11 May
2001
11 May
'01
3:25 a.m.
Well, I'm not sure you're entirely wrong here, but the ghosting is probably more
a result of poor signal termination than anything else, and the magnitude of the
reflection from a full-swing digital signal is much more likely to show up than
what you'd get form a 1-volt p-p analog signal. Moreover, since you've got much
smaller transitions on the analog video signal, and since you've got an
automatic gain control in the monitor to offset the losses, the line losses
aren't as much a factor as they'd be without it. Digital signaling was OK back
in the text-only and monochrome era, but once things went to full-spectrum
color, it was necessary to present them in analog. The EGA had 16 (?) colors
(maybe that was 16 shades of each of three colors), which already required some
digital=>analog processing at some stage in the process, and, from what I've
observed, it had to be on the adapter board, since the signal from there to the
monitor was already a small-swing analog signal.
Gosting is another problem that's never plagued me in this arena. Normally it
can be dealt with by proper termination of the video signal. Some monitors rely
on a 150-ohm resistor in the line and only provide that same amount themselves,
while other have a 75-ohm to ground/return. To accompany this some adapters
have a divider rather than simply a series resistor to the output. It depends
on what's in the combination, how well it will work, but I've not been
sufficiently bothered by any mismatch to warrant hacking either circuit.
Dick
----- Original Message -----
From: "Iggy Drougge" <optimus(a)canit.se>
To: "Richard Erlacher" <classiccmp(a)classiccmp.org>
Sent: Thursday, May 10, 2001 5:26 PM
Subject: Re: Converting TTL monitor to Analog
Richard Erlacher skrev:
What sort of digital interface would such devices
use? They'd have to have
lots of wires, e.g. 26 or so pairs, if the thing is using 24-bit color.
Perish the thought someone would use more color depth than that! It's hard
to imagine they'd do that. The number of wires that have to transport
high-frequency signal is the reason the analog is still so difficult to
surpass.
After all, a 26-pair shielded cable with an
appropriate connector, made in
the US, would immediately become the most costly component in a computer
system. I doubt a computer maker would go for that, as the 20" monitor
(@~$300-$400) is presently the most costly component. That would essentially
make the 20" display and cable more than 2/3 the cost of the computer.
But isn't the digital signal less dependent on absolute signal integrity?
After all, if there can only be two levels, making out the difference can't be
all that difficult. This also seems to hold true WRT my successful use of
long, thin, unshielded cables for EGA and CGA monitors without any ghosting.
--
En ligne avec Thor 2.6a.
I dunno, I dream in Perl sometimes...
--Larry Wall in <8538(a)jpl-devvax.JPL.NASA.GOV>