Hello to all,
The recent thread on call-puzzles leads me to believe that, rather
than add additional complexity to our usual CW procedures, we consider
alternative protocols and coding methods which will permit operation
with much lower signal levels. I have some ideas on the subject, but
lack both the operating experience and the math background of many on
this reflector. This post is an attempt to get people thinking and
start a dialogue which I hope will result in a design for a really
robust system.
Before getting into the technical details, a word about philosophy. I
know that many on this list are strongly opposed to computer decoding
of EME signals, saying "There is no challenge or fun in EME if your
computer simply tells you that you have a QSO." I agree with that.
But you must realize that, in spite of any advances in DSP or other
digital techniques, a majority of attempts at EME communication will
still fail. This is because people will exploit these advantages to
operate with smaller antennas, lower power levels, and under worse
conditions, until it again becomes challenging. When your computer
tells you that you *don't* have a QSO, that's when the fun begins.
You then get to use your "wetware" to improve the protocol, coding,
filtering, modulation scheme, etc., so that a future attempt under
similar conditions will succeed! If this is not your idea of fun,
you might as well hit the delete key now.
Some hams have claimed that by implementing DSP tracking filters and
detectors, and by using advanced modulation such as multitone FSK, we
could gain up to 20 dB over present practice. IMO, this is not
possible in the foreseeable future. Not that I doubt the performance
claims for the systems proposed. The problem is that they greatly
underestimate the performance of the present system! S/N calculations
for EME are typically based on a noise bandwidth of 100 Hz. But I
believe that the ears and brains of the best operators provide about
25 Hz, a 6 dB difference. Also, good operators effectively utilize
information from multiple repetitions of a message, gaining another
few dB. And while some FSK schemes may be about 3 dB better than OOK
(On Off Keying, the modulation used in CW), if you have a tube HPA
which is thermally limited, and cut power in half to run 100% duty,
then your 3 dB is gone. In all, I believe that no more than 10 dB
improvement is practical.
But, if we want to routinely communicate between two single Yagi
stations running brick amps, or between an OSCAR station and a typical
four Yagi setup, we need to gain about 15 dB. From where will the
extra 5 dB come? The answer is easy: we need better source coding,
channel coding, line coding, and protocols.
Using Morse code over an EME circuit has several problems. Morse
achieves excellent compression of normal English text, by using the
shortest codes for letters which occur most frequently. But except
for a few vanity calls, the letters in our callsigns are essentially
random, thwarting compression. Worse, as Juergen points out, the
variable length code makes it difficult to assemble received fragments
into a complete message. There is also little or no margin for error.
Suppose that the call W5ETN became active on the moon. Most of us
would mistake it for W5UN; the difference is only one dit. There is
an asymmetry in receiving CW under faded conditions that makes
distinguishing these calls especially difficult. If a weak message is
transmitted repeatedly, and you hear a distinct dit during a QSB peak,
you can be nearly certain that the key was down at that instant. But,
on average, it takes many more repetitions to be as sure that the key
was *up*, because it is not unlikely that a given bit is lost several
times in the fading. This effect can cost us up to 2 dB.
Guenter Koellner wrote:
Don't take this following words too serious!!!::: The only thing
that really works is very slow speed CW, allowing the computer to
decode the signal with FFTDSP, Hamview or anything similar using
FFT. The VLF people do so and I did on 13cm
DL4MEA, DSP and More
when I identified the first man-made signal on my 13cm.
We should take those words very seriously! IMO, it's EME's future.
We can further improve the robustness of very slow OOK by eliminating
the asymmetry. One crude way is to take two seconds to send each bit,
using Manchester coding: A zero is sent by having the key down during
an even second, and up during the following odd second; the opposite
is done to send a one. In a one minute period, allowing time for
propagation and T/R switching, we can send 28 of these bits. How can
we best utilize these bits to get our message across? Conventional
wisdom is to first compress the message, then add error correction
information. The problem is that the techniques used commercially are
all optimized for sending long messages over moderately impaired
channels, and would be useless or perform very poorly in an EME
situation, so codes especially suited to the application must be
developed. Please let me know of any which you feel are suitable.
Using Morse code, my callsign requires 60 bits; I believe that is
about average. How can we reduce that? Simplest would be to use a
six bit code for each character, e.g. shifted ASCII; 36 bits would
allow for callsigns up to 6 characters. If we limit our symbol set to
40 (letters, digits, blank, and no more than 3 punctuation), we could
code 3 symbols in 16 bits (40 ^ 3 < 2 ^ 16); a six character call
would now fit in 32 bits. For an application specific approach, we
could make a database of actual callsigns, and identify one by
transmitting the position where it appears in the list. If we allow
for up to 4 million hams, a 22 bit number would suffice. Still more
aggressively, we could limit the database to EME operators. Allowing
for, say, 3500 entries, we could use a 12 bit code and still reserve
several hundred values for control functions; use of such a code is
assumed in the examples below.
How much "gain" can we get from compression? If we are using
Manchester coded OOK with a bit rate which is slow relative to the
effective bandwidth, this is fairly easy to answer. Let's say we
compress by a factor of two, thereby doubling the time available to
send each bit. By integrating twice as long, the received signal
power is doubled, but the noise power only increases by the square
root of two. So we can reduce our transmit power by 1.5 dB for the
same bit error rate (BER). The actual gain can be somewhat greater
(the shorter message can tolerate a slightly higher BER for a given
probability of being correct) or somewhat smaller (ECC is not as
efficient on a shorter message). However, both of these effects are
quite small, so a value in dB of 5 * log(compression) is a decent
estimate.
How should we expand our 12 bit message into the 28 bits available, so
that the least power is needed for reliable transmission? I don't
have the best answer, or even a good one, but will suggest a code for
a first try. There are some interesting aspects of coding short
messages. The usual distinction of "block" and "convolutional" codes
is not meaningful. You are simply mapping a short string of symbols
into a longer one. You can easily encode a message using *any* code,
using a table lookup. More importantly, a maximum likelihood decode
is possible, for arbitrary codes, by simply comparing the received
data with all possible transmissions, and choosing the best match.
This allows "soft decisions" (where weak bits are given less weight
than those strongly received) to be used with any code. Also, you can
get any desired reliability (at the expense of power, of course), by
simply requiring that the best match exceed the second best by a
specified amount. On an early planetary mission, NASA employed a
Golay code which used 24 bits to encode each 12 bit pixel. Without
soft decisions, this code can correct all single, double, and triple
errors, and detect all quadruple errors. I don't think that with 28
bits it is possible to correct all quadruple errors, but by adding 4
more parity bits to the Golay code, you can fix 95% of them, and aid
the soft decision process considerably.
