USB-4431 and Sample Rate increments of 349.2 μS/s

[TimS.]

Hello, 

I will shortly be using the USB-4431 to store RF data to files at a sample rate of   25,625 Sps. However, I noticed that the USB-4431 only actually has sample rates between 1kSps and 102.4kSps in 349.2 uSps increments. So, I won't actually be getting the sample rate of 25.625kSps that I request in the returned data. It will be some fraction off. The closest one I calculated was 25625.0001376 Sps.

This difference in sample rate is small but it will have a large impact on my recordings since I'm planning on recording for 12 hours or more. Because I have file based analysis tools that aren't capable of reading in a file with fractional sample rates there will be a drift in time measurements over the 12 hour duration. About 5.94432 seconds off by the end of the file to be exact.

Is there an easy way to return data from the DAQmx driver control of the USB-4431 so that it is at the sample rate I request or is there an easy way to resample the 25625.0001376 Sps to exactly 25625 Sps?

Any help is greatly appreciated.

Thank you,

Tim S. 

[Mr.F · Accepted solution]

Hi Tim,

You’re right; you are limited to discrete sample rates that vary by a small increment. This will lead to drift over long periods of time. There is not a way to change that in the DAQmx driver, it is a hardware limitation and I don’t know of an easy way to resample your data. Another consideration I believe you are overlooking is the clock accuracy specification of the USB 4431 which is 100 ppm which means the rate you specific could be up to 2.56 Hz off from the true rate of the oscillator. The error you mean, 349.2 micro Sps compared to 25,625 Sps corresponds to an error of about 13 ppb. In order to achieve a clock at a designated frequency and have that be stable and accurate over extended periods of time you would need to use a clock that is correlated to GPS time. The PXI 6682 can take in a GPS signal, the GPS signal being directly correlated to atomic time. If you are sampling over 12 hours and that level of clock inaccuracy is bothersome then GPS can be used to prevent drift issues like that. Let me know if you have any questions about this.

[TimS.]

Hi Matt,

Thank you for the info on the sample rate accuracy. It definitely sounds like that could be more of an issue for me than the 349.2 micro Sps increment issue.

I haven't used the DAQmx code or the USB-4431 before so I'm not entirely familiar with it's clocking capabilities. This USB device will be part of a larger system that will have a GPS unit. The GPS unit will be providing NTP sync to the host PC that the USB-4431 will be connected to. Since there is no "Ext. Clock In" on the USB-4431 how can I ensure that the USB-4431's clock is correlated with GPS time?

Thanks,

Tim S.

[Mr.F]

Hi Tim,

Unfortunately, the USB 4431 doesn’t have the ability to share clocks. You can only import start triggers for AI and AO tasks. Is the USB 4431 the only DSA hardware you can use? You could bring in a GPS synchronized Pulse Per Second (PPS) signal into another AI channel and then process that signal and then resample your other data sets like you mentioned earlier in order to get the true rate you desired. This would be very computationally intensive. In order to correlate to GPS time recommend different hardware and the PXI platform. Do you have to use the USB 4431?

[TimS.]

Hi Matt,

Due to cost and I/O port requirements (3 in and 1 out) the USB-4431 is the only device that will work for my project. I think that we may need to just accept the sample rate accuracy limitations and figure out whether or not it will have a big impact on any of our post-processing. I know that the ppm accuracy will be more of an issue than the drift but I may restart the acquisition process periodically in order to remove any drift that has occured between requested and actual sample rates.

Thanks,

Tim S.