18

I'm reading a book called The Friendly Orange Glow, about the PLATO multiuser computer system developed by the University of Illinois in the late sixties, which is fascinating at several levels as a microcosm of an alternate universe of computing. The particular aspect I'm looking at here is the source of the orange glow, the plasma screen, a key enabling technology for the system.

PLATO used a 512x512 monochrome bitmap display, which apparently cost $2500. (I'm not clear whether that's just for the plasma screen itself or for the entire terminal, which contained additional components such as a 16x16 touchscreen input.) That's a quarter million bits (32 kilobytes) of data. On a CRT or LCD, it would need to be backed by a quarter million bits of video memory. According to https://en.wikipedia.org/wiki/Magnetic-core_memory in the early days of magnetic core, that would've cost a quarter million dollars; even by the early seventies, it would still have been $2500 of video memory, doubling the total cost of the display.

But apparently a plasma screen has the remarkable property that it does not need refresh, i.e. it doubles as its own video memory. And for a few years, that was a key advantage.

I'm trying to reason about what it would be like to use such a system. For example, scrolling would not be a primitive operation. You would always have to redraw the display based on internal data (whose format would be optimized for compactness, not speed of redraw), so it would be natural to design around 'page at a time' instead of continuous scrolling.

I'm also trying to figure out what would be the implications of the display persisting without refresh. It occurs to me that might depend on timescale; just because the display doesn't need refreshing every 1/60 of a second, doesn't necessarily mean it persists forever.

Just how long was the image on a 1970-era plasma screen stable for?

rwallace
  • 60,953
  • 17
  • 229
  • 552
  • 6
    PLATO applications separate the screen into regions and then update only the affected region based on user input. You can see for yourself using the PLATO instance at https://www.irata.online/ – Brian H Feb 21 '22 at 22:24
  • 4
    If all you're trying to do is get the experience of using a screen that serves as its own video memory and doesn't need a refresh, just use a Kindle or anything else with an e-ink display. As for the plasma experience, I have a perfectly functional Compaq lunchbox, haven't fired it up in a while but IIRC the fade is somewhere in the 50-250ms ballpark? Something like that. It's in my storage unit. But I'd be happy to take a video of it for you if I have time to go out there this week. But also there were plasma TVs as recently as 2014; they were the top tech for a couple years before OLED. – Jason C Feb 22 '22 at 17:29
  • 1
    Or if you find any old laptops or monitors where you have to have your head in a sweet spot to see it, and if you moved like a half an inch everything just disappeared; those were also older plasma. Those were really annoying times, heh. That is not a technology that I miss at all. – Jason C Feb 22 '22 at 17:35
  • 3
    As a former PLATO user, I don't remember anything weird about the screen. – Organic Marble Feb 22 '22 at 18:43
  • 2
    @JasonC the narrow viewing angle was in fact a security feature keeping plasma screens a thing over all the years. No unwanted peeking from the sides. – Raffzahn Feb 22 '22 at 22:22
  • 1
    @Raffzahn Heh; it was a "security feature" in an "it's not a bug it's a feature" sense. The narrow viewing angle was an eventually-overcome technological limit. But as a useful side-effect it happened to be handy in security contexts. There was never, however, an initial intent to invent a narrow-viewing-angle display technology. There were engineers spending late nights racing to figure out how to improve the viewing angle, and some security people on the side saying "well, ya knooooww......". – Jason C Feb 23 '22 at 00:01
  • 1
    Here, check this out, it's awesome: History of Plasma TV Reflected By Patents. Btw, for historical context, the first prototypes were in 1964 which is also the same year a CRT was first built directly into a computer as a monitor; i.e. the addition of monitors to computers was a relatively new idea, and CRT vs plasma were two of the initial competing ideas, with CRTs winning out (I think because the tech for plasma wasn't good enough yet). – Jason C Feb 23 '22 at 00:10
  • 1
    @JasonC Plasma terminals build in the 1970s and 80s were made exactly for that narrow viewing angle - and it was kept that way on purpose. These terminals sold for a premium. I remember a version by Siemens of ca. 1979, selling over 25,000 Mark, that#s like 8 times of a regular one, comparable to a good Mercedes car :) The improvements you're talking about are a very different line of development, one toward colour and usage as TV (still have one :)) – Raffzahn Feb 23 '22 at 00:12
  • 1
    @Raffzahn Right. But when Donald Bitzer and his two pals invented the plasma display in 1964, they didn't start with "we're going to invent a narrow-viewing angle display for security". They started with "we're going to invent a display that makes computer graphics look better than a CRT". The narrow-viewing angle was used as a security feature -- and obviously it was nurtured in that market, but it wasn't meant as a feature, it was a flaw that turned out to have a use. – Jason C Feb 23 '22 at 00:16
  • All I'm saying is, the way you said it made it sound like it was a feature right from the drawing board, but really it was a flaw in its original intent that ended up finding a use as a feature in a different market much later. Sort of like how Wellbutrin was designed as an antidepressant but turned out to be useful for quitting smoking and re-released as Zyban, lol. That's all I mean. :) – Jason C Feb 23 '22 at 00:18
  • @JasonC Sounds fine - but then again, we're talking terminal use here, don't we? And for terminal use there were only two use cases (beside the very early and specific of being it's own memory): a) the narrow view angle b) the ability to create a portable flat screen with better viewability than early LCD. Neither gains from a wider arc. – Raffzahn Feb 23 '22 at 01:01

1 Answers1

18

But apparently a plasma screen has the remarkable property that it does not need refresh, i.e. it doubles as its own video memory.

