Re: Higher Frequency Monitor?

["Michael J. Mahon" <mjmahon@aol.com>]

Bryan Parkoff wrote:
> > There is no such thing as an NTSC "dot".  There are only
> > color artifacts corresponding to the position of a pulse
> > relative to the 3.58MHz color subcarrier.  When a series
> > of short pulses is interpreted as color, the color displayed
> > is not properly a result of the juxtaposition of several
> > very small color dots corresponding to each pulse.  The
> > whole "constellation" of pulses corresponds to a single
> > color dot.
> >
> > A proper RGB decoding of the Apple II video signal will
> > similarly replace a "pixel" constellation with a single
> > color pixel as wide as the entire constellation.  To do
> > otherwise is to invent new graphics modes.
>
> Michael,
>
>     Thank you for explaining.  I still do not understand what do color
> reference mean.  I believe that it means to paint/display number of pixels
> in color at 3.5MHz.  Let say for example.
>
>     Deep Red is 1000 bits and Brown is 0001 bits.  Do color reference reads
> "1000" before it displays only one deep red pixel in the position 0 while it
> displays black pixel in the position 1 through 3?  If second 1000 bits come
> after current 1000 bits, it should display deep red pixel in the position 4
> while it displays black pixel in the position 5 through 7 before position 1
> through 3 has black pixels to become deep red pixels because position 0 and
> position 4 are joined together.  Correct?
>     Try to think that dark blue and dark green in position 1 and 2 to 
> produce medium blue.  Dark green and brown in position 2 and 3 to produce 
> light green (HIRES).  Deep red, dark blue, and dark green in position 0 
> through 2 to produce light blue.  Dark blue, dark green, and brown in 
> position 1 through 3 to produce Aquamarine.  Deep red, dark blue, dark 
> green, and brown in position 0 through 3 to produce white.  How can you 
> prove to be true?  Think about orange.  Brown in position 3 and deep red in 
> position 4 to produce orange.
>     Now, do you notice that 0001 bits for brown have black pixel in position
> 0 through 2 and brown pixel in position 3?  It is how you can see black and
> color fringe in the left and right block of low resolution.
>     Is it true that Apple II sends only one bit to NTSC monitor at this time
> before NTSC displays one out of four dots to produce pixel?  For example,
> 1100 bits is purple.  Apple II loads one bit which it looks like "1"000
> before NTSC monitor displays deep red pixel in position 0.  Then Apple II
> shifts next bit which it looks like 0"1"00 before NTSC monitor displays dark
> blue pixel in position 1.  Now, position 0 is deep red pixel and position 1
> is dark blue pixel which they are joined together to produce purple pixel.
> If you look very close at NTSC screen, you will be able to see deep red
> pixel in the left behind purple dot and dark pixel in the right behind
> purple dot.
>     I still don't understand why Apple II is designed to send one bit at
> this time instead of four bits.  Please explain.
>
>     Go back to DHR.  I have figured 88 11 22 44 in both aux even/odd and
> motherboard even/odd.  Even aux never rotate one byte.  Even motherboard
> always rotates one byte to the left three times, odd aux rotates one byte to
> the left two times, and odd motherboard rotates one byte to the left one
> time.  It would show the string like "10001000100010001000100010001000".  It
> is reversed to show "00010001000100010001000100010001" when it displays to
> NTSC monitor.
>     My question is -- why did Apple Computer, Inc make the decision to
> choose 88 instead of 11 in even aux to produce DHGR pixel?  You may notice
> that there are black pixels in the far left position before deep red pixel
> starts in position 3 through 559.  It is interesting.  I wish that 11 should
> be acceptable in aux even to produce deep red.

The theory of color encoding in the NTSC system is well explained
in numerous places on the web (Google: ntsc color encoding, for
example:

http://members.aol.com/ajaynejr/colordec.htm#Standard

If you study it, you will see that it is an analog system,
with no such thing as "pixels".  A receiver may use several
different color decoding methods, each of which results in a
slightly different gamut of colors, and a somewhat different
color resolution (the number of color changes which can be
reproduced in a single horizontal line).

Most analog decoders approximate the level of in-phase and
quadrature-phase sidebands of the 3.58MHz color subcarrier.
The actual subcarrier is suppressed, and only the sidebands
are transmitted.  The phase reference for the carrier, which
the decoder must re-create to demodulate the color sidebands,
is sent as a "color burst" of a few cycles of 3.58MHz signal
during part of the "back porch" of the horizontal blanking
pulse.

The approximation of the in-phase and quadrature components
of the sidebands of the color subcarrier convey the phase
angle (hue) and amplitude (saturation) of the color info,
with the luminance information being sent as baseband video.

When a digital video signal (such as the Apple II video output)
is sent into an NTSC receiver, its color demodulator syncs to
the 3.58MHz color burst, just as with a broadcast signal.  But
when a short video pulse occurs (and all the Apple II can do is
produce pulses, which may merge into a longer pulse if adjacent),
it is interpreted as containing a luminance component (its average
value) and a color component, determined by the amplitude and
phase of a color sideband which is imputed to be present because
of the bandwidth and timing of the pulse.  A short pulse will
have low luminance and a color depending on its position relative
to the reconstructed 3.58MHz color subcarrier.

So as a pulse varies in its timing relative to the 3.58MHz color
burst on the back porch of the horizontal blanking, it will
appear to change in hue while maintaining almost constant
brightness.  If two pulses occur in less than one cycle of 3.58MHz,
then they will be interpreted as having an average phase (and
therefore hue) between the two separate pulses.  Their average
luminance (brightness) will be double the luminance of a single
pulse.

By producing any of 16 combinations of pulses at four times
within a single 3.58MHz subcarrier cycle, it is possible to
produce 16 different hues (including black and white) with
five different luminances (corresponding to 0-4 bits on).
These are the Apple video colors.

The clock frequency of the Apple II was chosen to permit all
video pulses to be at multiples of 2x the subcarrier frequency
(HGR) or 4x the subcarrier frequency (lo-res and, later, DHGR).
The alternate HGR color set simply moves the two pulses that
can be produced in one 3.58MHz cycle by 1/4 cycle, giving
access to a set of hues 90 degrees displaced from the original
(purple, green) set.

As you can see, NTSC color is an analog concept, which the
Apple II designer (Woz) cleverly co-opted with digital signals
synchronized in frequency and phase with the color subcarrier.
To look at single pulses spaced so closely as to lie within
a single cycle of the subcarrier as having distinct "colors"
is a misconception.  The color displayed is the result of a
bandwidth-limited interpretation of average sideband amplitudes
and phases, not an instantaneous response to a pulse.

Even the luminance bandwidth is quite limited in a composite
color signal, so short pulses need not even be distinguishable
on the screen.

What any particular receiver displays, particularly a digital
receiver, is anyone's guess, and is very likely beyond the
intention or control of the designer of the Apple II.

-michael

8-voice music synthesizer using NadaNet 2.0 networking!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."