W1TAG / WD2XES / WE2XGR/3
  
John K. Andrews  
Holden, MA FN42ch  
[ w1tag@charter.net ]
42D 19M 12S North -- 71D 49M 43S West
Keyboard Modes for 600 Meters
If you want to skip ahead to the evaluations of the various modes, click here. Otherwise, read on...
In an earlier article, I reviewed a variety of modes for use at low frequencies, typically in the 160-190 kHz "Lowfer" band,
or the 135.7-137.8 kHz Amateur/Experimental band. These frequencies are characterized by high noise, inefficient transmitting
antennas, and in the case of 137 kHz, interference from Loran-C sidebands. Long distance communication by Amateur and Experimental
operators is very possible, but may require low data rates, tight frequency control and DSP techniques for reception. If you have
not already done so, I suggest reading through Low Frequency Transmission Modes for
an introduction to this sort of thing.
Amateur and Experimental operations are now taking place in the "600 Meter" band of 501-515 kHz, with various options on a
country-by-country basis. Being above 300 kHz, this band is in the "Medium Frequency" (MF) region, and a number of differences have
been noted with communication at LF. On the positive side, transmitting antenna efficiencies are better, and electrical noise levels
are lower. Thunderstorm static is present in both bands, but apparently from closer sources at MF. On the negative side, fading is
a real issue. Fades can be frequent and deep at the moderate distances used for regional communication. Long haul fading may not be
that much different than at LF.
As you would expect from the history of the 600 Meter region, CW has been the favorite mode for Amateur and Experimental operators
in this band. It's a very good choice, and skilled operators can negotiate around the fades to communicate over long distances. EU to
U.S. 2-way QSO's have already taken place in CW, as have contacts around the U.S. Much of the operation to date has been one-way
beaconing, useful in determining band conditions and evaluating equipment.
Two things have prompted the research for this article. First, an increasing amount of Amateur HF activity is being done with computer
soundcard-based programs that provide keyboard communications. Second, one of the justifications for the Amateur/Experimental licensing
at 600 Meters has been emergency communications: the premise that the band could be used on a regional basis for one-way or two-way
emergency messages in the absence of regular facilities. Computer-based setups at both ends would facilitate this, and provide
verbatim copy of messages. As the various existing modes are discussed in this article, comments will be made about their attributes
in both casual and emergency communication. Much remains to be done, but this is a start! Read on...
Bandwidth:
A plain carrier signal contains no real information; some sort of modulation process must take place for meaningful communication.
Whatever the modulation, it will increase the bandwidth of the signal. Here's a spectrum shot
of a simple carrier with no modulation. Watch what happens when that carrier is keyed on and off for
CW at 18 wpm . The bandwidth to the -20 dB points is now about 45 Hz. Note that this is happening despite some very normal
shaping of the envelope; it would be wider if faster rise/fall times were used. So,
as always, there's no free lunch. As we examine each keyboard mode, make note of the listed bandwidth, the spectral screen shot and
(if appropriate) the picture of the envelope. You may decide that some modes are too wide to be "neighborly" in a small band, or at least
should be used in an out-of-the-way corner.
Error Correction:
Think about what happens during a CW conversation in the presence of fading. If some words are missed, the receiving operator may ask
the sending operator to repeat them. Assuming that the fading doesn't happen again, the re-send will take care of the problem. This is
an example of a form of error correction that requires feedback from the receiving end. Another approach is the one frequently taken with
DX contacts: send everything important multiple times. This requires no feedback -- it is assumed that the repeats will compensate for any
fading. This is called Forward Error Correction (FEC), and is the only type we will consider in this discussion. It may be as simple as sending
everything twice (or more), or it may interleave characters or the bits that form characters in a repeatable fashion that hopefully "fakes
out" any fading or static crashes. The decoding end does the reverse of the pattern, throwing out any elements that don't fit the scheme, and
substituting repeated ones that do. As with the discussion about bandwidth, there is a price to be paid. Just as sending "Name is John John John"
to a DX operator takes extra time, FEC slows down the pace of the conversation. Since FEC would be most useful if the extra information was
sent after (as opposed to during) a disturbance, the slower modes seem to have some advantage in quick fades or static crashes.
