PDA

View Full Version : What is the proper aspect ratio adjustment of the IBM 5153 CGA monitor?



Trixter
March 18th, 2016, 08:27 PM
For decades, I was under the impression that the viewable area of an IBM 5153 CGA monitor (and, possibly, all CGA monitors) was supposed to look like it was evenly spaced from all sides of the bezel, such that the overscan border was of an equal thickness on all sides. My own hardware has been configured like that for decades, and looks like this:

http://i.imgur.com/fX0hdy7.jpg

Further reinforcing my impression of this was the fact that DeluxePaint from the 1980s has built-in aspect ratio correction, which you could turn on and off as needed. I used it to create the following aspect ratio test picture, which displays a perfect circle/square with the monitor adjusted as shown directly above. That test picture looks like this:

http://i.imgur.com/vOYrWf3.jpg

(The wording in the left circle is incorrect, please ignore the wording for now)

However, it was recently brought to my attention that the designers may have intended the display aspect ratio to be 1.37 (http://www.reenigne.org/blog/what-is-the-cga-aspect-ratio-exactly/), which is closer to NTSC than I would have expected. To match this, it means the correct adjustment of the viewable area would have very thin overscan borders on the top and bottom compared to the sides, like this:

http://i.imgur.com/V3r1DqW.jpg

(Again, note how there is almost no overscan on the bottom.)

Which is correct? I've checked the 5153 tech manual as well as the SAMS Computerfacts document and there is no mention of what the proper adjustment of the viewable area is supposed to be. I thought I could answer this question by grabbing samples of the 5153 from youtube videos, images, and old magazines in my collection, but after two hours I could only gather a decent sample size of 22 images (where the viewable area/overscan border were very clearly visible) and the results were inconclusive.

What did IBM's designers intend? What is the correct adjustment for viewable area and overscan on an IBM 5153 CGA monitor?

(I'm hoping this post brings some of the old monitor repair people out of the woodwork, as it's possible the only people who might know are those who directly serviced these monitors...)

vwestlife
March 18th, 2016, 09:41 PM
I don't know the exact answer, but the situation is additionally complicated for Tandy 1000 users, due to Tandy's 225-pixel-high text modes (which nicely eliminates the visual claustraphobia of having the bottom of descenders touch the top of ascenders on the line below them) vs. the standard 200-pixel-high graphics (and optionally text) modes. I usually adjust an RGB monitor so that the 225-line mode is approximately what you show in the dotted line photo, with about half a text row height's worth of overscan at the top and bottom. That way, it still fits on the screen, while the 200-line mode appears about the same as your first photo, with an even amount of overscan on all sides... as seen in this quick video:


https://www.youtube.com/watch?v=AWQQKi_1vCg

Shadow Lord
March 18th, 2016, 09:45 PM
Trixter,

I am not sure it matters. My reasoning: It is so simple to screw up the adjustment that there is no way IBM expected anyone to ever maintain some pre-conceived ratio. I've basically always set it so that the lettering looks good to my eye. If a screen mode after that looked bad I reached to the back made an adjustment. If this was some sort of critical ratio then the knobs to adjust things would be much harder to access IMHO. Of course as I said this is all conjecture on my part so if you do find a magical ratio do tell ;). Then of course we would have to go hunting for said magical ratio for the 5151, 5154, and 5157....

bhtooefr
March 19th, 2016, 01:30 PM
http://classiccomputers.info/down/IBM/IBM_PC_XT_5160/IBM_5160_Guide_to_Operations_6322511_APR84.pdf#pag e=181

As far as the height, the correct adjustment is to turn the computer off, turn brightness and contrast fully clockwise, turn vertical size fully counterclockwise, then turn the vertical size control clockwise until the black areas disappear.

That doesn't answer your question for width, though.

Trixter
March 19th, 2016, 02:04 PM
I think it might, actually, since the user width control is not user serviceable. Here's the relevant portion of the 5160 Guide to Operations: "Turn the Vertical Size control clockwise until the black areas just disappear. If one of the black areas disappears before the other, continue to turn the control until the second black area is gone." That supports my original calibration theory represented in the first picture I posted (and is, maybe not coincidentally, the exact method I used to adjust the monitor before I took the picture).

This is highly encouraging, but not definitive. I'm open to more opinions or specs.

On page 192 of that document, there are some "these are bad" sample images. Unfortunately, none of them are applicable to determining the proper ratio except that the "too narrow" is obviously not right and the "too shrunken" is obviously not right, so the correct result is somewhere in the middle.

