• Please review our updated Terms and Rules here

CGA colors as displayed on monochrome TTL monitors

VileR

Veteran Member
Joined
Jul 21, 2011
Messages
647
Location
Israel
hey all,

I'm not a collector of old systems (...yet?...), and it seems like the answer to my question can't easily be found online, so here I am on this fine forum.

Basically I'd like to know what the 16 CGA colors look like when translated into 16 monochrome shades.

The setup I'm talking about is a CGA/MDA dual-mode card, driving a TTL monochrome monitor, as was common back in the XT clone days (that was the combination on the first PC I ever used). Most emulators of old PCs don't tackle this kind of thing, so I've starting working on a DOSBox patch myself.. the catch is, I don't know anything about the exact correspondence of each CGA (RGBI) color to the resulting shade of green / grey / amber.

So this is kind of a mini research/educational project for me. All I've been able to find is this:

- An image showing 16 colors on a monochrome green monitor
- A video (maybe the creator posts here? :)) of an ATI Small Wonder card displaying CGA graphics on an IBM 5151 - check out 2:31 to 2:47 for the color bars test.

Interestingly it looks like the colors progressively increase in brightness, within each half of the palette... but this doesn't make much sense compared to the relative brightness of each CGA color. e.g., you'd expect color #4 (red) to appear darker than color #3 (cyan).
On the monitor in the first photo, color #6 (brown) is darkened, just like on IBM's color CGA monitors - but all other colors just follow the increasing brightness trend. I'd especially like to know if there were any implementations that worked differently.

I imagine there could've been at least two ways to do the CGA-to-mono conversion... it could either be done by the graphics card (and then sent to any TTL monochrome monitor), or on the monitor itself (if it's dual-mode and accepts color CGA signals). And the exact method could vary between different cards and monitors... so it gets still more confusing.

Here's how you can help:

  1. If you have a working setup with CGA graphics on a mono TTL monitor - and a few spare minutes:
    Could you post a nice, clear photo of it, showing the 16 color bars? (such as one of IBM's test utilities, or the "color testplate" from Trixter's CGA compatibility tester)
  2. Any technical info on how the color-to-mono conversion is done? (whether in principle, or with a specific card or monitor)
  3. Bonus question: the IBM 5155 PC Portable also had CGA graphics, but its monochrome monitor was composite (rather than TTL). Still, I'd also like to know what the 16 shades looked like on the 5155, since it would show how IBM meant it to look.
    So if you've got a working 5155, could you post a 16 color bar photo? That would be cool too.

Thank you for your time.
 
My model 25 has a monochrome monitor wired to MCGA, but the input internally is analog, would you still want that to run the color tests?

EDIT: http://img.photobucket.com/albums/v395/Evilweredragon/07212011223.jpg

Hey, very good picture, and could give me something to compare with - thanks :D
but yeah, analog MCGA/VGA works very differently, so it doesn't really give me the answers...

Anyone with mono-CGA / MGA(?) card and monitor, who could do the same thing?
(or even with an IBM 5155?)

cheers!
 
I've got an ATI Small Wonder card like the one in the video, and with the switches set for CGA emulation on a TTL mono display this is what I get:


(click for bigger)

The monitor is 18KHz mono only rather than dual-mode, so any conversion must be being done by the card rather than the monitor.

The card manual doesn't give any details, it just says "colours converted to shades". There's some slight flicker on some of the bars, so maybe it's doing it by toggling them on/off between frames, or by using moire patterns?
 
The card manual doesn't give any details, it just says "colours converted to shades". There's some slight flicker on some of the bars, so maybe it's doing it by toggling them on/off between frames, or by using moire patterns?

Yep, the card uses extreme flickering to create shades. On the long persistence phosphor used in the IBM 5151 and various other monitors, you won't even see the slight flickering.

As for the display, yes, it's not 100% emulated (50Hz vs 60Hz vsync, cursor differences, slightly different font, etc..). I assume it's based on the technology of ATI's fitst card, which purpose was mainly to display CGA graphics to MDA monitors.
 
Here is the IBM 5155, IBM CGA card on the built-in display.

http://home.earthlink.net/~russellbaker/cgatest.jpg

Thanks man. Interesting - looks like a monochrome composite screen really doesn't deal well with color modes... not only does it generate those vertical lines in place of color, but most of the color combinations make the text completely unreadable :D Guess it's the color-burst signal not being processed and noising things up?


The monitor is 18KHz mono only rather than dual-mode, so any conversion must be being done by the card rather than the monitor.

The card manual doesn't give any details, it just says "colours converted to shades". There's some slight flicker on some of the bars, so maybe it's doing it by toggling them on/off between frames, or by using moire patterns?

Guess it has to use some combinations of "video on/off" and "intensity on/off", but yes - sounds like it might be doing some quick toggling of the electron gun on top of that, to produce even more shades. On a white-phosphor monitor that could indeed create some flicker... green screen monitors had much longer phosphor persistence, so the flickering must have been far less noticeable on those.

Anyway, many thanks for the photo! Looks similar to what I've found so far, so we can at least assume that the ATI generates this "increasing brightness" pattern of shades whenever it's set to drive a 18KHz-only screen, such as yours or the IBM 5151 in the video.

I guess my next question is: Can anyone produce a similar photo, but from a different setup?

(a dual mode monochrome screen, receiving CGA-like signal in 15.75 KHz / 200-line mode... this could show what happens when the monitor itself does the conversion and assigns a shade to each color.)
 