Looking at each mode, we will include some notes about error correction, and the toll it would take on normal keyboard communication.
Speed of Copy:
The various programs used to generate these modes quote speeds in words-per-minute. We have found those speeds to be misleading. It may
just be that they are determined for some flavor of the mode and calculated or guessed for the rest. We decided to take a simple approach:
having used the text "WE2XGR/2 Burlington CT...The quick brown fox jumps over the lazy dog..." in our testing, I recorded the length of
time needed to decode that message. This should give a better feel for the relative speed of the various modes.
Linear Power Amplifier Needed?
Modes like PSK31 achieve a narrow bandwidth by careful shaping of the transmitted envelope, with phase reversals taking place at times
of minimum signal. A linear amplifier (being run in a linear fashion!) is required to preserve this shaping and tight bandwidth. Similarly,
some modes, such as THROB, transmit more than one tone at a time, and a non-linear amplifier would create mixing products of the tones.
If you wish to experiment with the majority of the modes mentioned in this article, then you will definitely want a linear PA. There are
basically three approaches for 2200 and 600 Meters:
1. A conventional Class B (or AB) RF amplifier, such as the amplifier
described by Ralph Hartwell on the 500kc.com website.
2. A mosfet-output audio amplifier, such as the Hafler 9500/9505
used by WE2XGR/2 and WE2XGR/3.
3. A non-linear amplifier that restores the input envelope, such as the
EER Transverter described by Jim Mortiz,
M0BMU.
Receiving Tips:
All of the modes described will run on computers using the various versions of Windows. Versions from 98 through XP seem to be OK, but
you may want to research Vista compatibility before heading in that direction. Sound cards of the "Sound Blaster" type and up appear to
work well. Many claim that on-board AC'97 audio is inferior for ham radio applications, but others use it without issue. We do suggest
that you do a minimum of multi-tasking when decoding these modes. DSP activity is CPU-intensive, and a number of us have experienced garbles
in received copy that coincide with email activity on the same computer. Ideally, your "QSO computer" would be a separate machine,
but you may at least want to experiment when using one machine for a number of activities.
One thing regarding receivers: There are two approaches to receiver setup for these digital modes. One is to keep the receiver bandwidth
only wide enough to pass the desired signal, with the receiver usually being in CW mode. The other is to run the receiver in USB mode,
allowing a wide audio bandwidth. That allows you to tune around on the computer screen, possibly seeing nearby signals. The disadvantage,
particularly if you run AGC, is that strong nearby signals may affect your copy of weaker ones. Experiment!
Station Setups and Test Methods
WE2XGR/2: Jay Rusgrove, W1VD, is located in Burlington, CT (FN31IS). His transmitter consists of an HP 3325A GPS-disciplined
oscillator feeding a phasing SSB exciter. His power amplifier for these tests was a Hafler 9505
audio power amp. Jay's antenna for transmitting and receiving is a 90' top-loaded Marconi.
Either an RF-590 or an Icom R75 receiver was used during the tests.
WE2XGR/3: John Andrews, W1TAG, is located in Holden, MA (FN42ch). My transmitter is a homebrew phasing exciter, which
includes a DDS synthesizer, and a Hafler 9500 audio amplifier. Power output can be adjusted from 0 to about 350 watts, and the excellent
linearity of the transmitter allows use of a calibrated attenuator between the exciter and the amplifier. Receiving is done with an Icom
R75. The antenna for both tranmit and receive is a large loop suspended from trees. An
interesting byproduct of the loop is more high-angle radiation than would result from use of a Marconi antenna. This increases the
likelihood of fading within the groundwave range, and provided some extra torture for our tests. A separate computer is normally used
for on-air work, keeping the email and other activity away from the DSP work.