I wish there was a way to track down the actual engineers. They are unknown, and most of the people (Don Estredge, William C. Lowe, etc.) who were close to the project are deceased.

vwestlife
March 19th, 2016, 03:32 PM
You'd think the proper width setting would be so that programs which shift the display to the left or right (such as MODE ,L or MODE ,R does) would not run off the side of the screen? I've come across some old games which automatically shift the display to the right, because old TVs with lots of overscan tend to cut off the left edge of the display at its normal centering.

Great Hierophant
March 20th, 2016, 05:32 AM
I am inclined to support the NTSC-aspect ratio with minimal vertical borders. My reasoning is that when IBM designed the CGA card, they did not design the 5153 at the same time. When CGA was released, the 5153 did not follow for two years and was very pricey when it was first released. Many graphics artists and programmers had to use the NTSC aspect ratio because all they had was a TV or a composite monitor to use. I'm not sure if there were many non-IBM CGA monitors prior to the 5153 or how obtainable they would have been.

Given the relatively small size of these screens, borders were seen as a necessary artifact but one that should be minimized. Most users probably would have preferred minimal borders to show the active display area as large as possible.

If you draw a circle in IBM PC BASIC, does it look perfectly circular with an NTSC aspect ratio?

bhtooefr
March 20th, 2016, 07:20 AM
CIRCLE is aiming at 4:3...

http://www.retroarchive.org/dos/docs/basic_ref_1.pdf#page=150


aspect affects the ratio of the x-radius to the y-radius. The default for aspect is 5/6 in medium resolution and 5/12 in high resolution. These values give a visual circle assuming the standard screen aspect ratio of 4/3.

And, that manual is dated May 1982, and has low-intensity yellow as brown, not dark yellow. Now to find a first-edition BASIC reference manual...

Trixter
March 20th, 2016, 08:42 AM
Given the relatively small size of these screens, borders were seen as a necessary artifact but one that should be minimized. Most users probably would have preferred minimal borders to show the active display area as large as possible.

This is not applicable to the Commodore 64, which has always had an equal-size border. Most purists consider it bad calibration to reduce or eliminate a C64 border.


If you draw a circle in IBM PC BASIC, does it look perfectly circular with an NTSC aspect ratio?

That is an excellent suggestion! I just did it, and the circles are only perfectly circular on my (calibrated, broadcast) NTSC monitor. However, that only proves bhtooefr's point, that they are only meant to be circular on a 4:3 monitor. It does not prove that the 5153 was meant to have a perfectly 4:3 viewable area.

Staying as objective as possible, I think it is a valid argument that the designers of the PC only ever meant for there to be one aspect ratio, and that they would want the RGB monitor to be a crisper version of an NTSC monitor or TV. Extending this argument further, if they DID mean for the aspect ratio of the 5153 to be changed, there probably would have been more noise about it in PC Tech magazine or programming journals or something.

Still looking for more proof or arguments... There is one strong argument on either side, so I'm afraid this is still up in the air...

retrogear
March 20th, 2016, 10:00 AM
I remember from when I did CRT repairs that we had to achieve convergence within a viewable area. The 5153 takes a 370mm diagonal tube which is 14.5". Part of it is hidden behind the bezel so considered a 14" display. Viewable area
is considered to be 13.2" (???). A 4:3 surface ratio correlates to a 3/4/5 right triangle. Doing the math places 11.2" width and 8.4" height of the 14" glass and 10.56" x 7.9" viewable area so roughly 1/4" recessed from all sides of bezel. 640x480 resolution
is a true 4:3 but cga is 640x400 or 320x200 which is an 17% reduction in height which reduces height by roughly another 1/2" top and bottom. Of course, where the raster starts until the actual video signal starts/stops will fudge the borders too.

PS - I came out of the woodwork but my math is still rusty ... :)

Trixter
March 20th, 2016, 10:45 AM
Thanks for the info, although I'm afraid you might be thinking of a different monitor, since the 5153 is considered a 12" diagonal monitor, not 14"...

This may help get the info we're looking for: What was the rule of thumb when adjusting the vertical and horizontal size pots? On CGA monitors, was the viewable area after adjustment meant to be a perfect 4:3?

retrogear
March 20th, 2016, 01:24 PM
I found an IBM announcement from 1983 that calls it a 13". I was going by the CRT# in the Sam's which is 370RKB2-TC14. Typically, 370 was the diagonal measurement in millimeters. I remember there was controversy how the measurement was advertised
whether it be viewing area, phosphor area or size of the glass. I do also remember some tube part#'s didn't follow the measurement either. As far as horizontal size, the control is a coil denoted as factory adjustment, so never touched by a service tech. Looking at schematic, that would also detune the flyback so would affect high voltage. I bet it was epoxied also. Since the width is fixed, that only leaves putting up a test pattern to get a centered, round circle at the optimal resolution. Whether it ends up 4:3 only a tape measure could tell.

