Synchronize NI-9205 with NI-9234

[golubovski]

Hello,

I have a cDAQ-9178 chassis with 4x NI-9234 (DSA) acquiring 16 PCB IEPE accelerometers. I await a NI-9205 which I intend to use to acquire voltage from 4 velocimeters. The NI-9234s are synchronized through their channel expansion in a single task. I would like to add the NI-9205 to the chassis and synchronize it with the NI-9234s. I have the following questions:

1. timing sync

How to properly achieve the START and CLOCK synchronization of the NI-9205 with the NI-9234s? In PXI I export both from the ST master DSA. WHICH device (NI-9234 or NI-9205) should export WHAT signals to WHICH?

2. same or paralle task

Since I produce an TDMS file within the NI-9234s task it would be nice if I could just ADD the NI-9205 to that task to expand that single TDMS file with the extra 4 channels from the NI-9205. Is this possible?

Thanks in advance,

[Brad_K]

Hi golubovski,

You can add the NI 9205 to the same task as the NI 9234s. NI-DAQmx will automatically share the necessary timing signals. The modules in a task all run at the same rate* without drifting apart over time.

How tightly do you need to correlate the data between the NI 9205 and the NI 9234s? The NI 9234 has a filter delay of 40 sample clock periods (plus 3.2 µs) and the NI 9205 does not, so you may need to compensate for the phase difference in your application.

Brad

* Ignoring slow modules like the NI 9213 and NI 9219, which have the ability to repeat samples so that they can be included in the same task as faster modules.

[golubovski]

Brad,

Thank you for the prompt response!

1. channel expansion of different cDAQ/cRIO modules

Are you saying my channel expansion task can embrase both NI-9234s and NI-9205 with its 32 multuplexed AI channels? How do I recognize which modules can be combined in that convinient way?

2. the DF delay

I am aware of it since I have so far been working with PXI-4472 DSAs. How tightly can I correlate between the NI-9234 and the NI9205? AFAIK, in case I use dummy DAQ with the NI-9234s to compensate the 40 clock cycles I guess I will need to separate the tasks - how to integrate the readouts to a single TDMS stream? Additionally, for the fixed 3.2 µs do I have to calculate how many cycles considering the actual sampling rate or is there another method?

We need the spectrum up to 30Hz, maximum 100Hz, so Fs of 200S/s is OK. However, NI-9234 makes a mess below the declared minimum of 1.6kS/s without reporting an error (?!?) so I use 2kS/s which would be convinient to decimate down to 200S/s. Nevertheless, with Fs=2kS/s the antialiasing filter gives me a hard time in the sampled spectrum :))) So considering the Fs of 2kS/s the sync requirement you ask for would be 500 µs against the initial delay of 20.0036ms... Yet, the current application is purely in the spectral domain - amient vibrations and resonant freqency searches which lessens the significance of the triggering delay. But it would be nice to know the limitations should a more demanding application requirement arrise.

Thanks in advance,

[Brad_K · Accepted solution]

Hi golubovski,

> Are you saying my channel expansion task can embrase both NI-9234s and NI-9205 with its 32 multuplexed AI channels? How do I recognize which modules can be combined in that convinient way?

Yes. With the cDAQ-9178, any set of AI modules can be included in the same task, and in most cases that applies to AO and DIO as well, but you can't mix AI, AO, and DIO in the same task. The NI cDAQ-917x Manual goes into more detail about what combinations of modules, channel types, and tasks are possible.

> I am aware of it since I have so far been working with PXI-4472 DSAs. How tightly can I correlate between the NI-9234 and the NI9205?

Good, I wanted to make sure you were aware that this is something you will probably need to take into account. Off the top of my head, I don't know the answer to this question.

> AFAIK, in case I use dummy DAQ with the NI-9234s to compensate the 40 clock cycles I guess I will need to separate the tasks - how to integrate the readouts to a single TDMS stream? 

If you use two separate tasks and you use the built-in DAQmx support for TDMS logging, then you will get separate TDMS streams for each task. Note that a dummy read will still log to the TDMS file, so maybe this won't work in your case.

If you use the TDMS API to write to the TDMS file, you can do pre-processing on the data before it is written into a single stream, but there are some tradeoffs between ease of use and efficiency: either you write floating point data to the file (easier but less efficient), or you have to deal with scaling the raw data (more efficient but harder to do).

Another approach is to put all of the modules into the same task and compensate for the filter delay after reading the data out of the TDMS file.

> Additionally, for the fixed 3.2 µs do I have to calculate how many cycles considering the actual sampling rate or is there another method?

If you're compensating for filter delay by removing an integral number of samples from the data, then yes. However, there are other methods of processing the data that can compensate for a non-integral number of sample clock periods. I don't have a recommendation, but perhaps someone else does?

Brad

[golubovski]

Thank you Brad!

I will integrate the NI-9205 within the NI-9234s task pre-read the first 40 samples (coercing the fixed 3.6 us). And I will have a pretty combined TDMS output. I would remove fisrt 40 smpls afterwards I guess. If I have to, I would separete the 2 tasks, waste the pre-read 40 samples and combine the 2 TDMS files after the acquisition is finished.

Marry Crhistmas and a Happy New 2012!