How much gain can we get by coding? Commercial systems now get 7 dB
or more, and some newer deep space probes come within one dB of the
Shannon limit (in theory, one could send perfect data with an Eb/No of
only -1.6 dB). Unfortunately, short messages cannot be coded as
efficiently as long ones, and it is harder to make badly corrupted
data right most of the time, than it is to make pretty good data
nearly perfect. My math is too weak to come up with a hard number,
but my feeling is that we have to settle for something around 3 dB.
The suggested code is a little better than 2 dB. Of course, in EME,
even that is nothing to sneeze at.
We now have a means for sending packets, fairly reliably, from one
station to another. In order to have a QSO, we need a suitable
protocol. Here, too, it appears that we must invent one, specially
designed for EME. That's because commercial protocols, although
they do an excellent job recovering from lost, garbled, duplicated,
and out-of-order packets, assume that all "good" packets are certain
to be correct. The headers and CRC checking which accomplish this are
a luxury we cannot afford. The overhead would be much larger than the
message we are trying to send, and would increase the power needed by
several dB. So a system is needed which will (most of the time)
recover from apparently good packets which are actually incorrect.
A good protocol for EME should do the following:
1. Allow successful completion of a QSO in a specified time, e.g. 1/2
hour, with the smallest possible signal.
2. Minimize the probability of a "false" QSO (one or both callsigns
incorrectly received, but it appears to be complete).
3. Minimize the chance that one party thinks that a QSO is complete,
but the other does not.
4. When signals are strong, complete the QSO as rapidly as possible.
5. Allow for transmission of arbitrary additional data, e.g. grid
square, special conditions.
Here is a "brute force" `protocol which might be an adequate first
try, although it is quite poor on (4).
Assume that we use our 12 bit message format, with a few of the
possible messages used for control functions, and the remainder used
for call signs. The control codes used represent CQ, QRZ, O, RO, and
R. This protocol always operates in "random" mode; a sked simply
means that someone will be listening on frequency, awaiting the CQ.
Here's how it would go under ideal conditions:
1. (A->B) CQ No variable information is sent in the first packet,
allowing the receiving station to search both
frequency and timing for the signal. Once found,
station B adjusts frequency and clock to match A.
2. (B->A) QRZ Here, we simply acknowledge the CQ and invite A to send
his call sign. A more advanced version could include a
crude signal and capabilities report.
3. (A->B) CA A, hearing the response, sends his call sign.
4. (B->A) CA B, reasonably confident about receiving the call, echos
it back.
5. (A->B) O A, having received his own call correctly, is now
highly confident that B has it right.
6. (B->A) CB B, hearing the acknowledgement, sends his call sign.
7. (A->B) CB A, reasonably confident about receiving the call, echos
it back.
8. (B->A) RO B, having received his own call correctly, is now
highly confident that both calls have been properly
received by both parties.
9. (A->B) R A, having received RO, has a complete QSO. He sends R
to B to indicate completion.
Sending a signal report in step two could be quite useful. Only a
couple of bits are needed (e.g. very poor, poor, fair, good). If the
station calling CQ has polarity switching, he could send a second CQ
in the other polarity, even though a reply was heard the first time.
He would then know which polarity was received better, and use that
for the remainder of the QSO.
One advantage of a simple protocol is that it facilitates "memory
ARQ". If you receive an unintelligible packet, you do not know if it
is has new information, or if it is a repeat (your last transmission
may also have been garbled). But, when you respond negatively, you
can be quite sure that the next packet you get will be identical. If
that is also unreadable, you can combine the two, bit for bit, to
improve the S/N. Using four copies should gain you 3 dB on that
packet, at the cost of an extra 6 minutes.
This protocol could never win a contest (if the rules were changed to
permit other than CW and SSB!), because even under ideal conditions,
it takes 9 minutes per QSO, longer than that *averaged* by the contest
winner, while a minimum CW QSO can be completed in only 5 minutes. I
hope that a protocol actually selected for routine QRP EME use would
not have this shortcoming, and would actually be tried in contests. A
four Yagi station could win!
Finally, although the system presented above is probably workable,
it's far from the best. I hope that you will respond by suggesting
better source coding, better ECC schemes, better protocols, etc. Then
we can construct a program, similar in concept to PSK31, which could
run on most any PC and would work with most any rig. EME contacts
between small single Yagi stations could then be commonplace.
Thanks for taking the time to read this rambling post.
73,
Stewart Nelson KK7KA
KK7KA
Uhmmm,
Every time someone posts something innovative on this reflector , its overwhelmed by
contrary or hilarius comments.
EME is absolutely nomore a top edge HAM activity, but a place for extremely conservative
high end plumbers?
This attitude annoyed me so much, and make me feel like the wheel barrow guy was right...
Go on Stewart, you are doing a really fine job
Time will tell who is right and who is wrong, I am afraid that like dinosaurs the BIG-GUN
arrays will be a happy reminder of the glorious analogic days.
73s Andrea, IK5QLO
Hi all,
I still have some comments to the newest discussion on this subject.
1. I do agree with all these stations, who think, EME on amateur radio
will loose a little bit of what it is, if you put too much computer-
help into decoding the signals, just relying on your dsp, that the
qso has taken place without ever having heard a single character with
your own ears.
2. I do agree with these stations, who think that the actually used
procedure on EME-skeds is not the most optimized.
Every station, even a "BIG-Gun" uses a continuous repeating procedure
of his informations. Why ?
The only reason I can imagine on this repeating is to add the n
repeat-cycles in your brain (with the help of paper and pencil, I think
this might be allowed) to make a complete callsign from 10-15 fragments.
So if this can be accepted by everybody, why not optimize the system for
the repeating a little bit, what makes it easyer to add the fragments ?
3. I do agree with those who said, to make the procedure not too complicated.
But I think, it should be allowed or everybody to try to change their
informations by procedures, computer-help-systems, however they like.
Amateur-Radio is (in some edges) still experimental hobby, what should
give everybody the chance to such (technical interesting) things, and
tell others of their results, without becoming smiled by the "serious"
stations, who do not use such "doubtful" helps.
Any experimenting is ok in my opinion and should be respected by
everybody.
But, back to the difficult-level of procedures, such things should not
become standard operations.
4. There are already severals different procedures on making a "QSO"
Most HAMs that I know, do not have any agreement on Sked-procedures
(MS, EME) at all, because they cannot understand that this exchanging
a call, some ooo, roo and rrr has anything to do with making a "QSO".
What many of them can hardly accept, is working MS or EME by Random.
But the difference between working Random and working Skeds is
magnificent in their eyes (in mine too). I think the magnificent
difference is, how shure you can be that a the contact was really
complete.
On MS, I know an abbreviation what says MS means "maximum swindle",
same think many people on Skeds on EME.