PS - to make the measurement would require measuring the lit part of the raster, not the video

Trixter
March 20th, 2016, 02:25 PM
Since the width is fixed, that only leaves putting up a test pattern to get a centered, round circle at the optimal resolution.

I think this will give us the answer we're looking for: What was the generator of the test pattern? What did you hook up to a 5153 to generate the circle?

Great Hierophant
March 20th, 2016, 03:25 PM
This is not applicable to the Commodore 64, which has always had an equal-size border. Most purists consider it bad calibration to reduce or eliminate a C64 border.


A correction about the C64 my friend. The borders on the sides of an NTSC C64 are thicker than on the top and bottom. I have an NTSC C64 and I took a screenshot showing its borders :

30218

I believe perhaps the PAL C64 may have equal size borders due to the extra blanking induced by the PAL 625-line standard.

Now, compared to the display of the IBM CGA on the same TV, :

30219

you can see that even though 200 lines are still used, the horizontal size of the display is longer. Both systems use a multiple of 160 pixels, 320 in the first picture and 640 in the second. Commodore's video is in no danger of being covered by the monitor's bezel, but IBM's is just a hair too much to the left.

I would note that many composite monitors did not have vertical and horizontal size adjustments, at least not intended to be accessible by an ordinary user. The Commodore 1701 and 1702 Monitors have a vertical hold and a horizontal position control. The Apple II AppleColor Composite Monitor does have a vertical hold and a vertical size control but a horizontal hold control.

I must also note that my TV is a modern one and shifts the picture upward from dead center in order to leave more room for closed captions generated on the bottom of the screen.

Also, even though my CGA monitor is only a Tandy CM-5, it is advertised as a 14" tube/13" viewable display, just as the 5153. I measured at least a fair 12.5" with a measuring tape.

Trixter
March 20th, 2016, 03:33 PM
I believe perhaps the PAL C64 may have equal size borders due to the extra blanking induced by the PAL 625-line standard.

They do, which is what I was referring to; apologies. I wasn't aware that NTSC C64s had less borders due to the timing.

(BTW, your images are reversed.)

bhtooefr
March 20th, 2016, 03:41 PM
Also, I wouldn't be surprised if the BASIC example is Microsoft expecting 4:3... except when properly adjusted, it wouldn't be 4:3 due to viewable area on NTSC anyway.

The only composite monitor that I'm aware of IBM selling - and, after the 5153's launch, so it counts here - is the 5155's internal display, which has no adjustability of height or width. Ignoring borders, it's 6 25/32" wide, 4 31/32" tall - slightly off of 4:3.

Active area with EDIT.COM set to display white on white is 6 3/8" wide, 4 3/8" tall - that's about a 16:11 aspect ratio, not 4:3 (16:12) or 16:10.

In SCREEN 2, a CIRCLE (319,99),238 (the largest drawable CIRCLE in SCREEN 2) is 4 7/16" x 4 5/16", indicating an error of about 2.9% in the circle. That's nowhere near the error that (16/11)/(16/12) presents, though - brightness could be affecting width.

To be honest, I'm wondering if IBM simply didn't care about getting the aspect ratio that precise.

Great Hierophant
March 20th, 2016, 03:54 PM
To be honest, I'm wondering if IBM simply didn't care about getting the aspect ratio that precise.

I think we have probably put more thought into the discussion than IBM did back then.

Trixter
March 20th, 2016, 04:07 PM
I think we have probably put more thought into the discussion than IBM did back then.

That's the conclusion reenigne and I came to about a week ago when we had this discussion over email. I wasn't fully satisfied with the outcome of that conversation, which is why I started this topic. (Which would be finished by now, were it not for the 5160 service manual instructions.)

I'm very interested to learn what retrogear used as a "circle generator" when servicing 5153 monitors.

bhtooefr
March 20th, 2016, 04:48 PM
The other question is what the correct adjustment for the 4863 is, and that would tell the rest of the story, I think.

5153 seems to be 16:10, 5155 seems to be 16:11, 4863 might be 16:12?

retrogear
March 20th, 2016, 04:55 PM
Just to clarify, our shop was not an authorized IBM service center so we didn't have any "official" generator. Just used software to produce the necessary patterns.
I don't even recall what it was. We had a generator for NTSC composite patterns, though.

Scali
March 21st, 2016, 12:27 AM
I believe perhaps the PAL C64 may have equal size borders due to the extra blanking induced by the PAL 625-line standard.

