fastest continuous data acquisition cDAQ

[Fibo]

Hello,

I've only done very slow data acquisition with LabVIEW, and I was wondering what's the fastest (conservatively fastest I'm not trying to push the limits) data acqusition and data logging that is possible on a compact DAQ device? and is the benifit of moving over to a cRIO from a cDAQ really just for more realiable speed and data acquisition? I'm interested in about 30kS/s over 5 analog input channels.  

much thanks! 

[Dennis_Knutson]

A hardware timed DAQ device is very reliable. Your stated rate is not especially very high speed. What do you want the rest of your program to do? Have you browsed the available products?

[Fibo]

Hi, the rest of the program won't be doing much, 4-20mA open loop control to a motor or something, nothign fancy.

I remember hearing that a Windows machine can only give you about 1ms loop time iteration, due to it not being a real time operating system.  I thought that this meant you could only sample at 1ms (1khz) continuously.  I've never had to sample at anything above a few Hz so I never explored this... I'm thinking that I'm off on how I'm interpreting this stuff.... if I have a compact DAQ how can I find out how fast I can continuously and realiably sample and log data? Is there a good link or something on the NI site I can read about this? I've been looking for something.

thanks!

[Dennis_Knutson]

The Windows time resolution has nothing at all to do with hardware timed I/O. The specs for each device are on the product page. If you would just browse the product pages, you would see the acquisition rates as part of the description - 250 kS/s, 500 kS/s, etc. If you've never acquired above a few hz, then you probably were using software timed acquisition of a single point.

[Fibo]

Yes I was just accquiring a sample every iteration

so these rates of 250kS/s are achievable with a regular windows computer?  I'm guessing just grabbing a bunch at a time and then during the loop iteration (which would be the slow part) take these 250,000 samples and write them to a TDMS file or something.... I guess keeping them in some buffer or FIFO or something? 

At what point would you say, ok this is too fast I need to switch over to an FPGA cRIO solution... would 500kS/s or something like that be the cutoff point where you need a real time OS?

thanks!

[Dennis_Knutson · Accepted solution]

The rates are easily achievable and you can stream to disk. It's only when you save to a text file that you need to start worrying about using a producer/consumer design and a queue (in general).

I would not be too concerned with real time unless you are doing some sort of control loop where the reactions need to be precisely timed (again, in general).

For very high acquisition rates ( high MS/s), NI sells digitizers that connect with USB so you have quite a ways to go before you need to start worrying.