"John Smith" wrote in message
...
Mike:

300 baud is ridiculous, in Dee's first post mentioning 300 baud I tossed
it out the window--that was fine up to about 1985, then only the mentally
challenged continued to run 300 baud modems!


Please show me and everyone else how we can run more than 300 baud on HF
without exceeding reasonable band widths. There are a whole lot of things,
not just video, that would be nice to do.

How can we do it? Bandwidth is directly related to baud rate.

Dee D. Flint, N8UZE

"Dee Flint" wrote


Hey I'm all for the "eureka" when it happens but the problem is that it is
unpredictable. Not only is it unpredictable in time but in the nature of the
breakthrough.


That's what makes ham radio some damn much fun! In my profession role I can
send a team of engineers off with some marketeers scribbling and know that
within 12-18 months I'll be shipping product. Bnt ham radio is not so
predicable --- we get these delightful surprises from unexpected places.

Some like APRS and PSK-xx gain traction and thrive in a niche, others like AX.25
packet radio and 2-meter autopatches which blossom like an Independence Day
firework, then fizzle to a few sparks on the ground after a short period of
glory.

Then there are a few genuine "revolutions" which fundamentally change the nature
of amateur radio. We're about due for one of those.

73, de Hans, K0HB

"Dee Flint" wrote


Please show me and everyone else how we can run more than 300 baud on HF
without exceeding reasonable band widths.

Why is this conversation hung up on 300 baud?

It's perfectly legal, for example to run digital voice (J2E) on HF under todays
FCC rules, and it will fit nicely in the generally accepted 3KHz band width now
used by traditional SSB. 97.307(f)(3) applies to rtty/data transmission, it
does not apply to voice or image.

73, de Hans, K0HB

Dee:

You don't understand binary compression techniques, ok...

.... it all has to do with binary trees (well, mostly, kind of), in
software, of the data stream, multiple bytes are converted into "data
streams", i.e., a pixel "byte" at this level is NOT necessarily 8 bits
long in a stream, and sometimes can be represented by a single bit
(automatic compression to 1/8 size just because it is stored in a
binary tree! (or, data stream) multiple occurring bytes can be sent in
the form of (five of these) or ((67 - number of bytes) of (00100100
- binary data byte to create 67 of)), in other words you tell it how
many of ONE type of data to create on the other end to fill in the
"video hole" on the screen, I would think you can visualize how small
a simple BW image can be transmitted--it grows bigger with grayscale
data, and much bigger with color data (in quality pictures an extra
few bits have to be sent just to describe the color/brightness of the
pixel being sent...)

.... there is also "variable bit rate compression" which I don't even
want to begin to try to give a simplified explanation of here...

I am sorry, my ability to describe these complex methods at work here
is lacking, and I realize this...
don't trust me on it, the web is loaded with papers on every aspect of
it...

The size of compressed data? That depends on the data types
compressed, BW video can be 90%+ compressed (resulting in data 1/10 to
1/20 the size, or MORE, it all just depends on the complexity of the
image.) For example, a completely white frame would be (for example)
1,024,000 bytes of color 00000000--this whole screen could be
transmitted in TWO BYTES!
and the same for an all black screen, at EXTREME RESOLUTION in this
simplified case.

Further "compression" can be had down at the hardware level where the
transmission software can "scan" data and "table-ize" streams of
duplicate bytes, or very similar bytes which can all be represented by
a common value with little or no detectable loss in "realized visual
quality." (can you really tell almost-almost black from "real" black?
Or, almost-almost-blue from "real" blue?)

In a very efficient compression scheme, it can be "mentally modeled"
as a onion, where many "layers" of compression are occurring in a
tight sequential loop creating very tightly compressed data packets,
with crc sums to ensure no data corruption and packets sequentially
numbered to provide a "sane" display stream (this can frequently be
rather lax with low quality audio (speech) and less than absolute
perfect video.)

Digital cell phones use very similar techniques on audio. Some of the
"trade secrets" there are closely protected...

It is really beyond the resources we have here to go into a deep
explanation on data compression techniques, and cheap tricks and short
cuts--a good book on the subject should bring one up to speed
quickly--perhaps amazon.com for those with a desire for a in-depth
understanding...

Now some "cheap tricks" examples:

you can actually throw away every other pixel (immediately cut the
size of a video frame in half!) by using a "normalized" colored pixel
in those "dropped" pixels place (and normalizing this "fill in pixel
color" as needed to fit the "general background" of the rest of the
picture--with NOT as great a loss of video quality as you would expect
(or a smaller percentage of "normalized" pixels if greater quality is
really needed)

and, a smaller than screen sized "picture" can be broadcast and
"expanded" by "image size extrapolation" (computer makes a lot of
guesses on how to represent it as a larger picture and "fill the
screen"--and attempts to have those "guesses" maintain a reasonable
quality of picture.

and, I could go on and on, however, this is quickly becoming WAY
beyond the scope of the arena we need to hold this
"argument/discussion" within... and there are papers and books which
can do a much finer job than I...