If someone likes, it is possible to make a "complete QSO", without
having copied even a single letter. If you only hear anything in the
noise, just send some periods "ro" and wait what will happen. Perhaps you
will receive a qsl-card.... that is, why ...
...I agree with all those people, who do not like skeds (me too).
...I absolutly prefere working random.
But...
...I now worked about 400 EME-QSOs with my (only) 4 yagis, from these only
10 skeds. After now abt 100 initials on 70cm, I notice, that there are
many stations with really nice equipment, which I sometimes hear in
QSOs with really excellent signals, but I never heard them calling CQ.
So what can I do to work them, beside making a sked ???
5. Now my opinion is to increase the efficience especially for random-
contacts, without getting too much new technics into it, just keeping
all by the force of my ears and my brain.
A simple thing to optimize is to syncronize the transmitting of the own
call with the time, so that you can avoid the call-puzzle and bring the
letters easyly into the right order.
I do not agree with anybody, who thinks, that this is too comlicated.
I do not agree with anybody, who says that this is not practicable.
I do not agree with anybody, who says that this might not be helpful
by detecting calls in week signals.
I do not agree with those, who reject any new ideas only the keep the
established things like the were all the time.
Such a small complement to the EME-procedure should nobody hurt,
nobody confuse, not cost any penny, but give smaller stations the
chance for some more random-qsos e.g. make difficult contacts
shorter. This should not become the standard procedure, but used
(only) if you do not get the call complete with the standard procedure.
6. I'm now licensed for 28 years and often find, that especially
these hams who are on the air for a long time (wouldn't say "old-
timers") don't like to change old and established procedures.
My sons (13 and 15) doubt on nearly anything, and think about making
everything new and better. It's sometimes not easy to convince them
of the establihed things. But I must say, that sometimes I found out
they were absolutly right in their doubt.
Sometimes it's only a small improvement what gives us a magnificent
progress.
So give all new ideas a real chance.
If you don't do, there will be no progress anymore.
Always try the "impossible" things again and again, until you have
success finally. Better to try a transatlantik-contact on 2m your
whole life without success, smiled by all those sceptics, than never
having tried it....
Don't smile on anybody, who wants to improve something.
Don't shout on anybody, who uses this reflector for his ideas. Nobody
is forced do read it. Everybody is alowed to delete emails before
reading them.
Thanks for taking your time to read this email.
wish you all the best 73 and hope to cu on the moon,
Juergen,DL8OBU
Hi all,
I think Stewart comes close to an important thing here:
> Some hams have claimed that by implementing DSP tracking filters and
> detectors, and by using advanced modulation such as multitone FSK, we
> could gain up to 20 dB over present practice. IMO, this is not
> possible in the foreseeable future. Not that I doubt the performance
> claims for the systems proposed. The problem is that they greatly
> underestimate the performance of the present system! S/N calculations
> for EME are typically based on a noise bandwidth of 100 Hz. But I
> believe that the ears and brains of the best operators provide about
> 25 Hz, a 6 dB difference. Also, good operators effectively utilize
> information from multiple repetitions of a message, gaining another
> few dB. And while some FSK schemes may be about 3 dB better than OOK
> (On Off Keying, the modulation used in CW), if you have a tube HPA
> which is thermally limited, and cut power in half to run 100% duty,
> then your 3 dB is gone. In all, I believe that no more than 10 dB
> improvement is practical.
Doing better than CW is VERY difficult!
The big problem in EME is the signal path. We have multi path
propagation that causes phase and amplitude modulation. The fundamental
thing for gaining a lower signal threshold is bandwidth.
There are two bandwidths in the receiver; pre- and post- detector
bandwidth. There are also two bandwidths for the signal; keying
bandwidth (KB) and path modulation bandwidth(PB).
There is NOT any fundamental difference between normal CW keying and
what the computer can do.
Commercial systems look for low error rate at maximum information speed
(as far as I know). As a contrast amateurs look for very low information
speed with a high error rate to convey just a small amount of
information.
As long as KB is much greater than PB it is possible to gain S/N by
bandwidth reduction. We can easily use coherent CW to get the same
advantage as the optimum digital system. No post detector filter is
used. 10 times lower bandwidth allows 10 times less tx power at 10 times
lower information speed.
When KB is approximately equal to PB it is not possible to transmit any
information at low signal levels. This is because the receiver can not
distinguish between the information intentionally modulated onto the
carrier (on of keying, phase shift, whatever) and the modulation
produced by the transmission path.
DSP or human mind, the problem is the same. (Practically we know that
keying too slow does degrade readability. The DSP has the analogous
problem)
If we want to make any improvement above the present state of the art
which at 144MHz is KB between 15 and 20 Hz (and PB perhaps 4Hz) we have
to look for some post detector filtering. This can be done in several
different ways and they all have the potential of improving S/N by the
square root of the pre- to post- detector bandwidths. Common to all
these methods is that the phase information is lost which is a 3dB loss
compared to coherent cw. On the other hand the pre detector filter can
me made to match PB which is approximately a 6dB improvement (on
144MHz).
Further there is a loss of about 7 dB because when post detector filters
are used with bandwidths well below PB the detect decisions have to be
based on the average signal level while the peak signal level can be
used when KB is well above PB. (Typically I can copy two or three
characters on the best QSB peak every second 1 minute period in a random
QSO)
Summing up: -3dB +6dB -7dB is 4dB or 2.5 times so we need 6 times
narrower post detector filter compared to the 4Hz assumed for PB. In
other words: To be compatible to todays CW, a system using average
signals must use a bit length in order of 1.5 seconds.
The good news is that the post detector system has a nice signal
threshold. Once the average signal is strong enough for a few bits to be
decoded properly, all the bits will be at very nearly the same S/N so
the communication easily becomes error free. (In contrast to normal CW
for which the QSB often goes nearly through zero causing frequent loss
of information even at power levels 10 to 20 dB above minimum for a QSO)
For two oscar class stations (100W, one yagi) we need to improve by
about 6dB (Compare to how I can hear such a station with 4 yagis)
4 YAGI Signals
For 6dB improvement a data bit has to be 16 times 1.5 seconds or so. Of
course the information rate can be improved by use of more frequencies
but any such scheme will easily become useless when several stations try
to work random.
***********************
The above text is intended to show that conventional CW keying together
with sophisticated detection methods (A well trained human brain or a
DSP with coherent CW and a slightly lazier operator) is close to the
optimum solution on 144MHz. This is because our brains happen to match
the optimum bandwidth that fits with the modulation caused by the signal
path.
Any averaging process such as post detector filtering or averaging of
several repetitions of the same message implies the use of average power
without phase information and leads to very low information rates.