Mode Specs and Results of Tests
Jay Rusgrove, W1VD/WE2XGR/2, and I ran a series of nighttime tests in November and December, 2007. Some involved 2-way QSO's, with 10 minutes
in one mode, and 10 minutes in another. Those events were well publicized, and attracted reports from as far as Europe. We are very
grateful for many listeners who patiently sent screen-shots and text grabs. We also have
done some one-way testing during the day and night hours, working through a long list of possibilities, giving each one a few minutes to
observe what happens with noise and fading. This project will continue, and it is important to carry the work into the summer, observing the
effects of the higher noise levels. The list below contains the modes we specifically evaluated at 600 Meters. You should also consider the
faster versions of Low Frequency Transmission Modes in my original article. Here
goes!
BPSK31
QPSK31
PSKFEC31
PSK63F
PSK10
PSK08
PSKAM10
WOLF
RTTY45
PACTOR1-FEC
MSK31
MFSK16
MFSK8
OLIVIA4-250
DOMINO-EX4
WSJT-JT2
Quick Summary of Tests
...And the Winners Are...
BPSK31
Author |
Peter Martinez, G3PLX |
Program(s) |
MultiPSK, Digipan, many others |
Bandwidth |
80 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
19 Seconds |
FEC? |
No |
Text Sample |
Yes |
Comments:
Most Amateur operators have heard of PSK31 by now, and it is widely used on the HF bands, particularly on 20 meters. The waveform is
very carefully shaped for minimum bandwidth, and if you compare its spectrum with
CW at 18 wpm , you'll be impressed. While amazing claims are
made for its weak-signal performance at HF, the noise and fades at 600 Meters do exact a price. PSK31 does recover quickly from missed
characters, however, and it should be good for casual operation in this band, particularly in the daytime. It does have lower case
characters, and they are decoded more quickly than upper case. You would have to be a skilled touch-typist to stay ahead of the buffer,
though much PSK31 operation on HF is done with "canned" macros.
QPSK31
Author |
Peter Martinez, G3PLX |
Program(s) |
MultiPSK, most other programs that support BPSK31 |
Bandwidth |
80 Hz |
Spectrum |
Similar to BPSK31 |
Envelope |
Similar to BPSK31 |
Linear Needed |
Yes |
Test Copy Speed |
19 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments:
QPSK31 uses quarternary phase shift keying (4 states instead of 2), with some modest error correction. The bandwidth is
about the same as BPSK31, but the throughput is reduced by the FEC. In tests at 600 Meters, it showed a little improvement
over BPSK31, but not enough to make any real difference with noise and fading. If BPSK31 doesn't work out for you, look at the
other alternatives.
PSKFEC31
Author |
Patrick Lindecker, F6CTE |
Program(s) |
MultiPSK only |
Bandwidth |
160 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
33 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments:
This is a very interesting mode, with good weak-signal capability. Like the other PSK Varicode modes, it re-syncs quickly (2 seconds) after
a signal loss, such as a deep fade. The increased bandwidth over straight BPSK31 comes from the envelope shaping: PSKFEC31 elements are
"fatter" than their BPSK31 counterparts. The faster rise and fall times are the culprits, but the longer duration of the full-power
signal makes life easier for the decoder. One downside of PSKFEC31 is the lack of a lower-case character set. You can get by with
upper-case characters for QSO's, but emergency traffic may include messages whose original formatting should be preserved. This aside,
PSKFEC31 currently has the best combination of speed and weak signal handling of any of the PSK Varicode modes.
PSK63F
Author |
Nino Porcino, IZ8BLY |
Program(s) |
MultiPSK, IZ8BLY STREAM |
Bandwidth |
140 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
19 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: PSK63F is another good weak-signal mode with reasonable bandwidth. Its threshold of clean copy is about 2 dB
worse than PSKFEC31. However, it does have the full PSK Varicode character set, giving the lower case characters missing in PSKFEC31.
Resync time is about 3 seconds, so the fade recovery is good. Typing speed is very good, similar to BPSK31, so only a very skilled typist
should spend time waiting.
PSK10
Author |
Patrick Lindecker, F6CTE |
Program(s) |
MultiPSK |
Bandwidth |
40 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
48 Seconds |
FEC? |
No |
Text Sample |
Yes |
Comments: PSK10 is a little-used mode that certainly has some applications at LF and MF. Its bandwidth is quite narrow,
but the throughput is definitely slow compared to BPSK31. There is no forward error correction, but the 10 bit/sec data rate is much more
tolerant of quick static bursts than its 31 b/s brother. Most typists will be able to stay behind the buffer...not a bad thing if
you like to think a bit before typing.
