Great information, Dave, On other thing that I can not understand is why THOR's performance proved to be so poor on Tony's tests. The robustness to multipath and Doppler does not seem to show up although sensitivity at -15 dB SNR seems quite good.
It might be understandable with the more severe amounts such as high latitude 7 msec path delay and 30 Hz Doppler, where the test indicated no copy with THOR11 even at -3 dB SNR. But even at more modest low latitude 6 ms path delay with 10 Hz Doppler and a -8 dB SNR there is still no copy. And most surprising is the Mid-Latitude 2 ms path delay with only 1 Hz Doppler at -8 dB and THOR11 was decoding only 80%. At most of these conditions, Olivia 500/16, and Olivia 500/8, and often MFSK16, provided perfect copy and THOR11 showed no copy at all. Can anyone explain how this can be? 73, Rick, KV9U David Freese wrote: > The following is an excerpt from the web page "Sights and Sounds of > Digital Signals", http://www.w1hkj.com/FldigiHelp/Modes/index.htm. > > THOR Modes > > General Description > > THOR is a family of offset incremental multi-frequency shift keyed > modes with low symbol rate, closely related to DominoEX. A single > carrier of constant amplitude is stepped between 18 tone frequencies > in a constant phase manner. As a result, no unwanted sidebands are > generated, and no special amplifier linearity requirements are > necessary. The tones change according to an offset algorithm which > ensures that no sequential tones are the same or adjacent in > frequency, considerably enhancing the inter-symbol interference > resistance to multi-path and Doppler effects. > > The mode has full-time Forward Error Correction, so is extremely > robust. The default speed (11 baud) was designed for NVIS conditions > (80m at night), and other speeds suit weak signal LF, and high speed > HF use. The use of incremental keying gives the mode complete immunity > to transmitter-receiver frequency offset, drift and excellent > rejection of propagation induced Doppler. > Protocol > > These are unconnected, manually controlled message asynchronous > simplex chat modes, using binary convolutional Forward Error > Correction. The default calling mode is THOR11. > Coding and Character Set > > A binary varicode with ASCII-256 user interface (same as MFSK16) is > used. Lower case characters are sent faster. An ASCII-128 secondary > character set extension allows a fixed (typically ID) message to be > sent whenever the transmitter is idle. Modulation uses two dibit > pairs, symbol synchronous, differential. > > The FEC system uses binary convolution to generate two dibits per > varicode bit, and halves the corrected data rate compared to the > equivalent DominoEX mode. Rate R=1/2, Constraint length K=7, > Interleaver L=10 (40 bits). > Operating Parameters Mode Symbol Rate Typing Speed1 Duty Cycle2 > Bandwidth3 ITU Designation4 > THOR45 3.90625 baud 14 wpm 100% 173 Hz 173HF1B > THOR55 5.3833 baud 22 wpm 100% 244 Hz 244HF1B > THOR85 7.8125 baud 28 wpm 100% 346 Hz 346HF1B > THOR116 10.766 baud 40 wpm 100% 262 Hz 262HF1B > THOR16 15.625 baud 58 wpm 100% 355 Hz 355HF1B > THOR22 21.533 baud 78 wpm 100% 524 Hz 524HF1B > > Notes: > > 1. WPM is based on an average 5 characters per word, plus word space. > Values based on sending 100 "paris " words. > 2. Transmitter average power output relative to a constant carrier of > the same PEP value. > 3. This is the "Necessary Bandwidth" as defined by the ITU. > 4. A summary of the ITU Designation system can be found at > http://en.wikipedia.org/wiki/Types_of_radio_emissions > > 5. Double spaced mode. > 6. Default and normal calling mode. > > > Implementation details are contained in the GPL software source code > for fldigi which can be downloaded from the following site: > > http://www.w1hkj.com/fldigi-distro/fldigi-3.03.tar.gz > > This is a tar zipped format that will be familiar to all Unix, Linux, > Free BSD and OS X developers. Windows developers can unzip this type > of archive using one of several archive managers including PKZIP. > > Fldigi is open source source software that is licensed under the > General Public License, http://www.gnu.org/copyleft/gpl.html. You are > free to use the source intact, to modify, to improve and even to > incorporate into a commercial product. You must however abide by the > the license under which it has been developed and published. To date > one other amateur product has used fldigi source with great success, > DM-780, by Simon Brown. > > DominoEX-FEC and THOR differ in two ways: > > The FEC table structures in DominoEX-FEC have been manipulated in a > way that prevents the transmission of control codes. THOR uses the > same FEC table as MFSK and can transmit the full ASCII character set. > The interleave used in THOR is longer than used in DominoEX-FEC, and > it will have a slower throughput but greater immunity againts multi-path. > > Fldigi can encode and decode both DominoEX-FEC and THOR. > > 73, Dave, W1HKJ > >