Stewart wrote:
> We can further improve the robustness of very slow OOK by eliminating
> the asymmetry. One crude way is to take two seconds to send each bit,
> using Manchester coding: A zero is sent by having the key down during
> an even second, and up during the following odd second; the opposite
> is done to send a one. In a one minute period, allowing time for
> propagation and T/R switching, we can send 28 of these bits.
This is too optimistic. To get the 6dB we look for, the data rate has
to be in the order of one bit per minute. Certainly possible but it
will not be very exciting. Anyone having a few dB more power would
like to use a higher bit rate. This kind of mode I think would be
useful for skeds only, not for random!!
Of course my assumptions above may be somewhat pessimistic but they are
certainly more realistic than any assumptions leading to a possible
reduction of tx powers by 20 dB!
************************
There is another very attractive way to improve EME communication.
Assuming good enough frequency stability (which already many stations
have) we could make an agreement on the transmission length. It could be
for example 18 seconds. Each minute would be 3 identical transmissions
and 6 seconds for rx to tx switching.
Each transmission would be in normal morse code. By adjusting both the
amount of repetition and the keying speed it will be possible to make
the message fit exactly to the agreed time.
When the signal is strong we just listen like we do now - nothing really
changed!
If the signal is close to the present limit we will gain a lot of time
because the fractions of letters we detect now and then can be combined,
each at it's correct place within the agreed time frame. (If the bits do
not fit we know that the station in question uses the old cw coding.)
We may gain a few more dB if the average 18 second signal is built from
those parts of the signal for which the coherent detect procedure was
successful - even though it did not produce a bit pattern. I would not
be surprised if a 6dB improvement could be obtained with about 10
periods (30 sequences to form the average from)
************************
Above I have assumed coherent cw to be used. It surely gives a 3dB
improvement when the carrier can be established but it is associated
with a detect threshold and that is not as low as to allow 3dB lower tx
power. The main advantage of coherent cw is that it is less tiring by
improving S/N above the threshold. The advantage in terms of required
tx power is smaller than 3dB. For a really good operator using his
memory to compare several sequences (automatically) it may not exist
or even be a disadvantage.
When KB/PB is very large, as in high speed meteor scatter, coherent cw
works perfectly and averaging of N sequences will gain a factor of N
rather than square root of N as we get on EME. (where phase is lost)
For EME the computer is a good tool. It makes things easier but it does
not allow communication with much less power so those of you who dislike
using DSP may well obtain equally good results just by putting in a
little more operating skill and by making sure everything works
properly. (And X-yagis do help if fully used!)
This was a long one! (for those of you reading this far)
Any comments ?
73
Leif/SM5BSZ
Hello to all,
A number of the (mostly negative) replies to my original post on this
subject used sports analogies, so I will continue in this vein.
CW operation is like running - it's done mostly for fun, sometimes
competitively. EME is the Marathon. I am not a good runner, so I am
trying to build a bicycle. Not knowing much about bikes, I outlined
what my first one would look like, hoping to receive suggestions for
superior designs.
Instead, most folks told me to continue running, many explained why
cycling is not a satisfying sport, one chastised me for failing to
attempt the Marathon, and one even suggested that bicycles do not
belong on the course!
John PE1OGF wrote:
> If you never heard him he must be very weak..... or using
> a strange type of code
Actually, both are true. My QTH has severe noise, so operation so far
has involved taking my rig to a quiet place, and running with battery
power and one Yagi. A "strange" code is needed to get good echos
under these conditions.
David Blaschke wrote:
> I notice you (KK7KA) are not in my eme log.
We each have our own esthetic sense. In my mind, the technical
challenge of ham radio, as opposed to say, the Internet or cell
phones, is that we are using our own gear, rather than depending on a
large infrastructure. Since Dave has his own large infrastructure,
I would not consider it remarkable if he could hear me, and so did
not make that effort, instead putting energy into a system which I
hope will eventually permit contact with hundreds of stations.
A number of hams said that EME is only interesting if the signal
can be decoded without a computer, in real time, and in your head.
Without a computer? Nearly all modern radios have one or more
embedded processors. So do most keyers, antenna positioners, etc.
I think it is a safe bet that 98% of EME setups involve at least one
microcomputer. What strange logic says that it is ok to involve
machines programmed by others, but not home-brew software? If
anything, the opposite should be true.
In real time? Everyone likes responsive systems. But EME is not that
way. In addition to the obvious 2+ second propagation delay, our one
minute sequences mean that you can't get a response to a message in
less than about two minutes. None of my suggested detection and
coding schemes would add more than 100 milliseconds, insignificant by
comparison. And, when reception is good, a message can be decoded
before all of it has arrived, just like with repetitive CW. But more
importantly, by providing a more robust path, it will often save many
minutes, perhaps hours, of mindless retransmission which would be
needed if CW were used. IMO, a principal advantage of this proposal
is operation closer to real time!
In your head? I can't argue with this one - it's just a matter of
philosophy. So let me ignore bicycles for a moment, and I will try to
describe a better pair of racing flats. The 12 bit source coding
scheme can easily be done manually, by having a printed list of
callsigns and corresponding codes. Detection by ear or eye is also
easy - every two seconds you decide is signal was present in the
even second "0", the odd second "1", or neither "?".
The problem is ECC; the proposed code is too complex to decipher in
one's head. But ECC is an important part of all modern systems,
including PACTOR and PSK31, and could help EME even more. Even a
very simple code can be quite powerful, especially when bits
which are likely to be incorrect can be identified by their
weakness.
Assume we are sending a three bit message. Naively, we could
improve reliability by repetition, like we do with CW. If our message
is 010, we might transmit 010010010. This "code" can correct one
error - if we receive 010011010, we can recover the original data
by "majority rules". But if we get 010011011, the wrong decision will
be made. Similarly, we can handle two erasures; reception of
01?01001? will be fine, but if we get 01?01?01?, then we do not know
what was sent.
A better way is to code as follows:
000 -> 0000000
001 -> 0010111
010 -> 0101011
011 -> 0111100
100 -> 1001101
101 -> 1011010
110 -> 1100110
111 -> 1110001
This code is not only shorter, allowing us to put 9/7 times the energy
into each bit, but it is also more powerful. We can correct all
messages containing one error, one error and one erasure, or up to
three erasures. For example ??01?11 can be correctly deciphered as
010, because it only matches one entry in the above table. Even when
four bits are missing, we can get the message correct 90% of the time.
Now, to complete our coding system, we divide the 12 bit source coded
message into four groups of 3 bits (easily done by simply using four
digit octal numbers in the table). Each digit is expanded into 7
bits as above - the whole message can be sent in one minute.
I believe that the resulting system can operate with a power level
6 to 8 dB below that needed for CW.
Comments and further flames welcome,
Stewart Nelson, KK7KA
Hello Stewart,
Keep it up and don't be discouraged! Some of us mortals with limited
systems and resources need all the help we can get!