If you are interested I can get you photos of a Compaq Portable II (which is a dual mode MDA/CGA card and dual mode green phosphor CRT combination) and of a Portable III (which is a fixed resolution amber colored 640x400 Plasma display that emulates CGA)
 
Hi RWallmow,

If you are interested I can get you photos of a Compaq Portable II (which is a dual mode MDA/CGA card and dual mode green phosphor CRT combination) and of a Portable III (which is a fixed resolution amber colored 640x400 Plasma display that emulates CGA)

that would be nice indeed - especially the Portable II in CGA mode. Not sure how the III's CGA emulation works, but perhaps it could also be relevant to what I'm looking for.

cheers!
 
DSCN0432.JPG
- Color bar test on Compaq Portable II
DSCN0435.JPG
- Contrast test on Compaq Portable II
 
This one surprised me on the Compaq Portable III, its CGA emulation is VERY poor, see photos below.
DSCN0433.JPG
- Color bar test on Compaq Portable III
DSCN0434.JPG
- Contrast test on Compaq Portable III


Full size images of these can be found in the respective PCs gallery album in the link in my signature.
 
Compaq Portable II:
very nice photos - it's weird how similar it is to the ATI Small Wonder... both use an "increasing brightness" pattern, don't darken the brown (#6), and show dark grey (#8 ) as black.
There may be a few subtle differences, but it still looks like the increasing brightness pattern was a sort of standard with Mono/CGA TTL combo cards... even though it doesn't line up well with CGA color luminances.

Compaq Portable III:
wow, I'm not even sure I get what's going on here. :p Looks like it does 3 brightness levels at most, but it doesn't even seem consistent in assigning those levels to palette colors...
e.g., in the color bar test, all solid colors except brown/yellow look the same (including black). But in the contrast test, the hi-intensity colors actually look darker than their low-intensity counterparts, and it's black/dark grey that are the odd ones out.
So yeah, pretty poor CGA emulation, but kinda curious in itself.

Thanks for these!
 
Let's be realistic--the Compaq P II uses a CRT, but the P III uses a gas-plasma display. The latter really doesn't have much capability to adjust brightness selectively. Different technology.
 
Let's be realistic--the Compaq P II uses a CRT, but the P III uses a gas-plasma display. The latter really doesn't have much capability to adjust brightness selectively. Different technology.
Agreed, I didn't expect GREAT CGA emulation, but I did expect better, it could have still "flickered" or dithered for more emulation of colors/shades.
 
Compaq Portable II:
very nice photos - it's weird how similar it is to the ATI Small Wonder... both use an "increasing brightness" pattern, don't darken the brown (#6), and show dark grey (#8 ) as black.
There may be a few subtle differences, but it still looks like the increasing brightness pattern was a sort of standard with Mono/CGA TTL combo cards... even though it doesn't line up well with CGA color luminances.
The Portable II uses a dual mode MDA/CGA card, and a true dual mode CRT, so CGA is not emulated, its true CGA, just displayed on a monochrome display.

And, you are most welcome for the screen shots, I could try to capture high quality high resolution RAW images if you really need for your research if these are not of high enough quality from my little point and shoot.
 
The Portable II uses a dual mode MDA/CGA card, and a true dual mode CRT, so CGA is not emulated, its true CGA, just displayed on a monochrome display.

Yeah, that's easy to tell from the obviously 200-line text mode :) I suppose that the method of generating the monochrome shades still has to be somewhat similar though, since TTL signals don't give you a lot of wiggle room.

In fact, that's what makes this photo especially helpful for me: my old XT-clone used to have the same sort of setup, a MDA/CGA card with a dual-mode green monitor. Unfortunately I have no idea about the make and model... but I do remember that the card switched between CGA and MDA in software, by running a tiny file named ST100A.COM.
(could that be a clue for some of the expert excavators on here, maybe?)

Anyway, thanks but there's really no need for even higher quality images - the full-size versions on your site are definitely more than good enough!
 
If I remember right, the signal conversions done in the Compaq Portable II are the same ones used on the monochrome EGA/VGA implementations... which Ferarro's "Programmers guide to the EGA/VGA" lists as:

L=0.3R+0.59G+0.11B

Wish I still had my H-161. It's CGA implementation for the internal mono display used a simple ladder DAC to turn the ttl signal into beam strength.... which I think worked out to similar to that. I think it made intensity be a 33% overall signal boost... If you could track down the assembly instructions for that machine, the parts used on the CGA board could give you the 'mechanicals' of how it worked.

Man I miss Heathkit.
 
If I remember right, the signal conversions done in the Compaq Portable II are the same ones used on the monochrome EGA/VGA implementations... which Ferarro's "Programmers guide to the EGA/VGA" lists as:

L=0.3R+0.59G+0.11B

If that was the case then you wouldn't have this simple increasing brightness trend with the palette colors - e.g. you'd expect color #4 (red) to appear darker than color #3 (cyan):
Color #3 - R=0, G=2/3, B=2/3 ==> L=0.4666...
Color #4 - R=2/3, G=0, B=0 ==> L=0.2

The Portable II clearly isn't showing it that way (although that formula is probably true for evildragon's analog monochrome MCGA/VGA shot from the previous page).
 
Yer right -- it appears to just do 0..7 in increasing brightness instead of even trying to do proper color conversion. My bad... Evildragon's pic does show it in action though quite nicely.

I was actually "stuck" on monochrome VGA from 1990 until sometime around 1995 -- as at that time I was too cheap to spring for a VGA color display.
 
Back
Top