PSK08
Author |
Wolf Buescher, DL4YHF |
Program(s) |
Spectrum Laboratory |
Bandwidth |
20 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
63 Seconds |
FEC? |
No |
Text Sample |
Yes |
Comments: We have included this mode as the slowest of the non-error-corrected PSK modes. It is available only in
Spectrum Laboratory, and works quite well. The user interface with SpecLab is a bit challenging for "newbies," but once you get it running,
QSO's at these low data rates are fairly relaxing. The bandwidth requirement is very low, but unlike the PSK02 and PSK01 modes in SpecLab,
receiver tuning accuracy and sound card sampling rate are not big issues.
PSKAM10
Author |
Patrick Lindecker, F6CTE |
Program(s) |
MultiPSK |
Bandwidth |
50 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
121 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: PSKAM10 has been used successfully in "Lowfer" communication on both U.S. coasts for a number of years. Its
weak-signal capability is very good, but it has some limitations. Only upper-case characters are available, and the throughput is quite slow.
The very worst typists should be able to stay ahead of its buffer! PSKAM10's biggest shortcoming, however, is its decoder, which is slow
to sync, and has considerable hysterisis, requiring more signal to re-sync than it needed to hold copy before a fade. In the absence of
fading or large variations in noise, this mode is quite viable for casual communication.
WOLF
Author |
Stewart Nelson, KK7KA, and Wolf Buescher, DL4YHF |
Program(s) |
WOLF GUI |
Bandwidth |
40 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
Yes |
Test Copy Speed |
Very slow! |
FEC? |
Yes |
Text Sample |
Yes |
Comments: WOLF (Weak signal Operation on Low Frequencies) was the brainchild of Stewart Nelson, KK7KA. It uses BPSK
at a 10 b/s rate, but in a very different manner than the preceding modes. A 15 character message is coded, given heavy forward error
correction, and interleaved with a known pattern to form a 960 bit signal. This is then repeated over and over until one of the operators
calls a halt. The result is a very weak signal DX mode that has worked over very long distances at LF. The throughput can be glacially
slow at the weakest levels, but it is easily the equal of QRSS60 (without the required visual interpretation). The original version was
command-line only, and required the user to record and analyze .wav files. The present program by DL4YHF uses a GUI interface that
makes QSO's much easier. This mode has little application for emergency or casual communication, but might be just the thing that gets
you into the distance record books.
RTTY45 (23 Hz Shift)
Author |
Older than I am |
Program(s) |
MultiPSK, MMTTY |
Bandwidth |
60 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
15 Seconds |
FEC? |
No |
Text Sample |
Yes |
Comments: Several European operators have experimented with 45 Baud RTTY using a narrow 23 Hz shift, with decent results.
Our tests, however, had limited success when compared to the various PSK or multitone FSK modes. Given good signal strengths with little
fading, you'll have good results and will be hard-pressed to keep ahead of its typing buffer. I tried a 200 Hz shift on a closed-circuit
basis, and the weak signal performance was worse. Perhaps there is a better RTTY implementation than the one in MultiPSK -- if so, please
let me know!
PACTOR I FEC
Author |
Ulrich Strate (DF4KV) and Hans-Peter Helfert (DL6MAA) |
Program(s) |
MultiPSK, others |
Bandwidth |
600 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
No |
Test Copy Speed |
20 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: Despite the forward error correction, this is a tough mode to recommend for weak signal work. As the text
sample shows, there is a squelch level below which copy is impossible. When the signal was above that level, it was fairly well
behaved, and there were some good reports to Jay's beaconing in PACTOR I FEC. It's also a bit wide for what it does, and we would
suggest that you stick with the multi-tone FSK modes for this application.