GL and 73, Lance, W7GJ
--
Lance Collister, W7GJ (ex: WA3GPL, WA1JXN, WA1JXN/C6A, ZF2OC/ZF8)
P.O. Box 73
Frenchtown, MT 59834 USA
QTH: DN27UB 2m WAS # 44 (all EME) 6m WAS # 1039
TEL: (406) 626-5728 FAX:(406) 728-6320
W7GJ's WEB Page
At 08:25 AM 1999-06-19 -0700, Stewart Nelson wrote:
> A number of hams said that EME is only interesting if the signal
> can be decoded without a computer, in real time, and in your head.
>
> Without a computer? Nearly all modern radios have one or more
> embedded processors. So do most keyers, antenna positioners, etc.
> I think it is a safe bet that 98% of EME setups involve at least one
> microcomputer. What strange logic says that it is ok to involve
> machines programmed by others, but not home-brew software? If
> anything, the opposite should be true.
You did not read carefully enough: they said DECODED without a computer.
DECODING and having your station AIDED by computers are two entirely
different animals. If you really want to get strict about the definition of
"computer power", you could consider the early Hallicrafters TO Keyer,
which used tubes (remember them??) and had digital logic circuits to
generate CW elements, to be a computer. Would you have everyone who objects
to being "aided" by computers in their ham shack go back to straight keys
and bugs?
Some do feel that way; but most of us have NO PROBLEM with having the
MUNDANE station tasks AIDED by a computer. To us, MUNDANE station tasks are
those which we could just as easily perform ourselves. MUNDANE does not
define how complex the algorithm has to be to let a computer perform those
previously-simple analog tasks. We DO, however, object to having the
computer do ALL the work, which is unquestionably what automatic DECODING
amounts to. We might as well use packet or PSK31. PSK31 has been out now
for over a year and it's still not certain whether anyone has bothered to
complete an EME QSO using it; most likely, because there's no real
challenge in using PSK31. Just let your US$2000 computer do the hearing.
Where's the fun in that? There is no doubt in anyone's mind that EME can be
performed using PSK31 under the right circumstances; but nobody is jumping
at it, and the simple reason is because PSK31 comes TOO close to total
operation of the station. You really need to understand the difference
before you can understand why many hams actually enjoy doing many things
the "hard" way!
Look at all those microprocessors in another light: their end task. What IS
the purpose of those chips of silicon? To AID us in performing what, to all
of us, is the most DIFFICULT task of all: the copy of the signal. Is that
simple enough?
Even using a computer to perform high-speed-meteor-scatter is not quite in
the same category; the computer is simply doing the same thing that a
75-year-old wire recorder could do, except with a lot less hassle and
making it a LOT more fun, by recording and then playing back, at slower
speed, the recorded CW for one to decode by ear; i.e., WE still have to
perform the ultimate "decoding". You still wind up sweating each and every
sked, each and every meteor reflection. But now, with a computer running a
modern program such as the newly-released WinMSDSP 2000, one can actually
concentrate on COPYING the signal. There's a difference, believe it or not.
There are, in fact, believe it or not, a FEW of us who can copy CW during a
MS sked at speeds that would take the breath of many other hams away. We
choose to go even faster and use recorders of one ilk or the other,
however, simply because the computer allows us to do even more of what we
really want to do, COPY signals; the computer actually AIDS us to sweat. WE
sweat, not the computer or recorder. Where's the fun in watching letters
appear, as if by magic, on a computer screen when you point your antler
into the sky? The computer screen is filling up with decoded letters and
callsigns?? How interesting. What's for dinner tonight?
> IMO, a principal advantage of this proposal
> is operation closer to real time!
I, personally, have no objection to realtime or delayed-time copying of a
signal; high-speed-meteor-scatter, whether performed by computer or
recorder, is another example of mandatory delayed-time copying. But I
INSIST that *I* do the copying. I *know* my computer can do it (in fact,
the new WinMSDSP 2000 actually contains decoding algorithms!!); what *I*
enjoy about operating my radios, however, is being able to copy a signal
MYSELF. My computer has the capability of taking that minor but very, very
important and unique capability of mine away from me. My computer has the
ability to DESTROY my enjoyment of copying signals. I choose to keep my
computer in its place... my SLAVE, if you will... by denying it the chance
to "enhance" my ability to communicate even more efficiently than I ever
could by myself. I could really care less just how many, and how weak,
signals my *computer* can copy. What interests *me* is just how many, and
how weak, signals *I* can copy, my computer be danged. It may *assist* me:
but it had better not be doing the actual *copying*, or it's going to be
rebooted until the cows come home!
If you wish to perform EME by letting your computer do the copying, then go
to it! If others like your method, they will adopt it. The marketplace
decides the product's value, NOT the seller. There are those who don't
believe in performing high-speed-meteor-scatter by using a computer for
recording and then playing back at slower speeds; so be it. The majority of
the folks herein on this EME reflector are telling you that they don't like
the idea of having the very last "sweat" task taken from them by a piece of
silicon. So be it. You know darn well that some day in the future, EME WILL
be commonly accomplished by computers; but the folks that set them up are
probably not going to be us, and it's probably not going to become popular
within a few years. Don't forget: you're talking to a bunch of old farts
with a few youngsters also in the audience. Your marketplace should be the
youngsters who haven't already become "contaminated" with our ancient ways
of doing things and our "strange" and "outmoded" ideas of "entertainment".
If your computerized copying methods become popular amongst the youngsters,
then as we old farts die out, the youngsters will eventually take over. So
be it: that's evolution. It's clear that's where the human race is going
since it seems to be human nature to find ways to reduce our sweat-factor.
But recognize that it is almost always the YOUNGER generation that comes up
with those new ideas, that actually performs the labor and emits the sweat
to bring new technology to fruition. Most old farts do wind up
incorporating much new technology in their daily lives; but we AREN'T
necessarily the ones who really WANT the stuff! We're set in our ways;
throughout our lifetimes, we've developed a way of living that has become
comfortable to us. We live a familiar routine upon which we've come to
depend each day. Most of us don't care to have our daily routines
continually disrupted and changing from day to day; most of us feel just
fine, thank you, doing things as we've done them for years. That's human
nature, and of course, there *are* exceptions.
The idea is not to get upset if others don't like your ideas. If you
believe they have value and initially nobody else does, then it's up to YOU
to prove your ideas. Nobody is going to do or use something in which they
have no confidence, no matter how glorious or colorful the advertising
campaign. I, personally, have no confidence that reading Jose's or Tim's or
Dietmar's reflected signals off the moon on my computer screen will give me
any greater sense of enjoyment and accomplishment than I can already
achieve by copying their signals with mine own two ears.
