Re: [stella] Number of Scanlines, was: F8 Bankswitching experiment

Subject: Re: [stella] Number of Scanlines, was: F8 Bankswitching experiment
From: "Mark Graybill" <saundby@xxxxxxxxxxxx>
Date: Fri, 6 Dec 2002 23:44:31 -0800
> > Understood. Now, I have a question in that matter: How many scan lines
should I have?
> > Reading Kirk's tutorial and the Stella guide I learned that in total I
should have 262 scan lines (3+37+192+30). I checked one game using z26 -v9
and noticed that it has 280 scan lines. Why would that be? Shouldn't it have
> Glenn answered this one sometime ago but I can't find it in the archives
now. In the Atari Graphics and Arcade Design book for the Atari 8-bit/5200
it states...
> "The electron beam retraces its 192 line path sixty times a second. A
television set actually scans 262 scan lines, but the average set can only
display slightly more than 200 lines. The area above and below your
rectangular playfield contains some of these extra lines. "
> I think Glenn said that 200 scan lines for your kernal should be okay but
I can't be sure.
My experience in pushing the limits of NTSC video with the 2600 is limited,
but I'll draw on my non-2600 experience here...

The 192 lines in the article I believe refers only to the lines where video
is drawn, and omits counting the scanlines in the vertical blank and
"overscan" timeframes.

For a stable display in NTSC on a standard TV set, you should have either
260 or 262 full scanlines in a non-interlaced display (60 frames per
second). This includes the drawn video area, the blank areas, and the number
of scanlines for which a vertical sync signal is sent. I've found that most
sets will handle down to 258 scanlines per frame OK, and up to 264
scanlines. Beyond that it starts to depend on the set itself. Most composite
monitors are more forgiving than most TV sets, though  current TVs tend to
be pretty flexible, due to the fact that they often don't base their timing
off the 60Hz power signal any more. I've gotten some pretty goofy video
signals to display on some current TV sets (like 397 scanlines of 40uS, for
instance. the flyback transformer whined like a banshee but the display was
rock solid. On precisely _one_ make of set that I tested.)

The number of  total scanlines that will produce a stable display will also
depend on what the rest of the timing parameters are like. They're all
related. With the 2600 your scanline time and horizontal blank and
horizontal sync times are already established for you, so your flexibility
in adjusting the number of scanlines is limited. To get a stable display
that doesn't "breath" or isn't "seasick", it's best to have exactly 262
lines. With, say, 260 scanlines the 2600's frame time is off, you would have
to be able to extend the time between the TIA's horizontal blank and sync
pulses by a few microseconds to get a nice display at 260 scanlines. But
chances are you'll still get a decent display on most sets, and if you're
looking at the signal after it's been folded, spindled, and mutilated by the
RF modulator and a long thin video cable with tin connectors you may not
even notice the difference from a regular 262 line display.

There is a fair amount of flexibility in the time for vertical sync and
vertical blank periods. In non-Atari video circuits I've run VSYNC for
periods as long as 8 scanlines with stable displays. In some circuits this
is the standard period. The blank period after should be at least 30
scanlines in my experience, though I usually run it longer for generated
video. The blank period after the drawn video display is also flexible. I've
run video down to within 4 scanlines within the start of VSYNC, though you
won't see it all onscreen even if the display is stable (which is was for me
on a variety of equipment, though I wouldn't make any guarantees.) It's best
to keep at least 20 scanlines of black before a vertical sync pulse.

-Mark G.

Archives (includes files) at
Unsub & more at

Current Thread