Hello David,


Thank you for the new greatly improved THOR25 test version '.17'!


To avoid a mere repetition of your already run flawless LinSim test, I have repeated my 'usual' long DL0BS intro text simulation in the new THOR25 mode, using PathSim.


PATHSIM SIMULATION RESULT FLdigi v4.2.06.17 vs. v4.2.06.15 for the condition 'Mid-Latitude Disturbed NVIS AWGN SNR 0dB' using THOR25 @ 1500Hz:

+ Impressive improvements in average FEC (45%) and SNR (27%). For details, see the respective 2 extra sheets (with graphs) in the attached spreadsheet.

+ Decoding has greatly improved from 7 (v4.2.06.15) to 3, or even 2 errors, see my note 2 below.

+ This equals the overall count of 3 errors in MFSK32, but there, all the issues were present while fading in, and not so in mid-text (THOR25), see my note 1 below.


Note 1:

PathSim has a slow signal fade-in (bug?) and thus the very first few letters / lines are not representative of the simulation.
This has been present (and noticeable) in all my simulations using PathSim.
See the attached Spectrogram which illustrates the issue.

I specifically mention this here, because, if we disregard such 'early' errors, the error count is still impressively reduced from 5 (test version '.15') to 2 in the latest alpha release (test version '.17'). But if we compare it to MFSK32, the THOR25-vs-MFSK32 'gap' increases, because MFSK only produced its errors during this (questionable)early fade-in phase, while THOR25 still did produce 2 errors (or 1 at times, see below) in mid-text.


Note 2 (just in case you have any recommendations other than the 'standard' settings used here):

I have attached a screenshot of my current THOR Config. Not sure whether any different settings should be used instead. I have not found truly conclusive clues in the Help Files, especially regarding the most suited number of 'Paths', and 'Filtering' on or Off' incl. the 'Filter bandwidth factor' (I understand this is relevant in case of CWI presence, which is not the case here during simulation). I tried tweaking the 'Filter bandwidth factor' anyway, and values close to '1' did slightly alter the occurrence of errors: the 'FLEXRADIO' line turned from error to OK, but others popped up sometimes, so it was rather inconclusive. Same happened with 'Filtering' OFF. I thus have set it back ON using the '2.0' default. Switching 'Soft-symbol decoding' OFF also turned the 'FLEXRADIO' line from error to OK, so we must be close to a limit there. I have added another entry in the spreadsheet, showing this (mostly repeatable) result with 2 instead of 3 errors total. On the other hand, I could not find noticeable influences of the other settings. No idea whether this is 'as expected', or unusual.


- - - - - -


So bottom line, this latest version clearly marks yet another huge improvement! Any attempt to tweak it just a bit more (in the same direction), if possible, might help stabilize the mode even further.


- - - - - -


And I'd still ask to kindly consider my requests regarding the Rx Window context menu (please shift the 'Clear' command to the very end, if possible), and should there be a reasonably 'simple' way to realize it, please add optional SNR (and FEC?) logging into the auto-created FLdigi log files.



Thank you very much in advance - and all my best wishes to you for the upcoming New Year 2025!


73s Tobias
.-.-.