I must say, Len was quite correct in the appraisal of your mental
aptitude, I would venture to say this with confidence, as damn few
women would have hung with this technical discussion as you have...

.... are you single, just how old are you? leering-smile

.... just kidding, well, mostly--I AM single yanno!
grin

Warmest regards,
John

"Dee Flint" wrote in message
...

"KXHB" wrote in message
ink.net...

"Mike Coslo" wrote


So here we are.


Yup, and no one has persuaded me it can't be done. I've only been
persuaded that we haven't figured out how yet. (Sorta reminds you,
doesn't it, of how those old-tymey hams must have felt when they
were told to take their party to "200 meters and below".) You,
Jim, and Dee bemoaning how hard it will be, and John raising the
tantalizing notion that we may only be a few "eureka!!!"s away from
something workable. Outside my area of competence, but I'll watch
the dialog with interest.

73, de Hans, K0HB

From my understanding of John's comments, he is saying it can be
done now with current technology. He does not however tell us how.
He just chatters on about "compressing it enough" without stating
the degree of compression, etc.

Hey I'm all for the "eureka" when it happens but the problem is that
it is unpredictable. Not only is it unpredictable in time but in
the nature of the breakthrough.

Dee D. Flint, N8UZE

wrote:
Mike Coslo wrote:

wrote:

Mike Coslo wrote:

Dave Heil wrote:
LenAnderson@ieeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeee.org wrote:


How we be gonna scale those pictures and live video to fit
into 2.5 KHz?


Two steps:


1) Convert the pictures and video into highly-compressed
digital formats for transmission.

Oooh, there could be a problem there! There are limits to the
compression, and we have exceeded them in some forms already.


Of course! But it all depends what you consider "acceptable quality"...

I'm assuming it has to be better than a very high quality SSTV image in
the case of stills, and present day OTA video signals....


Check to
see how many vertical pans there are on video signals lately.
The
compression on the digital signals (note that even if you are
getting
your feed via analog cable, you are still almost certainly
looking at a
digital signal) already calls for some major aliasing.


OTOH, if all you want is B&W ~CGA video....

Comes pre-aliased! ;^)


There are limits, and there are limits. How much more are we
going to throw away?


Always a tradeoff. Hams routinely use 1.8 kHz wide SSB filters for
"communications quality". Hardly hi-fi but it can make the difference
between QSO and QRJ.


2) Use different modes/modulations/protocols

Shannon's Theorem tells us that we can get very high data
rates through
very narrow bandwidths *if* we have adequate signal-to-noise ratio.
Note that "noise" takes many forms, not just the thermal
noise we're used to.

What're we going to do when the data rate that we need is darn
near(or above) frequency in use?

Use modes designed for the purpose. See below.

And a high enough frequency to handle the BW requirements.



For example, PSK has an advantage over OOK when dealing with thermal
noise. But when dealing with other types of noise, OOK can
have an
advantage. It all depends on the transmission medium. What
works on a
telephone line may not work on an HF path of the same apparent
bandwidth.


I thought that we were going to be able to send live video
and digital images on HF?


You can do that now - just need enough S/N.


Always?


No mode always gets through.

But if you have enough S/N, all sorts of things are possible.

Remember though that we are talking about our favorite playground - HF.
A high S/N is not often the case here.


Here's one way. I apologize if you are way beyond this simplified
example:

Consider how PSK31 works in BPSK mode. There are just two basic
modulation states - 0 degrees and 180 degrees. One bit per unit time.

But in QPSK mode, there are four basic modulation states - 0 degrees,
90 degrees, 180 degrees and 270 degrees. Two bits per unit time, but
the bandwidth is no greater than with BPSK. Only problem is that you
need a transmitter, receiver and transmission medium whose total
distortion is low enough that you can accurately tell the four states
apart.

Now consider a theoretical "256PSK" mode, in which there are 256
states: 0 degrees, 360/256 degree, 720/256 degrees, etc., all the way
to ~359 degrees. 8 bits in one unit time, in the same bandwidth! But
you need a transmitter, receiver and transmission medium whose total
distortion is low enough that you can accurately tell the 256 states
apart.

The error rate would be fantastic!