The PAL C64 borders are much larger than what I see here at least.
In your case, even part of the main screen is lost in the top border (could be a poorly adjusted TV as well).
On PAL you had very large borders, and many demos abused these for border-sprite tricks. Most of those wouldn't even be visible on a configuration such as yours.

https://youtu.be/3WYVSbpY91Y

Scali
March 21st, 2016, 12:50 AM
That is an excellent suggestion! I just did it, and the circles are only perfectly circular on my (calibrated, broadcast) NTSC monitor. However, that only proves bhtooefr's point, that they are only meant to be circular on a 4:3 monitor. It does not prove that the 5153 was meant to have a perfectly 4:3 viewable area.

Yes, I think I feel the same about this.
Instead of 'meant to have', what if we say 'supposed to have'?
Namely, CGA was released in 1981. The 5153 was released in 1983.
Until the 5153 was available, the de-facto display for CGA was NTSC. And if that results in 4:3, you would say that is de-facto the correct aspect ratio for CGA.
Which makes me think that the 5153 is supposed to have a 4:3 aspect ratio as well, in order to remain compatible with existing CGA content.

Perhaps there is a reason why IBM chose a CRT and a bezel that does not naturally lead to a 4:3 ratio. In which case I would think 'cost' is that reason.
As others have said, we shouldn't over-think this issue. Back then, a lot of things were done to keep costs down and to keep designs simple. That was more important than having perfectly calibrated systems (another argument can be the change in CGA colours from old to new, which was not officially documented afaik, so was probably not considered an important change. They probably considered both sets of colours to be 'in the ballpark', and left it at that).

bhtooefr
March 21st, 2016, 03:57 AM
Well, except it wouldn't be 4:3 - it'd be slightly off due to not using all of the available lines.

geoffm3
March 21st, 2016, 06:56 AM
The PAL C64 borders are much larger than what I see here at least.
In your case, even part of the main screen is lost in the top border (could be a poorly adjusted TV as well).
On PAL you had very large borders, and many demos abused these for border-sprite tricks. Most of those wouldn't even be visible on a configuration such as yours.

https://youtu.be/3WYVSbpY91Y

I wonder if that's because the VIC-II still supports the same resolutions regardless of PAL or NTSC. Therefore they'd have to add additional overscan lines to accommodate for the extra scanlines top and bottom.

Trixter
March 21st, 2016, 08:24 AM
If the 5160 operations manual didn't have the "equal overscan" bit in the documentation, I would have said 4:3, same as NTSC, done. Most logical analysis points to that. But those calibration/setup instructions are actual IBM documentation, so I owe it to the community to try to track down someone involved with the design of the monitor to be 100% sure. I've done some research this past week and come up empty, so I reached out to someone who might have connections as a hail mary pass. I'll wait a week before declaring that a dead end.

vwestlife
March 21st, 2016, 11:43 AM
Do we really need to overthink this?

1. Adjust the vertical height as per IBM documentation.
2. Use BASICA to draw a circle in 320x200 graphics mode (SCREEN 1).
3. Adjust the horizontal width until the circle is perfectly circular.
Done!

offensive_Jerk
March 21st, 2016, 12:33 PM
Funny, I was just reading the manual to Ashton-Tate's SignMaster program.

To correct the aspect ratio the program wants to know dimensions of the monitor.

YOU THEN WILL BE ASKED THE FOLLOWING QUESTIONS CONCERNING YOUR MONITOR:

MEASURE THE RECTANGLE ON THE SCREEN (NOT THE SCREEN ITSELF) FROM LEFT TO RIGHT AND ENTER IT IN DECIMAL UNITS (E.G. "9"). DO THE SAME FOR THE HEIGHT OF THE RECTANGLE AND ENTER IT IN THE SAME MANNER (E.G. "7.5").

The picture shows a rectangle border around the screen. It must be leaving out the overscan area.

Then in the picture it says:

ASPECT RATIO: THE PROGRAM MUST KNOW THE DIMENSIONS OF YOUR DISPLAY IN ORDER TO PRODUCE CIRCLES THAT ARE ROUND. HOW WIDE IS THE FRAME ON THIS SCREEN, IN INCHES (DECIMAL UNITS)?

Trixter
March 21st, 2016, 01:12 PM
Do we really need to overthink this?

I know it seems pedantic, but it would be nice to know what was intended. Wouldn't it be nice to know, exactly, what IBM CGA graphics are supposed to look like on the definitive CGA monitor?


3. Adjust the horizontal width until the circle is perfectly circular.

I think you're kidding, but in case you're not: The 5153 has no user-serviceable horizontal width control. Come to think of it, I don't know how many of my other CGA monitors do either. And as was pointed out earlier, the 5155 monitor has no horizontal adjustment on the circuit board so it's not possible even if you try to service it.

