Re: Apple II Video Generator

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

Immortal Nephi <Immortal_Nephi@hotmail.com> wrote:
> On Saturday, October 20, 2012 10:15:19 PM UTC-5, Kevin wrote:
>> ugh stupid google ... let me confuse you some more 10 HGR2 20 HCOLOR = 2
>> 30 HPLOT 0,0 40 CALL -3082 omit that part, I just hit enter while
>> editing the message, its the same, I thought I saw another waveform
>> packed in there but it was just noise. >Let's talk about luma wave while
>> color burst is NOT enabled. >How do NTSC monitor interpret square wave
>> into monochrome? I suspect all 15 >colors run at 7 MHz as true grayscale
>> as monochrome. >If NTSC monitor's frequency is higher enough to sustain
>> 14 MHz instead of 7 >MHz, then grayscale does not show and true black &
>> white monochrome shows. >Are you positive sure monochrome does not use
>> luma wave? >What is the filter name do NTSC monitor use to display
>> grayscale as monochrome? its pretty simple, NTSC is a analog signal, the
>> apple is sending out a digital signal so its either on or off, thus a
>> black or white only mono signal. In normal video applications that
>> voltage can be anywhere it wants to be between black and white, giving shades of grey.
> 
> I understand, but NTSC monitor does not provide sufficient information
> how square wave is converted into luma and chrominance waves.

Nor does it need to. 

The luminance response of any monitor will simply be the result of its
analog response to the video signal. 

If the video signal is changing faster than the monitor can respond, then a
partial response depending on the filter characteristic of the monitor will
be what is seen. 

For most purposes, responses to pulse signals far above the nominal high
frequency cutoff will result in displaying the average value of the pulses
in a moving window whose width varies inversely with bandwidth. 

> I will discuss chrominance later.  Let's talk about luma.  NTSC monitor
> is supposed to show only five colors in monochrome.  Five colors are
> black (0%), dark gray (25%), gray (50%), light gray (75%), and white (100%).
> 
> The NTSC monitor is supposed to use luma by recognizing the percentage between 0% and 100%.

That's a characteristic of Apple video as interpreted by an NTSC monitor
(with restricted bandwidth). 

> Let's talk about 16 colors.
> 
> If color=1, 2, 4, 8, 5, & 10 is chosen, then dark gray (25%) is used.

No. All single-bit colors are 25% gray: 1, 2, 4, 8.

> If color=3, 6, 9, & 12 is chosen, then gray (50%) is used.

No. All two-bit colors are 50% gray: 3, 5, 6, 9, 10, 12.

> If color=7, 11, 13, & 14 is chosen, then light gray (75%) is used.

Yes. All three-bit colors are 75% gray. 

> If color is chosen, then white (100%) is used.
> 
> Are you going to agree that all 5 colors in monochrome run at 7 MHz inside NTSC monitor?

No. You have been discussing the 16-color set, for which the pixel clock is
14MHz. 

> I do not know how, but I try to figure out.
> 
> If NTSC monitor runs monochrome at 7 MHz, then 80 column text is 75% readable.
> If NTSC monitor runs monochrome at 3.58 MHz, then 80 column text is 50%
> unreadable and 40 column text is 75% readable.
> 
> If RGB monitor runs monochrome at 14 MHz, then both 40 / 80 column text are 100% readable.

Actually *readability* is a psycho-visual phenomenon that depends on a lot
more than monitor bandwidth. It is enough to say that sharpness of vertical
edges and contrast of single pixel-wide lines is reduced as bandwidth
decreases. 

Since most analog color monitors are much like TV sets, their luminance
bandwidth is less than 4MHz, rendering 80-column text unsatisfactorily. 

As luminance bandwidth is increased to 7MHz or  so, 80-column text becomes
quite usable, though single pixel-wide vertical strokes in characters will
have lower contrast than horizontal strokes. (Since the monitor can respond
only partially to a pulse only as wide as a 14MHz clock period.)

Around 14MHz luminance bandwidth the monitor is able to fully respond to
such a fast pulse, resulting in 80-column text that is uniform in contrast
and therefore "very good". 

Rise time describes the response of a bandwidth-limited signal channel to a
step input, and therefore is useful in quantifying the channel's response
to sharp edges. Rise time, for channels with gradual high frequency
rolloff, is usually approximated by: Tr = 0.35 / f, where f is the
frequency where channel response is reduced by 3dB--that is, the usual
definition of bandwidth. 

As you can see, a monitor with 14MHz luminance channel bandwidth has almost
three rise time intervals to achieve almost full response to a pulse as
wide as a 14MHz clock period. 

A monitor with 7MHz luminance bandwidth would have only about 1.5 rise time
intervals to respond, and it's peak response would be less than 100%,
resulting in somewhat lower contrast vertical (single pixel) strokes, but
still quite readable. 

A monitor with only 3.5MHz luminance bandwidth (the typical case) would
have only about 0.75 rise time intervals to respond, and would display much
lower contrast vertical strokes--enough to render the 80-column text
display unusable by most people. 

-michael - NadaNet 3.1 and AppleCrate II: http://home.comcast.net/~mjmahon