You can see that if we just keep increasing the number of states, the
number of bits per unit time in the same bandwidth keeps going up. But
you need more and more accurate modulator/medium/demodulator - IOW,
better and better signal-to-noise. Or to look at it another way, the
mode carries huge amounts of data in a tiny bandwidth but has very
little tolerance for noise that takes the form of phase distortion.

Now for this system to be practical, there would have to be a way to
correct for all that phase distortion

Now imagine multiple spaced carriers in the 2.5 kHz bandwidth all
carrying data - lotta bits, huh?

Now yer cheatin! ;^) That is increased bandwidth


Of course you may find that in practice it's not that easy to get a
modulator/medium/demodulator setup that meets the requirements -
particularly if the medium is HF RF with relatively low power.

I have to imagine that there must be a lot of power, despite the
sensitivity being to phase distortion. When I look at my phase display,
there is a lot of noise showing up that can become bogus phase noise.


Simply by hooking our computers to our rigs via the
proper interfaces.


And software.

I really didn't think it was all that simple.


Nobody said it was simple!

Mr. Smith did!

Why don't we get together
and pop off a live video system for say the 160 meter band. The video
would be real time, 30 fps, and otherwise like broadcast video. Better
yet, Why don't we do it at computer resolution?


Ask the PROFESSIONALS, Mike. Remember, ham radio is a HOBBY, according
to them....

SNORT!


Now it seems that the *idea* is that we are going to use
DRM, and we're
going to need to get more spectrum in which to use.


There are all sorts of solutions. But there's a world of
difference between people talking theory and actual
application.

I did hear that DRM was capable of doing imagery. I couldn't find any
examples tho'. And they were very vague about it.


Most of all, some folks confuse the journey and the destination.

The journey beats all.....


Exactly.

Does complex and newer equal better?

Sometimes. Not always.


Is analog simpler than digital?

Sometimes!

Does having a computer that attaches to the Internet
make a person a digital expert?


Some folks think so! I don't. And besides - "digital expert" doesn't
mean someone knows much about radio.

Ain't that the truff?


I ask for enlightenment, I get invective.

Are you surprised?

Nope. It doesn't make for a very good discussion tho'.


Discussion is not what the invective-hurlers want, Mike.

You're right. Your multi-angle psk is the closest thing to possibility
that I have seen yet. For Satellites and other UHF applications, it
starts to become possible/practical:

http://www.tech-faq.com/qpsk.shtml

They also write about 8psk. Note that link degradation is an issue.


Here on HF, we just don't have the proper conditions. I can do
quadrature mode, but almost no one does. I've done some QSO's using
BPSK63, but of course that uses more BW.

But imagine! Someone (you) writing about something with a real
possibility, not just calling me stupid or ignorant, etc!!!!

Sunnavagun (with apologies to Hans)
BTW, 20 meters is going to town tonight!

- Mike KB3EIA -

Dee:

This "show me", "show me" you are repeating causes a complete
confusion on my part.

Will you agree that a 56K phone modem, does indeed, transmit this data
rate with an audio bandwidth of ~300Hz to ~5000K, and if you do so
agree, how can you argue this cannot fit in a HF AM RF signal which
only goes 2.5K each side of center frequency??????????

Are you NOT imposing an audio frequency of AT LEAST a 5K bandwidth on
the rf carrier with normal speech? (actually, most quality
transceivers have a wider audio bandwidth than this which can be set
+/-) and if you agree you are indeed, how can you argue that 5K
bandwidth can carry a 56K data rate over a phone line--and NOT a hf rf
signal???? That looks insane to me?

The modem is NOT using the whole 5K bandwidth--necessarily, there is
compression into a narrower bandwidth which can and is generally
software controlled--if necessary (the modems software is a LOT
smarter than most give it credit for, especially in the case of the
old "onboard processor" and "hardware logic" USRobotics external
modems.

You need to explain to me why it even begins to look difficult to you
for me to be able to understand what you are asking?

As, I have to be missing something here...

You do realize that a picture good enough to run a "webcam" on the
amateur HF bands and get an acceptable image from can be done in 28K
(or less depending on the fps), and 36K is really fine at 5 fps and
good at 10 fps--you will be able to see the wart on the guys nose you
are video conferencing with at 36K!!!

You know, I have not even looked to see on the web, but aren't tons of
people doing this right now as we newsgroup?

I suppose you could actually use the rf signal as data carrier itself
and modulate it directly through on/off switching, as opposed to
modulating the rf carrier with the audio data carrier... but that
would take some heavy duty equip mods/revamps, if it didn't wipe out
the neighbors cable tv! grin

Think about this:
at 100 mhz if you can precisely control the EXACT amplitude of each
and every cycle of rf out the back end of the xmitter, you have a
virtual 100mbs data carrier... most are working here... 450 MHz?
1Ghz? 12Ghz?