Also, your answer suggests you are deeming 4:3/NTSC the correct ratio based on the BASICA CIRCLE. All that proves is that NTSC monitors were meant to be 4:3, not the 5153 and other CGA monitors.


To correct the aspect ratio the program wants to know dimensions of the monitor.

That's an interesting data point -- It indirectly suggests that nobody followed a standard and that it was up to each program to query the user.

bhtooefr
March 21st, 2016, 01:37 PM
Another interesting data point is in the HMR: https://www.lo-tech.co.uk/downloads/manuals/ibm-pc/IBM_Hardware_Maintenance_Reference/PC_and_XT/HMR-6280088-Repair_Information-PC_and_XT.pdf#page=56

Corroboration of the adjustment procedure in the GTO manual, as well as the same procedure for "Mode 1" (read: CGA modes) on the EGA monitor, yet a different procedure (of equalizing the borders) for Mode 2 (read: EGA modes).

And, I didn't realize that there weren't /any/ adjustments on the 5155, but the HMR doesn't refer to any - the only adjustments that the 5155 HMR lists are to drive speed, the motherboard color trimmer, and the 5153 adjustments.

VileR
March 21st, 2016, 01:38 PM
I wish there was a way to track down the actual engineers. They are unknown, and most of the people (Don Estredge, William C. Lowe, etc.) who were close to the project are deceased.
What about Andy Saenz? Dr. David Bradley mentions him as "responsible for the video card": http://www.vintagecomputing.com/index.php/archives/790/the-ibm-smiley-character-turns-30
There are all those other little CGA quirks that'd be nice to have explained straight from the horse's mouth... I thought it'd make for some nice blog fodder, but never got very far in tracking the guy down. :)

Eudimorphodon
March 21st, 2016, 01:53 PM
I know it seems pedantic, but it would be nice to know what was intended. Wouldn't it be nice to know, exactly, what IBM CGA graphics are supposed to look like on the definitive CGA monitor?

As noted in the web page in the OP the difference between 1.37 and 1.33 is 3%; on a 11" diagonal (viewable) display the difference if the monitor were *perfectly* adjusted to the two ratios would be an 8.8"x6.6" rectangle for 1.33 vs. an 8.9"x6.5" rectangle for 1.37. I'm not entirely sure that little difference would be readily visible to the human eye.

(Since the width isn't adjustable on the 5153 if you take the 8.8" number as a fixed width the height of the display will be 6.42" for 1.37 instead of 6.62". On one hand that's almost a quarter of an inch, but that also means that if both displays were perfectly centered the 1.33 will be just one tenth of an inch closer to the top and the bottom of the frame. Not sure I'd notice that unless I had the two right next to each other *and* had a color border.)

offensive_Jerk
March 21st, 2016, 01:54 PM
What about Andy Saenz? Dr. David Bradley mentions him as "responsible for the video card": http://www.vintagecomputing.com/index.php/archives/790/the-ibm-smiley-character-turns-30
There are all those other little CGA quirks that'd be nice to have explained straight from the horse's mouth... I thought it'd make for some nice blog fodder, but never got very far in tracking the guy down. :)

I remember reading an interview with the guy who came up with the smiley face in code page 37. Not sure if it's the same guy or not, but he was open to talk about points that other people wouldn't give a second thought :rolleyes:.

vwestlife
March 21st, 2016, 02:16 PM
I know it seems pedantic, but it would be nice to know what was intended. Wouldn't it be nice to know, exactly, what IBM CGA graphics are supposed to look like on the definitive CGA monitor?

If you own a 5153 in good condition and a ruler, why not measure the aspect ratio for yourself? The calibration may have drifted somewhat over the years, but probably not any more than the original tolerance of the factory adjustments.

Or get one of these:

http://img.directindustry.com/images_di/photo-g/18583-7393193.jpg

VileR
March 21st, 2016, 02:43 PM
I remember reading an interview with the guy who came up with the smiley face in code page 37. Not sure if it's the same guy or not, but he was open to talk about points that other people wouldn't give a second thought :rolleyes:.
erm, that would probably be the interview I linked to. That's Bradley, but he does namedrop Andy Saenz as being the video card guy (CGA? MDA? both?), so maybe he (or Benj Edwards) would have a lead on Saenz?

Eudimorphodon
March 21st, 2016, 02:56 PM
Corroboration of the adjustment procedure in the GTO manual, as well as the same procedure for "Mode 1" (read: CGA modes) on the EGA monitor, yet a different procedure (of equalizing the borders) for Mode 2 (read: EGA modes).

