[Coco] S-video Prospects...
Gene Heskett
gene.heskett at verizon.net
Thu Jan 26 22:09:51 EST 2006
On Thursday 26 January 2006 19:59, Stephen H. Fischer wrote:
>Hi,
>
>This is the first I have heard about "A missing HSYNC", I never have
> seen it.
>
>As others clearly know about it, I must be lucky.
>
>I wonder if my TV (and other top of the line sets) keep a oscillator
> running and then phase lock it or similar when the source HSYNC
> arrives.
>
>Gene, perhaps you can educate me on this.
Somebody rang?
Yes Stephan, they do. The locking of them is done in a bit of trickery
during the vertical synch pulse. If you have a tv that you can make
the picture roll downward a bit, you'll see if you turn the brightness
up to max, what looks a little like the feather end of an arrow pointed
to the right between the bottom of the upper pix and the top of the
bottom pix.
This little cut between the fatter block, and the end of the synch pulse
to its left is about 2.4 millionths of a second, is put in there so
that when the synch pulse drops back to active state at the left edge
of this black block, its exactly in time with the hsync, but twice a
line for those 3 lines of active v-sync. You are seeing the one in
between the active ones in each vertical interval. There are 3 lines
(6 pulses) before, and 3 lines (6 pulses) after the 3 lines of vsynch,
of what we call equalizing pulses, and that block you see is also
duplicated in the hsynch's time, again so that the leading edge is
properly aligned in time to maintain h synch lock. If you can see the
hsynch and the vsynch at the same time, this block is the hsynchs
narrow set, with the leading edges matched to hsynch. This was worked
out originally in black and white days even before color was shoehorned
in by some clever math because of the alternate field interlacing done
to conserve bandwidth without dropping the vertical rate so low that it
flickers.
In the pro business, we buy monitors that can do this synch display in
both directions so we can quickly check how well its working just by
pushing a button. Handier than a loaf of sliced bread or bottled beer
to us.
FWIW, if all the timebases are stable, and this has now been mandated by
the commission for about the last 35 years, if a tv signal is looked at
with a spectrum analyzer, and the bandwidth of the analyzer is turned
down to very narrow, one can see that the sidebands generated by the
video consist of a spike of info every 15,734 hz, and that each of
these is straddled by lower level signals spaced 59.94 hz away from the
15,734 hz spikes. For B&W, those are 15,750 and 60 hz, a never mind
difference.
In Black and white, there is a huge amount of virtually empty space
between these 15,734 spaced groups of signals, so the color subcarrier
frequency was chosen to put the color information into these empty
areas. So all these frequencies are in fact locked to each other by a
very precise set of math formulas turned into strings of divider chips
and phase locked loop oscillators on our end of the signal. It works
surprisingly well most of the time.
Not widely known and not widely enforced is the fact that one of the
color phases has twice the bandwidth of the other, again because of
what the eye was sensitive to as much as any other reasons. I'd say
that 99% of the color encoders in use today generate a full, 2.4 mhz
bandwidth color signal, or up to 1.2 mhz either side of the base 3.58
mhz subcarrier. It's sent as the summed result of a pair of 4 quadrant
multipiers, one driven by the subcarrier, and one driven by the
subcarrier delayed 90 degrees, with the color signals mixed and gained
into them thru a complex matrix of resistors, The end result is a
signal that can be any amplitude and phase when referenced to the burst
reference we also send on the back porch of the horizontal blanking
interval.
Since the color is sent as a plus or minus from the black and white, let
me remind everyone that if your set is miss-adjusted or the crt is
going sour such that you cannot get a decent grey scale black and white
image, then no amount of tweaking of the color is going to fix it to
look really right.
And (offtopic) surprisingly, when watching the encoder output on a
vectorscope, there is no one alive who can look at the vectorscope and
tell from that faces color signal, going generally up the -I line on
the scope, whether it belongs to a white, red, or even a very black
person. To the vectorscope, which ignores brightness and only shows
the color and intensity of the color, they are all exactly the same.
I suspect that a Bell Tel. Fellow named Claude Shannon and his theories
had a lot to do with that math way back then. When that was done, I'm
not sure if the fourier transform had been discovered yet or not, and
full, realtime useage of that had to wait many years for both gigabit
ecl logic and simplfied methods, the most famous of which today is
probably the butteryfly transform, commonly used because you can check
your results by running it again on the output data and you should get
the input data back again. I was gonna port that to the coco way back
then, but got busy chaseing my current missus and putting table driven
crc checking into rzsz instead, which had more benefits for everyone at
the time. So much for good intentions. I hear tell there's a road
paved with them. :)
I hope I didn't lose too many of you. I don't think I did because the
folks who read this list are I feel, of well above average IQ's, I've
seen it demo'd many times.
--
Cheers, Gene
People having trouble with vz bouncing email to me should add the word
'online' between the 'verizon', and the dot which bypasses vz's
stupid bounce rules. I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2006 by Maurice Eugene Heskett, all rights reserved.
More information about the Coco
mailing list