Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!bionet!agate!ucbvax!hplabs!hpfcdc!hpfcdj!myers
From: myers@hpfcdj.HP.COM (Bob Myers)
Newsgroups: sci.electronics
Subject: Re: sync signal generator
Message-ID: <17660014@hpfcdj.HP.COM>
Date: 10 Aug 89 18:20:00 GMT
References: <89216.120136BHB3@PSUVM>
Organization: Hewlett Packard -- Fort Collins, CO
Lines: 83

>We just rolled in a couple of Ardent workstations that provide 1280 X 1024
>X 16,777,216 color resolution.  The manual only says RGB with sync on green
>60Hz noninterlaced, no RS numbers are shown but It appears to be RS-343.
>Does anybody spec the actual color resolution of their monitors?  From my
>experience, it varies considerably with the age of the CRT.

The "RS-343" designation on most monitors used with computer systems refers
only to the fact that the video signal *levels* conform to those set by
the RS-343 standard (white 0.714 up from blank, sync tips 0.286V down from
blank), and not the timing proposed in that standard.  (RS-343 was developed
as a high-resolution closed-circuit standard, and include a number of
recommended timings - I think four different timing standards, but I'm too
lazy to dig the standard out right now.  All were *interlaced*.)

I'm a little uncertain as to what you mean by "color resolution".  The number
you give for the Ardent workstation mentioned could be misleading if expressed
in this manner, as a display with 1280x1024 pixels can't possibly display
16 million colors *simultaneously*.  Now, in RGB monitors, the color
resolution is *not* artificially limited as it is in, say, NTSC color
encoding (where there were some very good reasons for imposing a limit on the
bandwidth of the color signals).  Assuming that the video amplifiers in your
monitor are up to snuff, and the dot pitch of the CRT was properly chosen,
the monitor is in fact capable of displaying 1280x1024 pixels, each a
different color.  Age should not affect this unless the amplifiers degrade
or there is damage to the CRT, especially in the shadow mask.  This latter
would affect only a certain area of the CRT, though.  Whether or not you
would actualy be able to SEE, say, 1280 distinct colors in a given line is
a completely different question - one reason that the NTSC TV standard was
able to get away with limiting the color resolution is that the human eye
isn't very good at resolving fine color detail - in other words, our "color
spatial resolution" isn't as good as our "monochrome" vision.  (Reason?
More rods, which respond to lumninance only, than cones, which respond to
certain colors.)


>Perhaps some well informed individual could post a table of common video
>formats and their associated scan frequencies, resolutions, interlace
>characteristics, etc.

This information is readily available for the various "PC" formats (CGA, EGA,
VGA, etc.), all of which (except for the new 8514A 1024x768 displays) are
non-interlaced.  I won't bother posting 'em here - partly because I can't
put my hands on those files right now (gotta get this drawer organized...
mumble, mumble...).

There are no real "standards" for the higher-end displays, such as those used
for the workstation market (as opposed to the PC industry, where everyone
simply heeds The Word: "And IBM said, 'Let there be CGA!' And lo, it was so!").
The problem is that there can be a number of different specific timings for
a given resolution, depending on the retrace times required by the monitors,
vertical refresh rates, etc..  However, a good ballpark set of numbers can
easily be obtained for a non-interlaced display as follows (examples are for
a 1024x768 image):

1. Find the total vertical period.  For a 60 Hz display, that's 16.67 msec.

2. Typically, about 5% of the total vertical time is used for the vertical
   blanking.  So, 95% of the above number gives us 15.84 msec "active" vertical
   time.

3. This active time must be evenly divided between the displayed lines.  
   So, divide this number by 768, and we get 20.62 usec per line, for a
   horizontal sweep frequency of about 48.5 kHz.

4. Typically, something like 20-25% of the total horizontal time is used in
   horizontal blanking (the percentage tends toward the high end of this
   range for the higher resolutions).  If we guess right in the middle, then
   77.5% of the total H time gives an "active" horzontal time of 15.98 usec.

5. This active time must be divided between the pixels in a single line.
   Dividing 15.98 usec by 1024 give 15.6 nsec per pixel, or a "dot clock" of
   about 64 MHz.

Just as a point of comparison, the display I'm using right now happens to be
a 1024x768 display - an HP 98545A display card in a Series 300 workstation.
It uses a 64.11 MHz dot clock, and a 47.7 kHz sweep frequency; so you can
see we got pretty close with our estimate.


Bob Myers  KC0EW   HP Graphics Tech. Div.|  Opinions expressed here are not
                   Ft. Collins, Colorado |  those of my employer or any other
myers%hpfcla@hplabs.hp.com               |  sentient life-form on this planet.