My memory is getting foggy on this point, but wouldn't the need for the separate procedure for "Mode 2" be dictated by the EGA 350 line modes not having a border? (IE, EGA basically has almost no overscan so the "crank up the brightness and center the whole scan block" is the correct tactic.)

In any case, if the official adjustment procedure for CGA is "turn the knob until both the black bars go away" it sort of implies they weren't really aiming for mathematical precision. ;)

Trixter
March 21st, 2016, 04:23 PM
Corroboration of the adjustment procedure in the GTO manual, as well as the same procedure for "Mode 1" (read: CGA modes) on the EGA monitor, yet a different procedure (of equalizing the borders) for Mode 2 (read: EGA modes).

Wow, the instructions are different for all three scenarios:
CGA monitor: "Turn the Vertical Size control clockwise until the black areas just disappear." (emphasis mine)
EGA monitor, mode 1: "Turn the Vertical Size control clockwise until both black areasdisappear."
EGA monitor, mode 2: "Adjust the Mode 2 Vertical Size control until the black areasat the top and bottom are approximately the same size as theblack areas at the sides."

Frustrating! #1 and #3 support "even border from all sides of the bezel", and #2 is just indeterminate.


If you own a 5153 in good condition and a ruler, why not measure the aspect ratio for yourself? The calibration may have drifted somewhat over the years, but probably not any more than the original tolerance of the factory adjustments.

I did, reflected in the first image in the first post of this thread. It is visibly not identical to 4:3 NTSC if calibrated according to IBM documentation. The question this thread is trying to answer is: Is that normal for the 5153, or was the 5153 always supposed to be 4:3 just like a composite monitor, and the IBM docs are wrong?


Or get one of these:

...which is clearly mis-marked when it comes to CGA! CGA wasn't a 67Hz, 35KHz horizontal monitor... This must be a device for calibrating arcade monitors.


In any case, if the official adjustment procedure for CGA is "turn the knob until both the black bars go away" it sort of implies they weren't really aiming for mathematical precision. ;)

I don't care about precision, I care about which setting is correct. It's possible to adjust a 5153 to match an NTSC monitor by nearly eliminating the top and bottom overscan areas, but is that correct? Especially when there are many photos in old magazines, including an IBM press release photo I dug up, that show a different calibration?

Research continues...

Trixter
March 21st, 2016, 04:29 PM
he does namedrop Andy Saenz as being the video card guy

I've reached out to Saenz; hopefully he will respond.

retrogear
March 21st, 2016, 05:08 PM
As far as the BK Precision generator, when switch is set to PC and CGA you get 15Khz CGA. The frequency spec column is for switch set to MAC.
Dang, Mouser has it for $320. Must be a market for it yet?

bhtooefr
March 21st, 2016, 05:17 PM
As an aside, I decided to take a different approach, and look through old PC Magazines to get an idea of what monitors IBM recommended before the 5153.

https://books.google.com/books?id=w_OhaFDePS4C&lpg=PP1&pg=RA1-PA68#v=onepage&q&f=false

It's looking like there were two color monitors that IBM and ComputerLand stores sold: the Amdek Color I (a composite monitor), and the Amdek Color II (a RGB monitor that later got an upgrade to RGBI specifically for the IBM PC). So, even if those monitors weren't the design target for CGA, they were the first semi-official CGA monitors, so to say.

Shadow Lord
March 21st, 2016, 09:41 PM
...which is clearly mis-marked when it comes to CGA! CGA wasn't a 67Hz, 35KHz horizontal monitor... This must be a device for calibrating arcade monitors.

Well according to the specs:


Specifications
Resolutions:
Type Resolution Horizontal Polarity Vertical Polarity
(kHz) (Hz)
MDA 720 x 350 18.4 + 50 -
CGA 640 x 200 15.8 + 60 +
EGA 640 x 350 21.8 + 60 -
VGA 640 x 480 31.5 - 60 -
SVGA 800 x 600 35.2 + 56 +
8514A 1024 x 384 35.5 + 87 +


Crappy table work is all me ;)

HoJoPo
March 21st, 2016, 10:03 PM
...which is clearly mis-marked when it comes to CGA! CGA wasn't a 67Hz, 35KHz horizontal monitor... This must be a device for calibrating arcade monitors.


If you notice the "MAC/PC" switch, the numbers on the right are for Mac monitors, the information on the left for PC monitors.

