Tony, Further complicating the static crash test conclusions is the effect of the static on the receiver AGC. If a long AGC constant is being used, the static crash is going to desensitize the receiver for as long as the AGC holds the receiver sensitivity above the decoding threshold. In such a case, the mode with the lower minimum S/N may recover sooner to the decoding threshold than the mode with the higher S/N. This may be why MFSK16 appears to beat out Thor (on the average). MFSK16 has both a low minimum S/N AND FEC, which appears to be a winning combination, especially as the band is starting to go out, as we experienced during our MT63-1000 trials (but without a lot of QRN, since we were on 20m). Depending upon the proximity of lightning strikes, and when signals are fairly strong, MT63-1000 may easily be the best mode - even better than Olivia - but there is ALWAYS some point that "the last mode standing" (probably the one with the lowest minimum S/N) is going to win when the band is going out.
The idea behind using NVIS antennas for NBEMS on HF is that propagation is more constant, since there is less dependence on the skywave, and also that noise arrives at a lower angle than the NVIS "cloud burner" signal. This reduces the effect of the static crashes, but limits the distance on 80m and 40m to about 300 miles. 73, Skip KH6TY NBEMS Development Team