keller@uicsl.UUCP (12/12/83)
#N:uicsl:18100001:000:1115 uicsl!keller Dec 11 21:54:00 1983 Recently in net.micro someone wrote about their quest for a high resolution monitor suitable for a high resolution graphics system they were building. The problem was that a sufficiently good monitor was $800 or so and that was just too expensive. After a lot of phone calls they final found a Clinton replacement tube that had the right long persistence phosphor and was plug compatible with some common cheap monitor. I would like to build a high-res graphics system like those found on LISP machines, PERQ's, Pixels, CPT word processors, etc. This doesn't seem to be a common hobbyist project. Clinton seems to be a good source for tubes, but what about the other components? I haven't priced the tubes, but how much should a monitor like this cost? Some company was selling a full page display option for the Apple ][ for about $1k. Surely the monitor wasn't more than $500. Seems like one could start a company making high-res BW graphics terminal for about $1500. Say 1024 by 800. Since I haven't built a monitor what are the design issues and what costs what? Shaun Keller ...pur-ee!uiucdcs!uicsl!keller
dmmartindale@watrose.UUCP (Dave Martindale) (12/13/83)
Before you go off 8Before you go off designing a graphics monitor, consider how often the screen has to be refreshed and how many lines you have to display on it. The simplest case is standard TV, or close to it: 480 to 512 lines visible, 30 frames/sec. Lots of hardware available which will do this, but it has a flicker problem: any given line is only drawn 30 times a second, even though one field is drawn every 60th of a second. If you don't want flicker, switch to a long-persistence phosphor. But then you have the problem that the screen will "smear" every time you change something on the display. To avoid this, you need to refresh every point on the screen at least 40 times/second, 60 is better. You can do this by displaying the entire image each frame, eliminating interlace. For that, you need a TV with a horizontal sweep frequency of 25-35KHz, depending on the number of visible lines and refresh rate. These are not cheap - they require more voltage and current to drive the horizontal sweep than ordinary TVs, and are thus made in much smaller quantities. And all this is just for 512-line resolution. If you want 1000 lines, you either double the horizontal sweep rate AGAIN, or start going back to interlace (with its flicker problems) or slow phosphors (with smearing problems) just to keep the sweep at 32KHz. Finally, consider the bandwidth of the video amplifiers. In ordinary 30Hz interlaced (TV) pictures, if you want 512 pixels/line, you are putting out one new pixel every 100-120ns. To distinguish the pixels clearly, you want about 10MHz video bandwidth. Going to non-interlaced 60Hz doubles that. Switching to 1024x1024 while leaving the frame rate at 30Hz quadruples it, since there are 4 times as many pixels in the same amount of time. Also, the horizontal and vertical oscillators and amplifiers need to be more stable so that the lines interlace properly. (If you are doing colour, you need finer dot pitch in the tube just so you can resolve the pixels.) All this costs money. The "ultimate" B&W bitmapped display would be, say, 1024x1400 pixels refreshed at 60HZ. This would require a horizontal scan frequency of around 65KHz and a video bandwith of 80MHz or so. This is going to cost you a LOT of money. Anyway, this is a basic guide to the tradeoffs involved. I hope this shows some of the tradeoffs that the designers of raster graphics systems of any type must make just in monitor selection. Dave Martindale
wdc@mit-eddie.UUCP (William Cattey) (12/14/83)
I hope there is someone who can answer the questions Shaun poses. I will add a few to his list: Does anyone out there know how to get high resolution COLOR monitors so that one could have 1000X1000 pixels in full color? The Japanese have been sampling monitors for high resolution color television, but how far are we from production models? wdc%ccc@mit-mc (arpa) ...!decvax!genrad!mit-eddie!wdc (uucp)
ksbszabo@wateng.UUCP (Kevin S. B. Szabo) (12/14/83)
I would just like to comment further about the `smearing' which occurs with long persistence phosphour. Robert Pike (one of the creators of the BLIT) mentioned that they used long persistence phosphour on their bit mapped display. Smearing is due to the decay time of the phosphour. The `attack' time (turn on time of the phosphour) is very short, probably the same as fast phosphors. Hence if your text is in reverse video the time for a line to erase (as might occur when scrolling) is the attack time of the phosphor, and the time for the line to become visible is the decay time of the phosphor. Apparently the effect of the phosphor fading to show a new line is not nearly as annoying as the fading/smearing of previous/erased lines. You might notice that all bitmapped displays (to my knowledge) use reverse video. This may be due in part to the above point. Caveat: I have not seen a blit, so I have not been able to judge for myself whether the long persistence phosphor is reasonable. P.S. I might have mispelled `phosphor' through this document. please ignore ( I can't stop vi and run spell! damn vi won't let you use ^z and redefine the stop character.....oh well). -- Kevin Szabo watmath!wateng!ksbszabo (E.E. U of Waterloo)
emma@uw-june (Joe Pfeiffer) (12/18/83)
Regarding smearing with long-persistence phosphors-- One of my most pleasant experiences with a terminal was on an Adage vector graphics unit programmed to emulate a TEC terminal. It had 35 lines by 132 characters, a long persistence phosphor, and a DMA link to the computer. The relevant part of all this is that it did a lot of smearing. Requesting the next screenful of data gave the illusion of the old screen fading out and the new screen fading in, with the actual scrolling completely unnoticeable. Far from being objectionable, I found this an extremely pleasant way of doing the job-- much superior to any slow-scroll terminals I've seen. -Joe P.