Eudimorphodon
March 21st, 2016, 10:40 PM
Thinking about the OP again, and after doing some ignorant Google-ing, I'm sort of wondering about the entire reasoning behind trying to say CGA's effective display aspect ratio "should" be any particular number based on SMPTE 170M. Strictly speaking does the NTSC standard say anything about pixel aspect ratio? I would genuinely like to see more of the math/assumptions that went into that post, is it relying on a constant standard for raster spacing vs. width or something? Analog NTSC strictly speaking doesn't have "pixels", it describes a raster, and technically the same scanning can be used for non-4x3 displays. (Perhaps notably the Rec. 601 standard for digitally encoding " 480i" content on devices uses a non-square pixels by design, with said pixels working out roughly to 10:11 in 4x3.)

Seriously, I'd like to genuinely know what the reasoning is, there totally could be something I'm missing.

Scali
March 22nd, 2016, 12:40 AM
Strictly speaking does the NTSC standard say anything about pixel aspect ratio?

I'm pretty sure it does, since test cards generally include circle shapes and such, so that people can adjust their set to them: https://en.wikipedia.org/wiki/Test_card

VileR
March 22nd, 2016, 05:19 AM
I've reached out to Saenz; hopefully he will respond.
I hope so. We'll have to prepare all our unanswered CGA questions if he does!

Trixter
March 22nd, 2016, 11:35 AM
I'm pretty sure it does, since test cards generally include circle shapes and such, so that people can adjust their set to them: https://en.wikipedia.org/wiki/Test_card

NTSC (the analog broadcast standard) as ratified in 1940 does specifically define a 4:3 display aspect ratio. There is no mention of pixels or individual display elements, of course, because those do not exist in the analog realm. There are 525 total lines per frame, of which 483 lines are visible. Later digital specifications shortened this to 480 lines, leading to the "720x480" standard with a defined PAR of 0.9.

So, you are right that NTSC test cards have circles in them for aspect ratio purposes; on a 4:3 display, they should appear circular. (All we need now is undeniable verification that the viewable area of the 5153 was intended to be a perfect 4:3.)


I hope so. We'll have to prepare all our unanswered CGA questions if he does!

You may want to start preparing them now, privately, just in case.

Scali
March 22nd, 2016, 12:04 PM
I hope so. We'll have to prepare all our unanswered CGA questions if he does!

Yea, once all our CGA questions are answered, we can finally move on to EGA...? :)

Trixter
March 22nd, 2016, 01:13 PM
Actually, EGA (in terms of monitors, anyway) is fairly well-defined. Very clear adjustment instructions (http://www.vcfed.org/forum/showthread.php?51820-What-is-the-proper-aspect-ratio-adjustment-of-the-IBM-5153-CGA-monitor&p=407095#post407095) are in IBM documentation.

Scali
March 22nd, 2016, 01:39 PM
I was thinking in terms of making an EGA demo after all CGA secrets have been unlocked ;)

vwestlife
March 22nd, 2016, 03:01 PM
If you adjust 200-line CGA to a 4:3 aspect ratio on a 5153, then Tandy's 225-line text mode would go off the screen, no? That's why Tandy gives you MODE 200 or MODE TV, because on a composite monitor or TV, 225-line mode does go off the top and bottom edges of the screen. That's why I'd rather adjust a 5153 monitor to the slightly shorter (more widescreen) aspect ratio -- not only to make the borders evenly sized on all sides, but also so just in case I connect it to a Tandy 1000, I wouldn't have to tweak the vertical size to keep the whole image visible.

KC9UDX
March 22nd, 2016, 04:18 PM
As an aside, I decided to take a different approach, and look through old PC Magazines to get an idea of what monitors IBM recommended before the 5153.

https://books.google.com/books?id=w_OhaFDePS4C&lpg=PP1&pg=RA1-PA68#v=onepage&q&f=false

It's looking like there were two color monitors that IBM and ComputerLand stores sold: the Amdek Color I (a composite monitor), and the Amdek Color II (a RGB monitor that later got an upgrade to RGBI specifically for the IBM PC). So, even if those monitors weren't the design target for CGA, they were the first semi-official CGA monitors, so to say.

For whatever it's worth, Amdek Color monitors do not have user-adjustable horizontal size, but vertical size is a screwdriver adjustment, for the user.

I'd be willing to bet if someone asked a CGA monitor engineer any of these questions, he'd get angry.

