[wsjt-devel] wspr two-pass decoder status and .c2 file problem

[Steven F. <s.j...@ic...>]

For those testing wspr mode in wsjt-x, I have placed a .pdf file containing information on the current status of the experimental two-pass decoder wsprd_exp at this link:

https://dl.dropboxusercontent.com/u/33211132/signal_subtraction.pdf

I have been running the two-pass decoder with wsjt-x and with my 8-channel gnuradio-based setup for the past several days without any problems. If you would like to play with the new decoder, the Makefile in the wsjt-x/lib/wsprd directory will compile it. For testing with wsjt-x, I renamed the executable wsprd and copied it into the <wsjt-x install>/bin directory.

While testing this new decoder I used a batch of .wav files and also a separate batch of .c2 files that were written by the latest wsjt-x. I eventually realized that all of the .c2 files that were written by wsjt-x were contaminated by signal dropouts. See “Case 3” in the pdf document for an example and more details. After discovering the problem with the .c2 files, I saved a batch of .wav and .c2 files using wsjt-x and found that the .wav files were fine, but that the .c2 files were contaminated. This makes me wonder if the problem lies in the writec2 routine in wsjt-x. I welcome input/opinion on this from those who are more familiar with the code. Could this problem somehow be specific to my setup? 

Steve k9an