11-30-2018 10:15 AM
I'm seeing very strange behavior when I acquire multi-span acquisitions on two NI-5644Rs simultaneously. It appears that portions of the spectrum are randomly mixed between the two devices. It is easier to see if you set the reference level of one device quite different from the other, so that the two device noise floor are several dB apart. It appears that some spans of the multi-span acquisition come from the other device's spectrum.
I first saw this behavior in an application I built, but I was able to reproduce it using shipping examples (RFSA Getting Started Spectrum.vi).
Steps to Reproduce: Run attached "RFSA Getting Started Spectrum (with stop).vi" and "RFSA Getting Started Spectrum (with stop) - Copy.vi" on two different NI 5644R devices. Observe the spectrum. I don't know whether the behavior reproduces on other RFSA devices.
Workaround: set # avgs to 1, and average outside the driver.
Software Versions:
LabVIEW 2015 (15.0f2 32-bit)
niRFSA.dll 15.0.0f1
ni5645RA.dll 14.6.1.3001
Is this fixed in a later version of the software? I don't see anything about this problem in the RFSA 18.1 readme:
11-30-2018 11:23 AM
Actually, looking closer, it seems the bug is not limited to multi-span. I'm able to reproduce the problem with a span of 6 MHz on both devices.
I've added additional notes in the word document, and re-attaching.
11-30-2018 11:45 AM
One more update: if I create a LV-built .exe out of the VI, I can no longer reproduce the problem.
Hint: was ni5645RA.dll 14.6.1.3001 built with LabVIEW 2015? I have run into issues in the past when a LabVIEW-built dll is used by the same version of LabVIEW that built it. The run-time engine can re-use threads from the development environment, I think. Perhaps this is what I'm running into, but I don't know what averaging has to do with it.
Anyway, another work around is to build my VIs into .exes and use those.
By the way, my OS is Windows 10 Pro running on a PXIe-8135.