well... Once the count wraps to 0, it will again take a long time before prev + 5 > current...
But the shift register is the key.
Something like this:
The subtract results in a valid dT, despite (or because) both values are unsigned.
If the U32 wrapped, you'd get e.g. 5 - (max U32-5) = 10
Of course you can save the U32 drama and use Get Data/Time In Seconds or High Resolution Relative Seconds.vi
- last edited on by
Content Cleaner
As far as speeding up your code, there are a few things to look at.
1. Your data processing is horribly inefficient. There is no reason at all to convert into a Boolean Array, combine, and convert back to numbers. Just use Unflatten From String on your original string data and you can go straight to your integer values before doing the math on those values.
2. You may need to go with a Producer/Consumer setup in order to separate the acquisition, processing, and logging into 2 loops. File IO tends to be slow, so this might be a way to gain up the processing rate.