MSK31
Author |
Wolf Buescher, DL4YHF |
Program(s) |
Spectrum Laboratory |
Bandwidth |
40 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
18 Seconds |
FEC? |
No |
Text Sample |
Yes |
Comments: Spectrum Laboratory has two MSK (minimum shift keying) modes, MSK31 and MSK08. We had real program stability
and decoding issues with MSK08, and are not reporting on it here. MSK31 appears to have weak signal, noise and fading characteristics
similar to BPSK31, which is not surprising. As with the other Spectrum Lab modes, the user interface is challenging, and at this data
rate, you have to stay focused. MSK has a lot of potential, but it will need better decoders and FEC capability to be competitive with
the multi-tone FSK modes.
MFSK16
Author |
Nino Porcino IZ8BLY and Murray Greenman ZL1BPU |
Program(s) |
MultiPSK, IZ8BLY STREAM |
Bandwidth |
316 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
20 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: MFSK16 and MFSK8 (below) are multi-tone frequency shift keyed modes, where only one tone is sent at a time. This
allows the use of a non-linear PA, and the setting of levels on a linear PA by the use of a single tone. MFSK16 uses 16 tones within a 316 Hz
bandwidth, so it is definitely wider than the PSK modes considered above. Is it too wide? You can certainly fit three signals in a 1 kHz
slot, which doesn't seem too bad. Not our call! MFSK16 is a very good weak signal mode, it's only shortcoming being a 6 second re-sync
after a loss of signal. It has a 1 dB advantage over PSKFEC31 in our steady-signal tests. It does have lower-case characters, however.
MFSK8
Author |
Nino Porcino IZ8BLY and Murray Greenman ZL1BPU |
Program(s) |
MultiPSK, IZ8BLY STREAM |
Bandwidth |
316 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
29 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: MFSK8 is similar to MFSK16, but uses 32 tones at half the data rate within the same bandwidth. It's slower
than MFSK16 as a result, but still in comfortable typing range, and a good typist will be able to stay ahead of the buffer. It has a 2 dB
weak signal advantage over MFSK16 (hence 3 dB over PSKFEC31), making it the most sensitive conversational-mode program we have tried. As
with MFSK16, there is a 6 second re-sync time after reacquisition of a signal. An interesting down-side is a tendency for the decoder
to "get lost" if the signal is lost for a while or changes frequency. This makes it difficult to use in a 3-way QSO, or to monitor a 2-way
where everybody isn't on the same exact frequency. If you are comfortable with the 316 Hz bandwidth, this is still worth a try for QSO's
or more formal communication.
OLIVIA 4-250
Author |
Pawel Jalocha, SP9VRC |
Program(s) |
MultiPSK, others |
Bandwidth |
250 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
40 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: There are a variety of Olivia modes, and we only evaluated the 4-250 version, which does FSK with four tones in
a 250 Hz bandwidth. The full character set is available, and typing speed is on the slow side, but reasonable. The weak signal capablility
is not as good as MFSK, so it did not rank near the top of our list. The OLIVIA 8-250 mode should be evaluated for better low signal
performance.
DOMINO EX-4
Author |
ZL2AFP, ZL1BPU and F6CTE |
Program(s) |
MultiPSK |
Bandwidth |
140 Hz |
Spectrum |
Yes |
Envelope |
No |
Linear Needed |
No |
Test Copy Speed |
65 Seconds |
FEC? |
Yes |
Text Sample |
Yes |
Comments: DOMINO EX4 is another multi-tone FSK mode, and offers good performance. It appears to run much more slowly than
the rated 27 wpm suggests, so that needs to be evaluated. The typing speed was definitely below comfort level for typical QSO's. The full
character set is available. Resync time after signal loss is about 7 seconds, the slowest tested. All of that being said, it only ranks
1 dB below MFSK8 in weak-signal capability, so it is a definite contender. We briefly tested DOMINO EX8, but its weak signal performance
was definitely less than EX4. The extra bandwidth would not appear to be warrented.
