LabVIEW
Recorded output data has a delay between input and output readings
Solved!
Hello imperialmanoj,
I looked into your files, however I could not fully understand what to look at. In the Excel file you attached I found that the time stamps or all 20k samples in columns A and C perfectly match up (from the first row to the last), however column F has just one sample. When you write "they should be the same" - do you refer to the time stamps (column A, C, E) or to your measured voltage (column B, D, F)?
- last edited on by
Content Cleaner
I just found that there is more than one value in columns E and F. Had missed those before. What modules are you using in your cDAQ? How have you connected your Voltage out to the Voltage in? Just a wire between ao0 and ai0? Or also GND?
You LabVIEW code could produce the behavior you see. I never use ExpressVIs, so I am not fully aware of potential side-effect. However, I'd assume that your DAQ Assistant (for acquiring) sets up a hardware task to acquire at exactly 500Hz, whereas DAQ Assistant2 starts and stops a new software-timed task each iteration - that could be indeed behind the data your LabVIEW code on your PC generates and logs.
Consider moving to the real DAQmx API: Learn 10 Functions in NI-DAQmx and Handle 80 Percent of Your Data Acquisition Applications. You have much more control about what is happening with these. There are many examples shipped with LabVIEW that you can use as a start as well. (menu Help » Find Examples » Hardware Input and Output » DAQmx.
Edit:
I just noticed your output task is configured as Differential, your input task as RSE. When I set both to RSE, and execute using my NI-9263 and NI-9201, it works as expected. Displacement is <0.01 all the time.
(I replaced the Write to File with the Graph bottom right.)
I think I figured out the cause. I used a cDAQ-9178 with NI-9215 as output (9125 was a typo, right?) and a NI-9263. The actual hardware should however not make a difference. When I just run the I/O, it works as expected. When I write to a LVM file instead of Excel, it works as expected as well.
When I write to an XLSX file, I notice the Front Panel update slower and slower (starting at once every 200ms which is what you configured, but then going down to maybe once every 500ms). It seems that the Write to Measurement file cannot keep up when appending to big excel files.
As your voltage output is coupled to the file write performance in your code, this leads to the delay you noticed. I'd decouple these two. I'll make a demo for you, give me some minutes.
Wow, I was not aware how slow ActiveX(?) access to Excel is - this is what the ExpressVI has to rely on when writing to Excel.
Attached I have one potential solution for you. I use a Queue to pass all data from the acquisition loop (where you need good timing) to a logging loop (which can hang behind). So the measurement data is buffered inside the queue until it has all been written.
When testing on my PC, it has piled up 34 sample waveforms when reaching 0V again, and then needs some additional seconds to write them.
Is this a suitable solution for you?
Please note that your output will always lag behind your input as the input is a hardware timed task running on your cDAQ, but the output task is controlled by your PC.