This depends of course on design, as there are many ways a plasma screen can be built and operated. For example, plasma TVs have next to no persistence. But for terminals a different approach would be needed - one that uses the ability to use a higher voltage to ignite a cell and a lower one to keep it lit.

A plasma display is basically the same as a fluorescent lamp. Like these it needs a high voltage to ignite, but a much lower voltage to keep 'burning' (*1). Except, it's not one fluorescent lamp, but an array of thousands.

For a plasma display, operation is done by a grid of horizontal and vertical wires located on front and rear of the display, crossing over each cell.

To ignite:

  • a positive voltage is applied on one horizontal line,
  • a negative voltage is applied on a vertical line,
  • both a bit more than ignition voltage;
  • the cell where both cross will light.

To keep a picture displayed, all lines will get a voltage applied a bit over half the holding voltage, resulting in all ignited cells continuing to create light, all othern staying dormant.

And yes, this only needs storage for a single drawing command at a time. In addition the commands can be sent in any sequence, as it's fully pixel addressable. Except for the clear command, that is :))

Just how long was the image on a 1970-era plasma screen stable for?

As long as the hold voltage was applied.

And for a few years, that was a key advantage.

There were more advantages that kept them being available even when memory became low cost. As already mentioned, it allowed extremely high resolutions (at the time) without large refresh memory. Plasma was for example used for high security terminals, as the viewing angle was rather narrow (*2) while the display was at the same time quite bright within that angle, allowing daytime usage without anyone peeking in from the sides. A typical use case would be in banks.

In fact, plasma terminals used by banks well into the 1980s when they were equipped with full memory, so they can operate exactly like regular terminals, but keeping the display advantage.

Similar for military/security usage - like in embassies, as they avoid a common attack channel: due their static nature they do not emit any repeated RF signal that could be captured to 'look' at the screen (*3).

I'm trying to reason about what it would be like to use such a system. [...] You would always have to redraw the display based on internal data

That's much like any other terminal using persistent storage - think Tektronix 4010 et.al.

so it would be natural to design around 'page at a time' instead of continuous scrolling.

Which for most parts wouldn't be a big deal. Scrolling was a non-issue for most mainframe applications anyway, as they usually operated in block mode.

The only real restriction compared with terminals with local storage would be that it was impossible to redraw just a portion of the screen. Due the way persistence is created, it can not be switched off for selected points/areas. (*4)

just because the display doesn't need refreshing every 1/60 of a second, doesn't necessarily mean it persists forever.

It will as long as hold power is applied.

*1 - It's the filling of Neon (for light conversion), Mercury (for energy conversion) and Nitrogen (creating a useful hysteresis).

*2 - It took quite some time to widen the viewing angle to make plasma TV a useful application.

*3 - A measure found only at embassies of rather wealthy nations.

*4 - If at all for whole horizontal or vertical regions, so it might be possible to clear for example the lower half, but I do not know if that was ever used.