WSJT-JT2
Author |
Joe Taylor, K1JT |
Program(s) |
WSJT Ver. 5.9.8 |
Bandwidth |
80 Hz |
Spectrum |
Yes |
Envelope |
Yes |
Linear Needed |
N |
Test Copy Speed |
13 char/min |
FEC? |
Yes |
Text Sample |
Yes |
Comments: WSJT6's JT2 mode was a late addition to our project. It is definitely NOT a conversational mode, but is in the
same category as WOLF, being aimed at very weak-signal QSO's. JT2 works in 60 second periods, sending a 13 character message which can be
decoded within that minute. Two stations would typically transmit on alternate minutes. Our test signals were copied by two stations in
Europe, so there are some interesting possibilities for dedicated operators. The earlier WSJT modes were aimed at weak signal work at
VHF and above. JT2 is specifically targeted at lower frequencies, and is somewhat fussier about frequency stability and timing. As with
WOLF, it has no real application for anything but long-haul QSO's.
Quick Summary of Modes Tested
ModePrograms | Bandwidth | Test Copy Time | FEC | Linear required? |
BPSK31 | MultiPSK Digipan | 80 Hz | 19 Seconds | No | Yes |
QPSK31 | MultiPSK | 80 Hz | 19 Seconds | Yes | Yes |
PSKFEC31 | MultiPSK | 160 Hz | 33 Seconds | Yes | Yes |
PSK63F | MultiPSK Stream | 140 Hz | 19 Seconds | Yes | Yes |
PSK10 | MultiPSK | 40 Hz | 48 Seconds | No | Yes |
PSK08 | SpectrumLab | 20 Hz | 63 Seconds | No | Yes |
PSKAM10 | MultiPSK | 50 Hz | 121 Seconds | Yes | Yes |
WOLF | WOLF GUI | 40 Hz | Very slow! | Yes | Yes |
RTTY 45 23 Hz shift | MultiPSK MMTTY | 60 Hz | 15 Seconds | No | No |
Pactor1 FEC | MultiPSK | 600 Hz | 20 Seconds | Yes | No |
MSK31 | SpectrumLab | 40 Hz | 18 Seconds | No | No |
MFSK16 | MultiPSK Stream | 316 Hz | 20 Seconds | Yes | No |
MFSK8 | MultiPSK Stream | 316 Hz | 29 Seconds | Yes | No |
Olivia 4-250 | MultiPSK | 250 Hz | 40 Seconds | Yes | No |
Domino EX4 | MultiPSK | 140 Hz | 65 Seconds | Yes | No |
WSJT-JT2 | WWSJT Ver. 5.9.8 | 80 Hz | 13 char/min. | Yes | No |
|
...And the Winners Are...
What follows is aimed at 600 Meter operation only, and might be quite different if applied to HF. Don't read too much into it at this
point! Based strictly on weak-signal performance, the rankings would be:
1. MFSK8 (0 dB reference)
2. DOMINO EX4 (1 dB) (note slow speed, however)
3. MFSK16 (2 dB)
4. PSKFEC31 (3 dB) (upper case only)
5. PSK63F (5 dB)
The weak-signal approach certainly would be an indicator of steady day-time performance, and would give a picture of the fading depth
that a signal could endure. As noted above, recovery from signal loss is another consideration, and the PSK modes definitely resync
more quickly than the FSK versions. In particular, beware of MFSK8, which may not recover from a long signal loss or frequency shift
without manual intervention. However, since there is a 5 dB range between MFSK8 and PSK63F, sychronization might not be lost
in a fade for the former. In general, MFSK16 has the greatest general utility, and is likely to re-sync under wider range of conditions.
The catch at the moment has to do with static crashes, a frequent enemy at 600 meters. This work was all done in the northeast U.S. in
the colder months, when thunderstorm static is not normally an issue. Absent a way to accurately simulate it, we will have to redo
some of these observations in the warmer months.
Note (happily) that the leading contenders above do not require the use of linear power amplifiers! A simple low-power transverter added
to your high-band rig should provide a good signal source. Just swear on a stack of Handbooks that you will not be tempted to run the
PSK modes without using a linear amplifier. The results are even uglier than some of those mashed-up PSK31 signals on 20 meters. But,
as long as you stick to the FSK modes, you can have a lot of fun on this band.
Changes will likely be made as modes emerge and suggestions are made for alternate methods and programs. Again, thanks to all who have
helped with the testing.