Large-screen analog (vector) color display: convergence?

The arcade game "Tank Battle" used a monochrome stroke writer, so no convergence issues with that. Around 27 years ago, I worked with some very high end graphics systems from a U.S. company called Vector General. They used a colour stroke writer (vector display) from a company whose name I can't remember now, but they were in San Jose. There was not any convergence issues, because they employed a special CRT type known as a Beam Penetration Display. It worked by having different colour phosphors layed on top of each other at the faceplate. Individual colours were 'selected' by modulating the high voltage supply to the final anode, which in turn modulated the 'intensity' of the electron beam, and determined how far into the layers of phosphor it penetrated, and hence which ones it excited. Obviously, its pallette was very limited - just the basic primaries and a couple of complimentaries as I recall, but still quite impressive.

The exact details might not be quite right, but good enough to give you the idea. It's been a long time since I did the factory course on it, and they were never very popular with our customers this side of the pond. I'm sure there must be references on the 'net that you could look at.

Arfa

Atari was big on "vector scan" monitors in the late 70's/ early 80's for their (arcade) video games. IIRC, Asteroids was one of the first of these. It, however, was monochrome. IIRC, Tempest, Space Duel, et al. used *color*.

At the time, the biggest annoyance in using vector displays was their inability to draw "curves". I.e., you have to resort to a piecewise linear approximation of those curves. This is fine if the opbject stays in a particular Z plane. But, since vector displays made it *so* easy to scale things (e.g., to zoom into or out of the screen), a linearization that looked good at one scale factor would look really bad at a larger factor. Conversely, drawing a good approximation at a large scale factor used up lots of unnecessary display bandwidth at low scale factors. (see below)

Someone (?) also had a vector graphic engine that only drew *arcs*. This was an interesting twist as it made curves a real possibility. However, it *only* drew arcs. So, it would have to "approximate" straight line segments (e.g., as used in *text*!) with arcs of suitably large radii. :>

I think one reason color seems so "pure" in most of these applications is that the number of colors actually used is usually very limited. And, since you aren't doing large area fills, you don't see artifacts in color gradations that would otherwise be visible, for example, on a raster scan monitor.

IIRC, the color monitors were essentially the same as raster scan monitors but the deflection amplifiers were controlled externally (via "X" and "Y"). The scan rate of most of these vector monitors was considerably less than that of a raster scan monitor of that period -- like one fourth of that! (N.B. Electrohome had a vector scan monitor that could actually *emulate* a raster scan monitor with its deflection amplifiers. I think they are/were located somewhere in Canada?).

A problem with vector scan displays is that you either adopt a "constant drawing speed" in the design of your graphics engine (i.e., slew the deflection amps at a constant rate) *or* you try to optimize the drawing *time* by maximizing the drawing speed (takes more smarts in the graphic engine). The flip side of this is variations in drawing speed -- stroke to stroke -- cause variations in the intensity of the displayed stroke. (think about it) Since the bandwidth of the vector display's amplifiers is so constrained, one would like to draw things in the minimum time necessary to allow you to get more on the screen at a given time.

Of course, nowadays, processing power makes the advantages of vector scan displays largely meaningless -- drive a raster scan display at an insanely high resolution and just *draw* the lines that you want to draw!

Interesting! I'm just about positive that Tektronix made 'scopes using peam-penetration CRTs for a limited color palette. Those CRTs can't change color quickly without a heck of a lot of fuss, because the accelerating voltage has to change by several kV, and that take a little while, maybe milliseconds. Afaik, the 'scope changed color to display a different category of data, perhaps alphanumerics, possibly a ground ref. line, multiplexed with the waveform display.

Furthermore, the CRT's sensitivity changes a lot, just about sure, so the gain of the deflection amps needs to change when color changes.

Thanks much, and best regards!

{Good grief; I must learn to check back for replies! Been away from USENET for quite a while. --nb}

On Wed, 22 Jul 2009 11:56:42 -0700, D Yuniskis wrote a very interesting reply; considering elapsed time, I'm quoting in full.

Now that pou mention it, it might have been Tempest that I saw (and played, a bit). I remember something like a hint of a tunnel into which you were looking, and a spiky creature that attached itself spider-like.

Most interesting.

Seems really silly not to be able to also draw straight lines. Reminds me of a US Navy sonar analog fire-control computer (ca. 1956) that could solve for a target moving with a constant turning radius. Unfortunately, the extra components were disabled, because the computer itself was unreliable. (I saw it at a Navy school, not on a ship.)

Indeed, in the game i saw, the palette was very limited. Convergence was amazingly good, though.

Good glory! Reminds me of a little low-cost consumer vector display game (Vectrex?); had mag. X-Y defl., and was probably monochrome. Could display little rasters. Also, the remarkable H-P wide-deflection-angle electrostatic CRT X-Y display. Its deflection amps were limited to small swings for fast writing.

Once again, much appreciated.

At a trade show, I once saw a monochrome large-screen (21 inch, maybe bigger) raster-scan monitor that did 4096 (just about sure) x 3480 (probably). IIrc, dot clock was around 1.2 GHz. One of their demo displays included one black pixel in a while field. You required a magnifier to see it.

They showed a three-view line illustration of a small boat, an engineering drawing minus most callout detail. What with the curves of a boat hull, the image was extremely susceptible to jaggies, and they weren't using anti-aliasing. Try as one might, without a magnifier, one could not see any jaggies at all.

Iirc, the resolution mas 300 dpi. Extremely impressive.

They were proud of their low-power horizontal deflection design. The deflection yoke was more like a "stator" yoke, used for X-Y displays, but dramatically different in that its individual coils extended out from the core like huge flower petals; I don't know why.

Unfortunately I've forgotten the trade name, but I think it was Megascan.

At the same trade show, Sony showed a square large-screen Trinitron, 4096 px square. Back then, they didn't have any images larger than 2K, so they tiled four different ones. A plastic housing was optional, and cost $7,000.

As to the original query, I guess I'll assume that convergence was done by some very well-designed analog circuitry. Maybe I should ask Analog Devices!

The displays were once available affordably as surplus, but I already had too much Stuff and a slim budget.

Best regards.

Yep, that's the Vectrex. Back in the '80s, a friend & I picked up two remaindered units for a song. Lots of fun.

One of the great videogame systems. I had one, and all the carts and accessories, but sold them when I was in bad financial straits.

It originally sold for $200, then dropped to $100. The games were generally very good.

The display was, of course, monochrome. However, there was an accessory pair of goggles with a rotating filter wheel (provided with the game) for color -- and 3D. (I don't remember if you could have color and 3D at the same time.)