Can this M-Series hardware sample different AIs at different freqs?

[hecmar.arreola]

Hi!

I'm looking to use the

NI USB-6229

on a project.

I have two channels I would like to sample at least at 13kS/s, where as my other 20 channels I only want to sample at <100S/s.  Can this be done by setting up two seperate tasks and having them run at the same time?  I know on E-Series devices and low-cost multifunctional USB DAQ this is not possible.  Otherwise I will have to sample at ~11kS/s on all my channels and handle them differently in software.  I'm not sure if 11kS/s will be enough for those two "high speed" channels.  If I can setup two tasks and have channels being sampled at different rates then I will be fine with this hardware.

Thanks for your input.

-Nic

[Dennis_Knutson]

None of the DAQ cards with a single clock and muxed inputs (i.e. the one you bought) can have separate tasks on the same resource. Since the ~11 S/sec may not be enough, I don't see what you can do other than use another DAQ device or run two different tasks sequentially.

[Mark_E]

Hi Nic,

Unfortunately, what Dennis has said is correct. (Well, it’s good that what somebody posts is correct, but it’s unfortunate that it’s not the answer you were hoping for.) The only two hardware solutions are:

1.     Get a device that has the ability to sample as quickly as you want, or

2.     Get a second device (one for faster sampling, one for lower sampling)

The option Dennis mentioned would make your acquisition software-dependent, so don’t expect consistent timing between samples, but you could choose to do that if you’re aware of the consequences.

I did a search on ni.com/daq to narrow down USB devices with more than 16 channels (32 is the next step up) and 100kS/s. For 13kS/s for all 22 channels, you would need a device that can sample at least 286KS/s, which would leave the USB-6289 (625 kS/s), USB-6259 (1.25 MS/s), or the USB-6255 (1.25 MS/s). They have different specifications for resolution, number of AO/DIO channels, etc., and if you want to ask more specific questions about each one, feel free to visit ni.com/contact to speak to a Sales representative.

I hope that helps. Best of luck with your application.

[hecmar.arreola]

Hi Mark,

Thanks for elaborating on the answer to my question and proposing some solutions.  In the past, what I have done is sample ALL my channels at the speed my fasts channel requires.  I collect all my data using a Consumer/Producer VI pattern, and in the consumer loop I average down the samples in my slow channels while generally "window/trigger" the fast channel.  This works fine for me.  Since I'm looking at purchasing hardware now, my method for deciding if it is fast enough is:

Sample_Rate / No_Channels > Fastest_requried_sampling.

When it comes to the device I'm looking at, it appears that I am right on the edge.  The upgrade from this that will definitly work is also about $1000 more.  So now I'm looking at the alternative of running two seperate DAQmx devices.  Although we may get "enough" data from the device at a lower sampling frequency, which I'm about to explore.

Are there any articles NI has put out on the limitations of the USB 2.0 bus?  It seems like USB is a shared bus across multiple devices and at some point there will be problems with hooking too many things up to it such as Multifunctional DAQ devices from NI.  I would like to educate myself more on this limitation because I need to buy more multifunctional DAQ devices in the near future and I like the USB interface but it does seem like there would be some limitations involved in its maximum transferrates.

Thanks for your input!
-Nic

[ChrisR.]

You shouldn't need different hardware. You can use multirate scanning, as mentioned here: http://zone.ni.com/devzone/cda/epd/p/id/4795

The basic idea is to have your fast channels appear more frequently in your scanlist than your slow channels, so they get sampled more often. For example, say you have one channel ("A") you want to sample 10x as fast as 10 other channels ("B" through "K"). You could generate a scanlist like this:
ABACADAEAFAGAHAIAJAK.
In this case, the total sample rate is only double the sample rate of the fast channel, even though you're sampling 11 channels.

Hope this helps. Sorry I couldn't find a better link describing the technique. It's evidently not well known, although it's been around for ages, and is one of the reasons we allow long scanlists on our boards.

Chris