.... and of course, the receiver has to be able to decipher the
amplitudes of each cycle back to a data stream for the video card...

.... this is the land where dreamers are...

John

"Dee Flint" wrote in message
...

"John Smith" wrote in message
...
Mike:

300 baud is ridiculous, in Dee's first post mentioning 300 baud I
tossed it out the window--that was fine up to about 1985, then only
the mentally challenged continued to run 300 baud modems!


Please show me and everyone else how we can run more than 300 baud
on HF without exceeding reasonable band widths. There are a whole
lot of things, not just video, that would be nice to do.

How can we do it? Bandwidth is directly related to baud rate.

Dee D. Flint, N8UZE

KØHB wrote:
"Mike Coslo" wrote


So here we are.



Yup, and no one has persuaded me it can't be done. I've only been persuaded
that we haven't figured out how yet. (Sorta reminds you, doesn't it, of how
those old-tymey hams must have felt when they were told to take their party to
"200 meters and below".) You, Jim, and Dee bemoaning how hard it will be, and
John raising the tantalizing notion that we may only be a few "eureka!!!"s away
from something workable.

He also give a lot of solid technical ways in which this can be done, eh?


Outside my area of competence, but I'll watch the
dialog with interest.

Hey, Hans, ignorance is not a crime! Note that Jim brought up an
*actual* method of trying to do a lot of BW using 256 or more phase
angles that are decoded by the receiving station. That is not likely to
work at HF, but a simplified version of this is used for some satellite
comms. they (see my link in my post to Jim) note that QPSK is more
reliable - or at least suffers less from link degradation - same thing,
than 8PSK. But there is some theory there that can be discussed.

And as for "bemoaning", I have been asking for something based in
solid theory since early in this thread. Most of what I have gotten in
return is that I am an olde tyme ham (untrue) stuck on CW with my Bug
(paraphrased, but laughably untrue), and topic shifted to DRM voice
(technically working, but beside the point). That ain't substance.

John Smith wrote:
Mike:

Oh no. Now someone is going to have to explain ccd cams and pixels to
you, huh? Take a course!


Challenge, "John Smith"!


DO IT!

Post the method in which you and I can send live Video to each other via
whichever HF band will propagate between our QTH's, and I will build a
duplicate. We can set up a sked. Once we have established live
communications, I will most certainly apologize for my olde tyme
hamminess. Let us keep everything in the group so that I may apologize
publicly when proven wrong.

Anxiously awaiting your system outline and diagrams..

- Mike KB3EIA -

Dee Flint wrote:

Please show me and everyone else how we can run more than 300 baud on HF
without exceeding reasonable band widths. There are a whole lot of things,
not just video, that would be nice to do.

How can we do it? Bandwidth is directly related to baud rate.

Dee D. Flint, N8UZE

John has been challenged. His system for real time video via HF will be
posted soon, TTPUOSU!

- Mike KB3EIA -

Mike:

The clock in a ~4GHz computer and DDR memory makes modem data xfr look
incredibly s-l-o-w.... with spaces miles long between marker bits...

100Mbs nic cards are not even close to a challenge to that clock
speed...

John

"Mike Coslo" wrote in message
...
wrote:
Mike Coslo wrote:

wrote:

Mike Coslo wrote:

Dave Heil wrote:
LenAnderson@ieeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eeeeeeeeeeeeeeeeeeeeeeeeee.org
wrote:


How we be gonna scale those pictures and live video to fit
into 2.5 KHz?


Two steps:


1) Convert the pictures and video into highly-compressed
digital formats for transmission.

Oooh, there could be a problem there! There are limits to the
compression, and we have exceeded them in some forms already.


Of course! But it all depends what you consider "acceptable
quality"...

I'm assuming it has to be better than a very high quality SSTV image
in the case of stills, and present day OTA video signals....


Check to
see how many vertical pans there are on video signals lately.
The
compression on the digital signals (note that even if you are
getting
your feed via analog cable, you are still almost certainly
looking at a
digital signal) already calls for some major aliasing.


OTOH, if all you want is B&W ~CGA video....

Comes pre-aliased! ;^)


There are limits, and there are limits. How much more are we
going to throw away?


Always a tradeoff. Hams routinely use 1.8 kHz wide SSB filters for
"communications quality". Hardly hi-fi but it can make the
difference
between QSO and QRJ.


