q: how do the various digital modes compare?
a: this is a hard one to answer. i have done a few tests with the most promising modes. from my experience i would rank the olivia modes first: 8 tones 125 hz is very sensitve but also very slow, then come 8 tones 250 hz, 16 tones 500 hz, and 32 tones 1 khz, to name the most common. behind olivia comes domino ex with fec, then pskfec31, mfsk8 (which is slightly better than mfsk16), then psk63f, pskam50, qpsk31, and bpsk31. the very narrow band and slow modes like pskam10 easily lose sync when subject to fading. jt65 should be usable with a s/n ratio of -30 db, but you can exchange only very little information with it. last is rtty, which has no error correction capabilities, and requires around 200 hz bandwidth. i haven't included other modes like thor, mfsk4, mt63 or fmhell for instance, since i have not enough qso data with them to make a comparison meaningful.
here a few figures from patrick, f6cte (author of multipsk):
lowest s/n for 2% errors: -9 db for bpsk31 (for white noise in a 3 khz bandwidth), -6 db for qpsk31.
lowest s/n for 2% errors: -11 db for pskam50 and -19 db for pskam10 (for white noise in a 3 khz bandwidth).
lowest s/n for 2% errors: - 7 db for psk63 (for white noise in a 3 khz bandwidth).
however the "white noise" scenario does not match reality very often. in every test i did, qpsk was better than bpsk. also it is possible to use sharp cw filters for all the psk modes, mfsk 8 and 16, olivia with 250 hz bandwidth and rtty, all of which fit through a 250 hertz filter.