Toby Speight
  • 1,611
  • 14
  • 31
Raffzahn
  • 222,541
  • 22
  • 631
  • 918
  • 1
    Could plasma display contents be corrupted by things like the light from a camera flash? By my understanding, neon bulbs' striking voltage can be effectively reduced by the presence of light. – supercat Feb 21 '22 at 22:37
  • 2
    @supercat Never noticed anything alike. While theoretical possible, it may still be a hard to produce. it would need a lot of light to do so. – Raffzahn Feb 21 '22 at 22:49
  • So for programming such a display, the two primitive operations were 'write these pixels at these coordinates' and 'erase' but erasing was done by briefly turning down the voltage, so it always applied to the whole screen at once? That would be an interestingly different model! No mice, for example, because you couldn't xor a mouse pointer with the contents of the screen. You could do hypertext, but you would have to do something like numbering links and selecting them with keyboard input. – rwallace Feb 21 '22 at 23:12
  • 1
    @rwallace Well, yes, except the drawing commands were more complex, like set position, draw string, draw line and so on. None of those needed much memory, beyond some registers to keep track of a line or an area that is. As said, technical clear was only possible for a whole horizontal or vertical area, but I belive it was only done as whole screen. Further, mouse wasn't an issue, it was touchscreen. Like modern tablets don't do a mouse cursor - after all, ones finger already points where the touch will happen. No need for a mouse pointer as a crutch :) – Raffzahn Feb 21 '22 at 23:26
  • True! Though the touchscreen was only 16x16, so you couldn't use that for something like a drawing program. Just to check to make sure I understand this correctly: could you erase a single pixel? You could set a pixel by raising the voltage on the corresponding X and Y wires. Could you erase a pixel by slightly lowering the voltage on the X and Y wires? – rwallace Feb 21 '22 at 23:40
  • 1
    According to Electronic Information Display Technologies, page 92, it looks like the answer is yes, you can set a pixel with one kind of pulse along X/Y wires, erase with another kind. So as far as the fundamental capabilities of the hardware are concerned, you could set or erase arbitrary quasi-rectangular areas just by selecting which wires to activate, though whether the interface provided that is a different question. – rwallace Feb 22 '22 at 00:31
  • 3
    @rwallace I would be careful about the erasure part. Lowering the voltage on one axis will lower it for all pixels on that row or column, carrying the risk to loose more than just the desired pixel. As mentioned, there are many ways to create plasma displays, so unless the cited article is explicit about the PLATO IV display (that's the one from the 1960s, the later are of different design) , id' still be sceptical about deleting a pixel. Not to mention that deleting and overdrawing is a way too complex operation software wise. A look at the command language might help as well. – Raffzahn Feb 22 '22 at 00:47
  • Right. But if you couldn't erase pixels, the system would be much less useful, e.g. no cursor. Here's another reference that seems to confirm PLATO could do such: search for 'mode erase', about 1/6 of the way down: https://irata.online/assets/s0ascers-045c83081e9ada2008378c3ae6aa62564b213a71decf9fe04608909b91d20ad1.html#C3 – rwallace Feb 22 '22 at 04:49
  • 1
    @rwallace Could be, or it just refers to inverted drawing. Maybe more important, the paper does describe a terminal interface working on colour (3.1.1.6) as well as allowing a screen copy (3.1.1.7). So this may cover commands not available on orange plasma stations. There is no real date or version on this paper, but the earliest change date is 01/03/84, so this might be of later/enhanced origin. Mind you, I'm not saying it can't be done, it would be an interesting design for a terminal without a screen buffer. – Raffzahn Feb 22 '22 at 09:13
  • 2
    @Raffzahn: I would think it would almost certainly be possible to design a display which could handle individual pixel erasure, but displays that don't need to accommodate that could be a lot cheaper. Further, I'd expect that erasing a few complete rows and redrawing them could often be faster than trying to erase pixels individually, since the time to reliably extinguish an arc is longer than the time to establish one. – supercat Feb 22 '22 at 13:58
  • 1
    @rwallace: Not all concepts in that document would be applicable to all devices (color, for example, would be meaningless on a plasma display). There are ways of handling cursors without being able to erase. I remember seeing someone do text editing on a storage-tube-based terminal which couldn't erase anything less than a full screen. IIRC, each row of text had a stroke below it, and a row worth of space below that. The system could repeatedly draw the part of the stroke representing the cursor position (making it flash brightly), and changes to a row would be shown... – supercat Feb 22 '22 at 14:06
  • 1
    ...in the blank space below it, with deleted parts of the row above being struck out, in a manner somewhat similar to the way people would edit pen-and-paper manuscripts. – supercat Feb 22 '22 at 14:07
  • @supercat The question is not what could have been done, or what was done later on, but what was the system of these first generation Plato terminals. What-If stories should reside on some RF places. – Raffzahn Feb 22 '22 at 14:48
  • 1
    I think some amount of explanation is useful for why these terminals didn't support pixel-erase. I came looking in comments for any technical reason why erase would have to be whole-screen or whole-region like your answer assumed. The interesting points were: (1) time to reliably extinguish is much longer than ignition. (2) lowering the voltage on one row would extinguish a whole row (but still not a whole region)... unless you raise the voltage on all columns except the one you want to clear. So it makes sense that pixel erase would take extra hardware and thus be something to omit. – Peter Cordes Feb 22 '22 at 22:00
  • @PeterCordes the second part it is - it would not only need more hardware, it would also need rather complicated control. It may not be impossible, still, I would rather like to find any reliable source before assuming that it was done. – Raffzahn Feb 22 '22 at 22:03
  • I'm not assuming it was actually done, I was just wishing (/ suggesting) that your answer contained that summary of why single-pixel erase was hard and therefore not done in this or most other plasma terminals. – Peter Cordes Feb 22 '22 at 22:06
  • @PeterCordes a) it was not asked for; it's only a topic within the comments b) it's already covered due the basic workings of an X/Y grid used ti light/keep alight - as you have noticed yourslef; and last but not least c) since I have neither information aboutthis feature being available at all (for the original version that is), I do not think it would add. – Raffzahn Feb 22 '22 at 22:19