Minium sample frequency for PCI-6220 and USB-cDAQ-9172, with hardware loops ?

[hemmerling]

Hello,
what is the minium sample frequency for

PCI-6220 M-series
http://sine.ni.com/nips/cds/view/p/lang/de/nid/201761

and
USB cDAQ-9172
http://sine.ni.com/nips/cds/view/p/lang/de/nid/202545
with "any" module
, for example with
NI 9205
http://sine.ni.com/nips/cds/view/p/lang/de/nid/202633
, with hardware loops ?

Well there are two different  kinds of sampling
1) Timed loop - where Labview starts a new sample conversion each x seconds

2) Hardware Loop - Labview sets a hardware timer, and the data is transfered automatically by the
hardware to the driver and the Labview application. Until the arrival of the data the loop is "suspended".
This is the "prefered" sample  method, as here the hardware guarantees the sample frequency
AND the (block) transfer to the Labview application, so that there is no CPU-use burden for the Labview computer.

"Cheap" DAQ hardware like USB-6809
http://sine.ni.com/nips/cds/view/p/lang/de/nid/14605
don´t have such a hardware timer, AFAIK,
but with cDAQ and "any" modern DAQ PCI card can do this.

But as there is a hardware timer - in the hardware, I expect, what is the slowest setting ?
So there is a register which is counted down, or up. And this value is limited, either 1 or the highest value is the limit).
( In opposite, the fastest setting is of course limited by the speed of the A/D conversion, the number of channels ect - well known.).

So if I want to write an application which samples ( temperature ) signals from very-slow to very fast ( 10Hz! - very unusual),
I assume that I must write two different loops for data aquisition:

1) One for the hardware loop for fast and medium sample frequencies,
2) one for the timed loop for very-slow sample frequencies ( 1/minute or 1/h)

and I would like to know the low limit for the  hardware loop.

Sincerely
Rolf

Message Edited by hemmerling on 02-08-2007 03:04 AM

Message Edited by hemmerling on 02-08-2007 03:12 AM

[Brad_K]

Hi Rolf,

The PCI-6220 and cDAQ-9172 both use the NI-STC 2 system timing controller ASIC for timing and data transfer. You are correct that software-timed acquisitions can be as slow as you want and that hardware-timed acquisitions have a minimum rate. The NI-STC 2 generates the sample clock for hardware-timed analog input by dividing down a 20 MHz or 100 kHz timebase with a 32 bit counter. The minimum sample clock rate is about 4.7e-3 S/s (one sample every 3.5 minutes) using the 20 MHz timebase, or 2.3e-5 S/s (one sample every 12 hours or so) using the 100 kHz timebase. DAQmx automatically chooses the timebase based on your sample rate. The NI 9205 and NI 9211 don't impose a minimum sample rate limit, but be aware that the NI 9233 and NI 9237 do.

Note: You can only have one analog input task per timing controller. The cDAQ-9172 has one analog input timing controller, and it uses that timing controller for software-timed analog input as well. Assuming that your application will be using a single chassis, you will need to put all of your channels in one analog input task and throw away data for the "very-slow" channels. If you need to sample your other channels at a rate that the NI 9211 can't handle, don't worry. DAQmx will repeat the newest 9211 datapoint rather than limit the rate at which you can sample your other channels.

By the way, when you set up a USB-6008 to perform a hardware-timed analog input acquisition, the USB-6008 performs the timing, not LabVIEW. USB-6008 CPU usage shouldn't be any worse than with a USB-62xx or cDAQ-9172.

Brad

[hemmerling]

Thanks for the impressive infos,

you write
"The NI 9205 and NI 9211 don't impose a minimum sample rate limit, but be aware that the NI 9233 and NI 9237 do."

1)
How can I find out by specification ( from the user manual or data sheet ) ?
What's the point ?

2)
How about
http://sine.ni.com/nips/cds/view/p/lang/de/nid/202713
here the sample rate is "fixed" to a maximum of 100 Hz per channel ( 400 Hz per module ).
So here if the overall sample rate of all modules in the cDAQ 9172 is higher than 100 Hz, I get "repeated" values, right ?!

- currently this module does not work with cDAQ but just with cRIO, according to the support, but is expected to work with cDAQ 9172 soon or later ( we don´t believe / trust on this but were frustrated that we got a written quote from NI althought it does not work right now ) -

3)
****************************************************************
How about PCI-6220 if there is a timer chip like on cDAQ 9172,
is there a minimum sample rate like with the NI9205,9211 modules of cDAQ or not ? From what you say, with cDAQ it depends on the modules.. so I can´t conclude vom cDAQ to PCI.

I assume and expect yes, due to the nature of PCI cards.

Is there a difference with OTHER PCI cards of NI ( S-models, E-models,...) ?

****************************************************************

4) Btw, is there a difference between cDAQ and cRIO concerning this point, minimum hardware loop sample frequency ?

Sincerely
Rolf

[Brad_K]

Hello Rolf,

1) The NI 9233/9237 manuals have the specs:
http://digital.ni.com/manuals.nsf/websearch/5B0E2985BE2C3D4F8625720300133D8C
http://digital.ni.com/manuals.nsf/websearch/5759E6CBA6DF1AC9862571CA0068B5D6

Look for "data rates" in the specifications. These two modules have a minimum rate because they use delta-sigma ADCs which have built-in filtering that is designed to work across a certain range of sample rates.

FYI, there is a bug in DAQmx that limits the 9237 rate to 3.8 kS/s instead of 1.6 kS/s, but those are probably still much faster than your "very slow" rates.

2) That's correct, the NI 9217 is not currently supported in CompactDAQ.

3) My answer about the NI-STC 2 was for both the PCI-6220 and cDAQ-9172. Sorry if I didn't make that clear.

4) With CompactRIO, there is no NI-STC 2. Instead, you design the hardware yourself using LabVIEW FPGA, and so you can do hardware-timed acquisitions as slow as you want.

Brad

[hemmerling]

Thank you VERY much