NI 9209 DAQmx sampling rate

[Blokk]

Hello! I need to monitor 8 analog input channels using a NI 9209 module inserted in a cDAQ unit. I have 5 channels with RSE terminal config, +1 channel also as RSE but different range, and 2 channels with Differential terminal config. I would like to sample these 8 channels with a rate of 5 Hz. The maximum speed I can get now is a bit below above 2 Hz, unless I change the DAQmx ADC Timing Mode to High Speed. In this way I get the required speed. I would like to know what is the trade off in accuracy how much I lose? I could check the accuracy of my sensors, and it is easily possible that I am still OK with a lower DAQmx ADC accuracy. How can I get this information? Is there a white paper, or DAQmx property node to read out? Thanks!

[stevem181]

NI-9209 Data Sheet

They don't list the accuracy by Conversion Time, just typical and maximum.

steve

[mcduff]

Do you need to get the data point by point? Is it possible to set up the cDAQ to acquire continuously at 5 Hz? If so, set it up so and use DAQmx events to read every 50 samples or so, whatever update rate you need. In the NI examples search for DAQmx with events.

Lastly, the 9209 is rated for +/-10V what is the benefit of setting up the range for anything else? My understanding is that the 9209 has 24 bits over +/-10V, changing the range does not give more resolution or am I wrong?

Cheers,

mcduff

[Blokk]

---

@mcduff wrote:

Do you need to get the data point by point? Is it possible to set up the cDAQ to acquire continuously at 5 Hz? If so, set it up so and use DAQmx events to read every 50 samples or so, whatever update rate you need. In the NI examples search for DAQmx with events.

I do not think using Events can speed up the DAQmx Read. I think either I use an explicitely pre-defined sampling rate or I sample data points "on demand" with software timing, the limitation is the same (of course the jitter is another question, but I do not need accurate timing between data points). But I will try it, just to be sure. Anyway, changing the ADC timing to "speed mode" did the trick as I noted above...

Lastly, the 9209 is rated for +/-10V what is the benefit of setting up the range for anything else? My understanding is that the 9209 has 24 bits over +/-10V, changing the range does not give more resolution or am I wrong?

Hmm, if so, we do not gain any benefit setting the input to a lower range. In this case I will just use +-10 Volts for all channels... (actually based on my sensors' accuracy and resolution 24 bits is already an overkill even over a range of +-10Volts).

Cheers,

mcduff

---

[mcduff · Accepted solution]

Check out the Voltage (with Events) - Continuous Input Example, you can find it in the NI Example Finder. The vi changed my DAQmx life, . Using a heavily modified version of that VI I was able to continously acquire 8 channels each at 2MSa/s continuously with a 10 year old laptop.

Anyway looking at the datasheet, it probably will not help you with the High-Resolution Mode. The data sheet gives the conversion time per channel for High Speed mode as 2 ms, or 500 Sa/s. For 8 channels this gives you a max rate of 62.5 Sa/s. For High Resolution mode the conversion time is 52 ms per channel or 19.2 Sa/s. For 8 channels this gives 2.4 Sa/s, or what you saw earlier.

So the moral of the story, which you know, is if you want 5 Sa/s rate then you need High Speed mode.

Cheers,

mcduff

[Blokk]

---

@mcduff wrote:

Check out the Voltage (with Events) - Continuous Input Example, you can find it in the NI Example Finder. The vi changed my DAQmx life, . Using a heavily modified version of that VI I was able to continously acquire 8 channels each at 2MSa/s continuously with a 10 year old laptop.

This should not depend on the age of your laptop, but the DAQ device you connect to it. Could you share us what kind of HW you used to reach this sampling rate? 8 * 2 MHz ... hmmm, really, are you sure?

Anyway looking at the datasheet, it probably will not help you with the High-Resolution Mode. The data sheet gives the conversion time per channel for High Speed mode as 2 ms, or 500 Sa/s. For 8 channels this gives you a max rate of 62.5 Sa/s. For High Resolution mode the conversion time is 52 ms per channel or 19.2 Sa/s. For 8 channels this gives 2.4 Sa/s, or what you saw earlier.

Yep, numbers look like this...

So the moral of the story, which you know, is if you want 5 Sa/s rate then you need High Speed mode.

Cheers,

mcduff

---

[mcduff]

Could you share us what kind of HW you used to reach this sampling rate? 8 * 2 MHz ... hmmm, really, are you sure?

NI - USB - 6366 ( http://sine.ni.com/nips/cds/view/p/lang/en/nid/209076 )

This should not depend on the age of your laptop, but the DAQ device you connect to it.

USB unfortunately goes through the processor, there have been improvements with newer processors, even just plain USB-2. Eight channels at 2MSa/s gives 32MB/s just over half the max theoretical speed of USB-2. One the 10 year old laptop it would work only on one USB port, the other ports were shared and would not work. Also newer computers have SSDs instead of 5400 RPM spinning platters.

Cheers,

mcduff