Synchronizing AO, AI task in multiple PXI device

[NarayananK]

Hi All,

Well I am using a PXI 4461 and 4472 on slot 2 and 3 of the PXI consecutively.

However, i am not able to acquire the data from the 4461 and 4472 synchronously i.e. i have a finite delay about 3 ms in the meaurement and this increases with the increase in frequency of acquisition.

I cannot use the multi task synchronization technique provided in DAQmx 8.0 and later because, i am using an AO task for the 4461.

This is done to perform a swept sine analysis of a UUT.

I would like to know few things:

1. Does the Swept Sine VI of the Sound and Vibration toolkit have the capability to acquire multiple device data?

2. I have used the sample clock time base, the sync pulse and the PXI start trigger to synchronize the devices. Is there any mistake?

3. The sampling rate i have used is 30 Ks/s.

Kindly let me know if any clarifications are required.

Regards,

K.Narayanan

[Travis_W]

Hi Naru,

Unfortunately, the swept sine example program that comes with the Sound and Vibration Toolkit was written for a DSA or DAQ device with both inputs and outputs, but is not intended to acquire data from multiple devices.  You would have to modify the example to gain this functionality.

Part of the problem you're seeing could be caused by the different filter delays of the 447x and 446x DSA cards.  Page 5 of the 447x specifications, and page 6 of the 446x specifications show the ADC filter delays of the respective cards.  At 30kS/s, the difference looks like it will be about 25 samples which comes out to about a 1 ms delay.  This, however, doesn't account for all of the delay you are seeing, because this should be a constant 1ms delay at 30kS/s and up.

Do you see the same 3ms delay if you use the multidevice task support as mentioned in the above Knowledgebase?  For instance, if you feed in an external signal (not generated by the 4461) to both the cards with equal length cables and specify all channels in the same DAQmx channel constant, do you see this delay?  If the delay doesn't exist in this scenario, then could you attach your other synchronization code to this post so that we can take a look?  From what you've explained, synchronizing the oversample clock and the start trigger should be the right steps be synchronized.  There are examples that ship with the DAQmx driver, as listed in this KnowledgeBase, that should help get you started with this and verify the techniques you are using.

Have a great day!
Travis W

[NarayananK]

Thanks Travis,

Thanks for providing the link to knowledge base.

However as i am using a AO task for generation from 4461 and start the AO task before the AI task, if i used the synchronization method provided in the knowledge base i.e. Multidevice Task Support For DSA and SMIO Devices the AI task results into error the error message says that already a task is running (AO task ) hence, AI task cannot be started.

Hence i had to use the normal way of synchronization method as given in the example i.e. using the "Multi-Device Sync-AI and AO-Shared Timebase & Trig-DSA.vi" routing the start trigger.

However if i use this vi to synchronize my task then i get a finite delay of 3 ms between the devices.

What is not understood here is why i cannot run multiple task on the same device?

Regards,

K.Narayanan

[dsb@NI]

Is there a reason why you chose to use the inputs on the 4472 as opposed to the inputs on the 4461? Not trying to challenge your decision, but more trying to figure out how the application might grow in the future before making a suggestion that will be obsolete in the final application.

[NarayananK]

Well, i am using the inputs of 4461 as well as 4472.

The idea is to use multiple channels for perfoming Frequency Response Analysis for multiple responses.

The generation of single tone to the UUT is done using 4461 and acquiring the multiple response from different places of the UUT is done using two AI channels of 4461 and 8 AI channel of 4472. Thus we need 8+2 10 channels in all.

Any other clarification required kindly let me know.

Regards,

K.Narayanan

[dsb@NI]

You can try to synchronize the 4461 and 4472 via channel expansion. Program the AO channel to start on the digital ai start trigger. Add your 4472 AI channels and then add one of the 4461 AI channels to the AI task. Channel expansion and the explicit start trigger for the 4461 will synchronize all the various subsystems. Then, for Fs equal to 30 kHz, you measure synchronization with a constant 84.5 sample propagation delay. This delay is ~3 ms when converted from samples to time. This propagation delay is not dependent on the test frequency but is dependent on the sample rate, and the delay from 4461 AO to 4472 AI is due to the anti-imaging and anti-aliasing filters, respectively.

[dsb@NI]

One challenge in this application will be that there will be different delays through the 4461 inputs than through the 4472 inputs. This is something that can be measured and removed, but it is some work. The SVT Swept Sine API currently supports measurements where ao and ai are on the same device and only two ai channels (one reserved as a wrap back stimulus channel and the other the response channel from the DUT). I can send you the SVT Swept Sine VIs that I modified to enable the synchronization between the 4461 and 4472. Can you please send your email address to doug.bendele@ni.com and I will then send you the VIs. We can then look at what might need to be done to expand the number of response signals in the swept sine measurement. By the way, what is your DUT? It is getting late for me now, so I probably won't get back to you until morning.