2) Use different modes/modulations/protocols

Shannon's Theorem tells us that we can get very high data
rates through
very narrow bandwidths *if* we have adequate signal-to-noise
ratio.
Note that "noise" takes many forms, not just the thermal
noise we're used to.

What're we going to do when the data rate that we need is darn
near(or above) frequency in use?

Use modes designed for the purpose. See below.

And a high enough frequency to handle the BW requirements.



For example, PSK has an advantage over OOK when dealing with
thermal
noise. But when dealing with other types of noise, OOK can
have an
advantage. It all depends on the transmission medium. What
works on a
telephone line may not work on an HF path of the same apparent
bandwidth.


I thought that we were going to be able to send live video
and digital images on HF?


You can do that now - just need enough S/N.


Always?


No mode always gets through.

But if you have enough S/N, all sorts of things are possible.

Remember though that we are talking about our favorite playground -
HF. A high S/N is not often the case here.


Here's one way. I apologize if you are way beyond this simplified
example:

Consider how PSK31 works in BPSK mode. There are just two basic
modulation states - 0 degrees and 180 degrees. One bit per unit
time.

But in QPSK mode, there are four basic modulation states - 0
degrees,
90 degrees, 180 degrees and 270 degrees. Two bits per unit time,
but
the bandwidth is no greater than with BPSK. Only problem is that
you
need a transmitter, receiver and transmission medium whose total
distortion is low enough that you can accurately tell the four
states
apart.

Now consider a theoretical "256PSK" mode, in which there are 256
states: 0 degrees, 360/256 degree, 720/256 degrees, etc., all the
way
to ~359 degrees. 8 bits in one unit time, in the same bandwidth!
But
you need a transmitter, receiver and transmission medium whose
total
distortion is low enough that you can accurately tell the 256
states
apart.

The error rate would be fantastic!

You can see that if we just keep increasing the number of states,
the
number of bits per unit time in the same bandwidth keeps going up.
But
you need more and more accurate modulator/medium/demodulator - IOW,
better and better signal-to-noise. Or to look at it another way,
the
mode carries huge amounts of data in a tiny bandwidth but has very
little tolerance for noise that takes the form of phase distortion.

Now for this system to be practical, there would have to be a way to
correct for all that phase distortion

Now imagine multiple spaced carriers in the 2.5 kHz bandwidth all
carrying data - lotta bits, huh?

Now yer cheatin! ;^) That is increased bandwidth


Of course you may find that in practice it's not that easy to get a
modulator/medium/demodulator setup that meets the requirements -
particularly if the medium is HF RF with relatively low power.

I have to imagine that there must be a lot of power, despite the
sensitivity being to phase distortion. When I look at my phase
display, there is a lot of noise showing up that can become bogus
phase noise.


Simply by hooking our computers to our rigs via the
proper interfaces.


And software.

I really didn't think it was all that simple.


Nobody said it was simple!

Mr. Smith did!

Why don't we get together
and pop off a live video system for say the 160 meter band. The
video
would be real time, 30 fps, and otherwise like broadcast video.
Better
yet, Why don't we do it at computer resolution?


Ask the PROFESSIONALS, Mike. Remember, ham radio is a HOBBY,
according
to them....

SNORT!


Now it seems that the *idea* is that we are going to use
DRM, and we're
going to need to get more spectrum in which to use.


There are all sorts of solutions. But there's a world of
difference between people talking theory and actual
application.

I did hear that DRM was capable of doing imagery. I couldn't find
any
examples tho'. And they were very vague about it.


Most of all, some folks confuse the journey and the destination.

The journey beats all.....


Exactly.

Does complex and newer equal better?

Sometimes. Not always.


Is analog simpler than digital?

Sometimes!

Does having a computer that attaches to the Internet
make a person a digital expert?


Some folks think so! I don't. And besides - "digital expert"
doesn't
mean someone knows much about radio.

Ain't that the truff?


I ask for enlightenment, I get invective.

Are you surprised?

Nope. It doesn't make for a very good discussion tho'.


Discussion is not what the invective-hurlers want, Mike.

You're right. Your multi-angle psk is the closest thing to
possibility that I have seen yet. For Satellites and other UHF
applications, it starts to become possible/practical:

http://www.tech-faq.com/qpsk.shtml

They also write about 8psk. Note that link degradation is an issue.


Here on HF, we just don't have the proper conditions. I can do
quadrature mode, but almost no one does. I've done some QSO's using
BPSK63, but of course that uses more BW.

But imagine! Someone (you) writing about something with a real
possibility, not just calling me stupid or ignorant, etc!!!!

Sunnavagun (with apologies to Hans)
BTW, 20 meters is going to town tonight!

- Mike KB3EIA -