On the other hand, I DO know that if I have to sweat all night to hear W5UN
or KB8RQ working everybody else through power line noise and microprocessor
birdies that should, if the FCC was really on the ball, be totally illegal
and would raise the hackles of any concientous EMI-freak, that at the end
of the night, I WILL feel a sense of accomplishment that at least I was
able to not only hear Dave, but to COPY him, too, even if he never heard me
calling. And I done it myself. If I can use a DSP filter to eliminate all
that QRN, then clearly, I will enjoy that night's "sweat" even more, simply
because the filter makes it EASIER for me to copy, NOT because the DSP
filter or program algorithm can perform the copying by itself, with little
or no intervention on my part.
> The problem is ECC; the proposed code is too complex to decipher in
> one's head. But ECC is an important part of all modern systems,
> including PACTOR and PSK31, and could help EME even more. Even a
> very simple code can be quite powerful, especially when bits
> which are likely to be incorrect can be identified by their
> weakness.
If it's anything like CW in that hearing and recognition of bit patterns by
the human ear and brain is required and can be achieved by the average
human, then it should, eventually, be recognized as nothing more than CW
performed with a different pattern; and it may even eventually gain some
small degree of popularity. But the more complex it is, and the more
technology it requires, the steeper the acceptance curve. That's just the
way of the marketplace. Say, for example, that in order to use our VCRs
just to watch realtime TV and/or play movie tapes, folks actually HAD to
learn how to program the VCR itself BEFORE we could even get it turned on:
you think the VCR would be as common a household item as it is?
I have a feeling that none of this windy verbage has done anything to help
you understand the objections to your proposed methods; if so, so be it. If
not, so be it. We really don't give a dang; if we want to do it your way,
we will. If not, we won't, and you have our permission to call us old farts
and be done with it. So be it.
73, Steve, K0XP
Just stop all these useless discussions, switch off your
computers and switch on your stations and come on the band !!
R.
DF6NA
DF6NA's WEB Page
I agree! What's else is there? What's left? Some seem to be headed
into an area of turning on the radio- go fishing-come back check the
screen/printer to see what you've
worked. please preserve the human factor. Cell phones are boring.
> This probably won't be popular; however I strongly feel that if
> you do not copy a signal in real-time in your head it is not a
> valid QSO in my book...
Good DX DE Sam, WA7TDU CN92
i3dli
> David Blaschke wrote:
>> I notice you (KK7KA) are not in my eme log.
>
> We each have our own esthetic sense. In my mind, the technical
> challenge of ham radio, as opposed to say, the Internet or cell
> phones, is that we are using our own gear, rather than depending on a
> large infrastructure. Since Dave has his own large infrastructure,
> I would not consider it remarkable if he could hear me, and so did
> not make that effort, instead putting energy into a system which I
> hope will eventually permit contact with hundreds of stations.
>
> Comments and further flames welcome,
>
> Stewart Nelson KK7KA
>
Hi all:
I believe Dave's "large infrastructure" is a system to permit
other minor stations to make an EME QSO their own, using
only one frequency at the time.
AND the "large infrastructure" is also him "own gear", I think.
On the other side, digital modes, along with their algorithms,
are another way to explore about the best mode to extract
informations from the EME path IN THE RESPECT
of the fundamental rules for radio amateurs.
Among others, Leif SM5BSZ and Jan G3SEK pointed out,
to me, the entire question:
- Leif demonstrate by math & him own impressive experimental
work what one can expect by the "entire digital" and "digital aided" small
signal detection technologies.
- Jan substantially declarate, by sport analogies, the ham must
be the ultimate ring of the whole EME path chain. Otherwise
is'nt ham-radio anymore.
In effect, by comparison, I thougt to the chess games nowadays: it's
a fashinating thing to see the uman being in figthing against
a computer. A computer against another computer maybe
even more astonishing....but only for the unknown programmers.
It's still a "chess game" ?
Would be that in the future chess games will be played only
in the more efficient way to mate the king ?
Back to the topic. I believe we must proceed to enhance
our cabability in making EME QSOs using up-to-date
technologies. The limit, uncertain, lying in a pursued
situation where the natural sensing system
(ears-brain) will be the final discriminator for the QSO.
73s
Paolo, I3DLI
Stewart Nelson wrote:
> A number of the (mostly negative) replies to my original post on this
> subject used sports analogies, so I will continue in this vein.
> CW operation is like running - it's done mostly for fun, sometimes
> competitively. EME is the Marathon. I am not a good runner, so I am
> trying to build a bicycle. Not knowing much about bikes, I outlined
> what my first one would look like, hoping to receive suggestions for
> superior designs.
>
OK, it's a good analogy to use, so let's pursue it.
> Instead, most folks told me to continue running, many explained why
> cycling is not a satisfying sport, one chastised me for failing to
> attempt the Marathon,
You have a right to decide what you - individually - find enjoyable
about amateur radio. I'd support that, and defend it against people
trying to make you enjoy it their way. But...
> and one even suggested that bicycles do not
> belong on the course!
>
In a sense, they don't - at least, they don't belong on the same course
that the runners are using.
Nobody "owns" the EME path, but we do have to share the amateur bands.
These so-called ultra-weak-signal modes still involve very strong TX
signals in the immediate locality (read: anything up to 50-100 miles
radius). What worries me even more is the fact that these signals will
be controled by computers that take no account of the other band users.
If the cyclists want to enjoy their own sport, they must find a
different course that won't interfere with the runners. In particular,
the cyclists have no right to decide what is and isn't OK for the
runners.
Us cyclists and the runners had better decide to work together.
Otherwise, we'll all be mown down by the motor-bikers and the 4x4s!
73 from Ian, G3SEK
Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
G3SEK's WEB Page
Hi all,
Here are some comments on Leif's response, interspersed with his text.
For brevity, I have deleted those sections where we are in complete
agreement.
> The big problem in EME is the signal path. We have multi path
> propagation that causes phase and amplitude modulation. The fundamental
> thing for gaining a lower signal threshold is bandwidth.
>
> There are two bandwidths in the receiver; pre- and post- detector
> bandwidth. There are also two bandwidths for the signal; keying
> bandwidth (KB) and path modulation bandwidth(PB).
Allow me to add a third one - message bandwidth (MB), that needed to
carry the actual information as a baseband signal. IMO, EME is unique
in that MB (only a few bits per minute) is small compared with PB.
> There is NOT any fundamental difference between normal CW keying and
> what the computer can do.
Not quite true, using M-ary FSK, the computer can distinguish
frequencies only a few Hz apart, very difficult for the human ear.
> Commercial systems look for low error rate at maximum information speed
> (as far as I know). As a contrast amateurs look for very low information
> speed with a high error rate to convey just a small amount of
> information.
Agreed.
