Initial FSK Application using Zedboard with FMCOMMS3; under-sampling Error while writing large amounts of data to the output file.


I am working to install and develop with OpenCPI. I am using the Zedboard with FMCOMMS3 as my front-end. I have succeeded in installing OpenCPI, the IDE with eclipse neon, and I have been able to check my OpenCPI installation with 'test-opencpi-rpm' and I have been able to successfully use the FSK Application 'filerw'. I run into an issue when I attempt to use the FSK application 'txrx' mode, I receive an error stating that my sample rate is too high and that bytes were dropped. While stating this, the program writes more data to the file then the amount in the initial file. I have attempted to use this on two separate installations of OpenCPI, but I received the same results. When I received my Zedboard and FMCOMMS3, I was able to run the default analog devices' transmit and received function; to view a waveform on a spectrum analyzer. I do not have a second FMCOMMS3 board to compare with or to attempt other debugging methods with the FSK modes 'tx' and 'rx'. I have attached a sheet of inputs and outputs that I have received while attempting to work with your FSK application model. I am wondering if you have seen any issues similar to this or if you know of any debugging tools or procedures that can allow me to trouble shoot this issue. I am unsure if this is hardware or software related, seeing that the 'filerw' and Lab1 'Loopback' functions have worked for me but the radio communications in FSK are not.


Virtual Machine with the Specifications of:
CentOS 7 Version 3.22.2 64-Bit
4GB of ram
4 threads of Intel I7-3630QM
OpenCPI Release 1.3.1


James Kulp
October 17, 2018, 10:26 PM

Given the error, what would you expect to happen? Simply less erroneous data?
Given that the data stream is corrupted, the amount of data will also be corrupted.
(anonymous logins can not notify the reporter, so I don't know if this comment will be looked at...)
Presumably the core issue is simply that you are getting overruns on that platform when running the data in both directions, which should be analyzed to find out what is slow.
Note that if the files being written are NFS network mounted files, they are subject to network congestion etc.




OpenCPI Guest