Michael Coslo on Fri, 16 Mar 2007 09:42:30 CST wrote:

John Smith I wrote:
wrote:

However, if a new market comes forth, one composed of amateurs with
little or no knowledge of CW and only using digital voice and digital
data transmission--it would be market controlled also, and one would
suspect it would self-correct and frown on the use of the bands for
wasteful analog and cw communications.

I would disagree. Those modes are not wasteful. On the other hand, a
vision of using the HF bands for data transmission would indeed be a way
of filling up our bands pretty quickly, and for not a lot of gain. If
I'm interested in Data Transmission, I would design a system for
frequencies where there is less natural noise - VHF and up. Then
bandwidth issues would be less of a problem too.

I disagree with both of you...:-) For one thing, 300 WPM
equivalent data rate at 170 Hz "Spread" on HF does a
credible job of sending text in only a half-KHz of
bandwidth.

The presumption is that "data" somehow MUST have "perfect"
conditions to avoid errors is false. The BER or Bit Error
Rate rules the show and is a function of noise and
transmission rate (in units per second) and bandwidth.
Claude Shannon used the example of a teleprinter signal on
his seminal 1947 paper...which became boiled down to the
more familiarly-known "Shannon's Law." That was 60 years
ago and Claude wasn't considering OOK CW modes. :-)

"Data" can have a wide BER range depending on the design
of the data coding, all compared data systems having the
same data rate, signal-to-noise ratio, and channel
bandwidth. Forward Error Correction improves the BER
but isn't an absolute necessity.

An example is the ordinary modem we use on-line. If you
have a human handset as well as the modem line, try
picking up the handset and making random noise in it while
the modem is on-line. That's an extreme case, but
survivable without data disaster. You might be surprised
at how well it can survive without messing up the screen.

Let's face it, digital voice is the only way to go.

I would disagree. What I have seen in digital voice so far offers no
particular advantage over SSB, unless we are talking about digital for
it's own sake. Most schemes that I have seen have some fatal flaws, such
as the received transmission must be received in toto - IOW if you don't
catch the beginning, you don't catch anything.

I disagree with that and I have seen/heard many such systems
but - certainly - not all of them. The digitized bit stream
can be structured to enable a receiver to ID it and lock onto
it quickly. It there are lots of tones in the multiplexed
digital signal (such as with OFDM) that should be enough for
an ID and lock-in.

The solution to that
would be channelizing HF, or assigning specific frequencies to Digital
voice. In addition, unless there have been some big advances recently,

The "big advances" have already come, like in the late 1980s.
I'm mentioning a hint to the U.S. military SINCGARS in its
digital mode (with or without frequency-hopping). DSSS
essentially. Such can be slowed down or scaled to reduce its
bandwidth without disabling intelligibility (no encryption
needed or allowed by amateur regulations).

Digital voice does not have any particular bandwidth advantages.

Maybe not, but the decoded voice can be crystal clear all the
way to the threshold point (where it breaks up suddenly). It
sounds like an FM link with lots of amplitude variation, yet
there isn't any decoded speach amplitude variation.

A case in point is HDTV that we've had in this house for a
year. I've put an attenuator in the TV cable line and NOT
see a bit of difference in video nor hear any in the audio
until there is lots of attenuation reaching the threashold
of input.

Give me a Digital voice mode that I can tune across the band and pick up
a conversation at any point in the transmission, and a 1 KHz or less
bandwidth signal, and then I'll be interested.

Those are already in the works.

And of course, I'll need
to see that some other folks are buying the digital radios too, so I'll
have people to talk to.

Hmmm...what if they are thinking the same thing? :-)

Case in point: PSK31, Peter Martinez' clever brainchild was
spread all over Europe and tested by many on the Continent for
four years before it was first publicized in ARRL publications.
Not many in the USA were aware that PSK31 even existed, let
alone proven under "field conditions." Publicity caused its
spread over on this side of the pond.

PSK is too slow for
data transmission of LARGE and multi-megabyte amounts of data, end of
story.

Sorry, John, but you haven't justified any NEED for "large and
multi-megabyte amounts of data" in the amateur bands below
30 MHz. Please think harder on how much data throughput CAN
happen with existing data bandwidths and rates first. It is
quite large, although that is in subjective terms.

As for PSK, you are correct that it is too slow for data transmission.
But that little 31 baud signal was never intended for Data transmission.
It was intended for correctable text at a rate that a reasonably
proficient typist could tap away at the keyboard in real time.

Absolutely so and that was a design goal of G3PLX way back
in the begining.

Also a
mode that can be efficiently operated at QRP and lover levels. And for
that, it is an excellent mode.

Not necessarily true. PSK31 is efficient in terms of bandwidth
reduction versus data rate, still well within Shannon's Law, but
it can be used at high RF powers just as easily as low RF powers.

It seems to me to be Conventional Wisdom (a new form of "CW")
that "high power" in USA amateur bands is associated only with
OOK CW or SSB. All other modes seem to be ignored in the
literature as a general rule. That's not a technical thing,
just a subjective thing of the high-power types' desires. I've
observed that most of them are ultra-conservative (as a general
rule) insofar as mode use is concerned.

Sometimes one has to look "outside the box" of Conventional
[amateur] Wisdom to see where contemporary limits are in the
practical, working sense.

73, Len AF6AY