> As long as KB is much greater than PB it is possible to gain S/N by
> bandwidth reduction. We can easily use coherent CW to get the same
> advantage as the optimum digital system. No post detector filter is
> used. 10 times lower bandwidth allows 10 times less tx power at 10 times
> lower information speed.
True, but I also believe that noncoherent systems can be designed to
operate at signal level too low for a coherent system to "lock".
> When KB is approximately equal to PB it is not possible to transmit any
> information at low signal levels. This is because the receiver can not
> distinguish between the information intentionally modulated onto the
> carrier (on of keying, phase shift, whatever) and the modulation
> produced by the transmission path.
For coherent detection, this is correct, but I believe that a stronger
statement is also true: if the signal power is less than the noise
power in PB bandwidth, you can't recover information, regardless of
the value chosen for KB.
Another way of looking at this is that an OOK signal is equivalent to
a BPSK signal plus an unmodulated carrier. You use this carrier to
recover a phase reference. If it's too weak (relative to PB), you
can't lock the signal; this is independent of how fast anything is
being sent.
In the noncoherent case, information is present at all signal power
levels. Assume that you have a symmetrical system, e.g. one where the
carrier is on during an even second to send a zero, and on during the
odd second to send a one. The detector integrates the received energy
during each second, and subtracts the even value from the odd value.
If the result is positive, the bit is most likely a one; if negative,
most likely a zero. The magnitude (signal strength) gives you an
indication of how likely the decision is correct. As the signal level
is decreased, the bit error rate goes up, but it does not reach 50%
( no information) until the signal is zero!
Now, bits that are wrong 25% of the time are not worthless. If you
feed lots of them into a good ECC, you can get a few reliable bits
out! If you preserve the soft decision information, you don't need
as many flaky bits to do the job.
PB, of course, is not constant, varying over about a factor of ten.
It can be predicted from ephemerides and knowing the approximate
location of the other station. In theory, you should need 5 dB less
power at minimum PB than at maximum. A good challenge is developing
a system which takes advantage of this. I foresee a time when the
value of PB is taken as seriously as lunar distance or sky noise,
when choosing a time to operate.
Please do not take any of the above as an indication that I am not a
fan of coherent detection. If, for example, we want to bounce images
or decent quality speech off the moon (I believe this is possible
between a big gun and a four Yagi station, and would love to see
someone try it), then I would say in a heartbeat that BPSK is the
way to go.
> DSP or human mind, the problem is the same. (Practically we know that
> keying too slow does degrade readability. The DSP has the analogous
> problem)
IMO, this is mostly because in CW, "zero" is the same as "nothing".
A symmetrical system mitigates this greatly.
> If we want to make any improvement above the present state of the art
> which at 144MHz is KB between 15 and 20 Hz (and PB perhaps 4Hz) we have
> to look for some post detector filtering. This can be done in several
> different ways and they all have the potential of improving S/N by the
> square root of the pre- to post- detector bandwidths.
Indeed, an important design decision is the choice of KB. This can be
equal to MB (low performance, because all bits must be correctly
received), larger than MB but less than PB (better, that is the system
proposed in my original post), about equal to PB (probably optimum,
but more complex), or greater than PB (required for coherent systems).
> Common to all these methods is that the phase information is lost
> which is a 3dB loss compared to coherent cw. On the other hand the
> pre detector filter can me made to match PB which is approximately
> a 6dB improvement (on 144MHz).
Our numbers differ, partly because I was comparing a proposed system
to existing CW practice, not to an optimized coherent system. So I
assumed a 7 dB gain (25 Hz -> 5 Hz).
> Further there is a loss of about 7 dB because when post detector filters
> are used with bandwidths well below PB the detect decisions have to be
> based on the average signal level while the peak signal level can be
> used when KB is well above PB. (Typically I can copy two or three
> characters on the best QSB peak every second 1 minute period in a random
> QSO)
If you are using a symmetrical modulation, then the detect threshold
is always zero, regardless of signal level. Every bit, no matter how
weak, contributes some information to solving the puzzle. If you are
using just the peaks, you get no information at all most of the time,
and end up worse off, IMO. However, I will concede that there is 3 dB
additional loss, because the detector must receive KB+PB.
> Summing up: -3dB +6dB -7dB is 4dB or 2.5 times so we need 6 times
> narrower post detector filter compared to the 4Hz assumed for PB. In
> other words: To be compatible to todays CW, a system using average
> signals must use a bit length in order of 1.5 seconds.
If MB is 0.2 bits/second (A 12 bit message in one minute), then 6.5 dB
is gained from noncoherent averaging (4 -> 0.2). There is also 2 dB
of ECC coding gain.
My summation (compared to conventional CW): +7dB -3dB +6.5dB +2dB
is 12.5 dB. I admit that this is overly optimistic, for at least two
reasons: Redundancy in CW permits correction of some errors; I have
not attempted to calculate this but will guess at 1 dB coding gain.
Also, the messages sent are generally repetitive, and, to the extent
that the operator can combine pieces of different repeats, there is
some integration gain, which I estimate at 1 dB. So, maybe 10.5 dB is
realistic.
> The good news is that the post detector system has a nice signal
> threshold. Once the average signal is strong enough for a few bits to be
> decoded properly, all the bits will be at very nearly the same S/N so
> the communication easily becomes error free. (In contrast to normal CW
> or which the QSB often goes nearly through zero causing frequent loss
> of information even at power levels 10 to 20 dB above minimum for a QSO)
It's true that such integration quickly stabilizes the signal level.
But even for ham quality reliability, you would need to integrate
until the effective S/N was on the order of 10 dB. So it still makes
sense to use the best ECC practical.
> For two oscar class stations (100W, one yagi) we need to improve by
> about 6dB (Compare to how I can hear such a station with 4 yagis)
> .WAV files
Ouch. I had estimated a 15 dB shortfall (for my 300W, single 3.2 WL
station), naively working from the widely published "fact" that a 600W
four Yagi setup is what is needed to work a similar station under
"good" conditions. But I cannot in any way dispute Leif's 6 dB figure
- it is well documented and his experience is far greater than
mine. (Leif is even using 100W in his example, but is also assuming
5 WL, so that's comparable.)
Now, it's really embarrassing for me to not know by 9 dB, what my goal
is, and then talk about 1 dB differences. I can think of a few
factors which partly explain this disparity:
1) Leif's PC radio, quite impressive, receives optimum polarization.
2) His adaptive filters are more effective than those in most rigs.
3) His ears and brain are superior to those of the typical operator
assumed for my base "fact".
4) We are comparing "once in a lifetime" conditions with "good" ones.
I have no information to attach numbers to these factors, let alone
show how they add up to 9 dB, but would like to hear opinions from
Leif and other experienced operators.
> For 6dB improvement a data bit has to be 16 times 1.5 seconds or so. Of
> course the information rate can be improved by use of more frequencies
> but any such scheme will easily become useless when several stations try
> to work random.
I believe strongly in M-ary FSK, but feel that less than 0.2 dB
further gain can be had by using more than 16 frequencies. Although
there is some benefit from spacing wider than orthogonal, about 15 Hz
is more than adequate. So the widest signal I would propose would fit
easily in 300 Hz. There is room for lots of these on the band.
> ***********************
>
> The above text is intended to show that conventional CW keying together
> with sophisticated detection methods (A well trained human brain or a
> DSP with coherent CW and a slightly lazier operator) is close to the
> optimum solution on 144MHz. This is because our brains happen to match
> the optimum bandwidth that fits with the modulation caused by the signal
> path.
IMO, only for a loose definition of "close" :)
> Any averaging process such as post detector filtering or averaging of
> several repetitions of the same message implies the use of average power
> without phase information and leads to very low information rates.
Of course. But if you beat your brains out to complete a QSO in half
an hour, the information rate is also very low - about one bit/minute.
> Stewart wrote:
> > We can further improve the robustness of very slow OOK by eliminating
> > the asymmetry. One crude way is to take two seconds to send each bit,
> > using Manchester coding: A zero is sent by having the key down during
> > an even second, and up during the following odd second; the opposite
> > is done to send a one. In a one minute period, allowing time for
> > propagation and T/R switching, we can send 28 of these bits.
> This is too optimistic. To get the 6dB we look for, the data rate has to
> be in the order of one bit per minute. Certainly possible with clever
> coding but it will not be very exciting. Anyone having a few dB more
> power would like to use a higher bit rate. This kind of mode I think
> would be useful for skeds only, not for random!!
I was merely noting how many bits fit in the available time, and did
not mean to imply that these bits would all be "good". I will admit
that this modulation scheme is quite weak, but was chosen for these
reasons:
1) It is easy to explain and implement.
2) Each bit is independent in a two second interval. This allows
the complete system to be tested and debugged using the echo
path.
3) It is well suited to human detection, both by eye and by ear.
You only need to make a decision once every two seconds.
I thought that this would make it more popular politically.
Silly me.
Weaknesses include:
1) It is more sensitive to timing error at the receiver than
other OOK codes.
2) It is a repetition code which consumes transmission time that
could otherwise be used for better error correction.
3) M-ary FSK is much better!
> Of course my assumptions above may be somewhat pessimistic but they
> are certainly more realistic than any assumptions leading to a
> possible reduction of tx powers by 20 dB!
I agree that there is no magic bullet, and don't even dream of 20 dB,
but am optimistic that 15 dB is possible. We all know that to operate
EME, you need a good antenna *and* a powerful amp *and* and a low
noise front-end *and* a good receiver, etc. Likewise, if we use the
most robust protocol, the best source coding, the most powerful ECC,
the best modulation scheme, soft decision detection, *and* the most
effective pre and post detection filters, tuned to the ever changing
PB, we will get there!
73,
Stewart Nelson, KK7KA
From:
"Ian White, G3SEK"
i3dli wrote:
> Among others, Leif SM5BSZ and Jan G3SEK pointed out,
> to me, the entire question:
>
> - Leif demonstrate by math & him own impressive experimental
> work what one can expect by the "entire digital" and "digital aided" small
> signal detection technologies.
> - Jan substantially declarate, by sport analogies, the ham must
> be the ultimate ring of the whole EME path chain. Otherwise
> is'nt ham-radio anymore.
>
I did not say that. I believe that all-computer modes like RTTY, packet,
PSK31 etc ARE ham radio. They have a right to exist, and people have
a right to enjoy them.
But they are a different kind of ham radio, and I don't want them
running all over the frequencies where my kind of amateur radio lives.
To be fair, if you read what Stewart KK7KA really did say, he was not
suggesting anything like that. He was only discussing alternative
modulation methods, which is a valid topic.
Stewart, please be aware that you are not writing on a clean slate here.
In the spread-spectrum debate about a year ago, some S-S advocates tried
(a) to tell us what we should be enjoying, and
(b) to use bogus technical arguments to "prove" that we wouldn't experience
any interference from the raised noise level from S-S. That episode
strongly polarized opinions in this group.
Also, as I3LDI has said, for most of us the modulation/detection method
is part of the basic choice we have already made about the style of
operating, as well as to the EME path.
73 from Ian, G3SEK
Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
G3SEK's WEB Page
HI all.
Rereading more carefully on my previous mail
and the mail of Jan G3SEK, I've to admit I have
misunderstood the letter of Jan on the subject.
Nevertheless I hope to be not so far from him
opinion.
My apologies to Jan, anyway.
73,s
Paolo, I3DLI
>> David Blaschke wrote:
>>> I notice you (KK7KA) are not in my eme log.
>>
>> We each have our own esthetic sense. In my mind, the technical
>> challenge of ham radio, as opposed to say, the Internet or cell
>> phones, is that we are using our own gear, rather than depending on a
>> large infrastructure. Since Dave has his own large infrastructure,
>> I would not consider it remarkable if he could hear me, and so did
>> not make that effort, instead putting energy into a system which I
>> hope will eventually permit contact with hundreds of stations.
>> Comments and further flames welcome,
>>
>> Stewart Nelson KK7KA
>>
>
>
>
> Hi all:
>
> I believe Dave's "large infrastructure" is a system to permit
> other minor stations to make an EME QSO their own, using
> only one frequency at the time.
> AND the "large infrastructure" is also him "own gear", I think.
>
> On the other side, digital modes, along with their algorithms,
> are another way to explore about the best mode to extract
> informations from the EME path IN THE RESPECT
> of the fundamental rules for radio amateurs.
>
> Among others, Leif SM5BSZ and Jan G3SEK pointed out,
> to me, the entire question:
>
> - Leif demonstrate by math & him own impressive experimental
> work what one can expect by the "entire digital" and "digital aided" small
> signal detection technologies.
> - Jan substantially declarate, by sport analogies, the ham must
> be the ultimate ring of the whole EME path chain. Otherwise
> is'nt ham-radio anymore.
>
> In effect, by comparison, I thougt to the chess games nowadays: it's
> a fashinating thing to see the uman being in figthing against
> a computer. A computer against another computer maybe
> even more astonishing....but only for the unknown programmers.
> It's still a "chess game" ?
> Would be that in the future chess games will be played only
> in the more efficient way to mate the king ?
>
> Back to the topic. I believe we must proceed to enhance
> our cabability in making EME QSOs using up-to-date
> technologies. The limit, uncertain, lying in a pursued
> situation where the natural sensing system
> (ears-brain) will be the final discriminator for the QSO.
>
Paolo, I3DLI
Return
W6/PA0ZN