Eudimorphodon
March 22nd, 2016, 05:41 PM
I'm pretty sure it does, since test cards generally include circle shapes and such, so that people can adjust their set to them: https://en.wikipedia.org/wiki/Test_card

Well, yeah, the *display* aspect ratio is intended to be 4:3, but that doesn't necessarily say anything about *pixel* aspect ratio, so my question was more along the line of wondering how you'd get a (such a precise) pixel aspect ratio from the SMPTE standards doc which appears to be mostly about how the color space is represented. I mean, for instance, I gather the Rec 601 standard settled on its 10:11 pixel ratio (and accompanying 720 pixel ideal line length) to be close to a multiple of the NTSC color carrier phases (or something, my eyes start to glaze over), I guess I'm just a little lost on the assumptions used to come up with such an exact number for the *display* aspect ratio by working backwards from an ideal NTSC color space equation; my impression was that the pixels should be shaped to fit the display, not vice versa, so working it backwards like that is just going to be amplifying the error bars.

Also not totally relevant, or maybe it is, is the fact that SMPTE-170M should: A: apply equally stringently to or more so to Composite monitors over the 5153, since they're closer to being "TVs"... and in fact the whole discussion about color space doesn't apply to the 5153 at all because it doesn't use NTSC color space, and B: Neither the composite or RGB outputs on the CGA card are in close compliance with broadcast NTSC standards anyway. Both of them use the "progressive NTSC" shortcut that was so common in the home computer/video game space in the 1970's through early 90's. They don't generate the half-frame that regular NTSC uses to shift the rows between the even and odd frames, they just spit out 262 (usually, although the exact number varies) progressive lines and give the monitor an oddball Vsync pulse that makes it just scan the same lines over again. (Giving CGA that "radiator grill" look we all love so much.) Granted that shouldn't change *too* many assumptions about other aspects of the video signal like the NTSC color space, but it's another point to argue that a number derived from working backwards through the broadcast standard equations was probably not a design goal of the system.

Trixter
March 22nd, 2016, 07:45 PM
I'd be willing to bet if someone asked a CGA monitor engineer any of these questions, he'd get angry.

My secret fear is that exact scenario. It is possible the person who designed the 5153 housing made a mistake, or simply didn't care, or didn't know any better, etc. I don't care what the real answer is, and I won't judge anyone for any mistakes made in what was a very rushed design scenario -- I just want to know the truth.

NTSC is known, so I'm going to rewrite The CGA Compatibility Tester's aspect ratio portion for composite monitors, and wait until the results of this investigation to handle the RGB monitor portion.

Eudimorphodon
March 22nd, 2016, 07:49 PM
A long, rambling, but somewhat amusing rant about how trying to make sense of aspect ratios from the published specs is hopeless. (http://www.lurkertech.com/lg/video-systems/)

Trixter
March 22nd, 2016, 08:12 PM
Fascinating read! Funny and depressing at the same time. I always wondered why VCD anamorphic 352x240 MPEG-1 files never displayed properly all the time...

reenigne
March 23rd, 2016, 12:25 AM
Well, yeah, the *display* aspect ratio is intended to be 4:3, but that doesn't necessarily say anything about *pixel* aspect ratio, so my question was more along the line of wondering how you'd get a (such a precise) pixel aspect ratio from the SMPTE standards doc

The pixel aspect ratio is fixed by the scanline distance and the pixel width, and the latter is fixed by the fact that there are 4 CGA (high-res) pixels per color carrier cycle. I did the calculations (http://www.reenigne.org/blog/what-is-the-cga-aspect-ratio-exactly/) last time this came up, and the standard gives a display aspect ratio (for the CGA's 640x200 active area) of about 1.37:1 (which is what the dotted rectangle in the out-of-focus image at the start of this thread is) with a tolerance of something like 0.005:1. Of course, the questionable assumption here is that an image on the 5153 was supposed to have the same aspect ratio as the same image on an NTSC-standard composite monitor.

On Trixter's monitor, doing the recommended adjustment (making the black borders outside the overscan area just touch the bezel) gives a slightly wider DAR than 1.37:1, suggesting that these two different ways of fixing the aspect ratio aren't quite compatible. It's possible that they are compatible for some adjustment of the image width. The image width is not adjustable without taking the case off, but could still have drifted over the years on any given real 5153 (even if it was originally adjusted that accurately to begin with, which is debatable). Perhaps not coincidentally, 1.37:1 is very close to the bezel aspect ratio of the 5153 (measuring at the widest and tallest points).

From the point of view of people creating CGA images (including aspect ratio calibration images, as in cgacal which sparked off this discussion), I don't think there is any reason to create separate images for 5153 and composite. Even if there is a nominal difference it's pretty small, so we can continue to do what we have always done and assume a DAR of 4:3 (PAR of 5:6) without it looking "wrong".

reenigne
March 23rd, 2016, 04:43 AM
A long, rambling, but somewhat amusing rant about how trying to make sense of aspect ratios from the published specs is hopeless. (http://www.lurkertech.com/lg/video-systems/)

That is indeed a fascinating read - thanks! If we take the 12+3/11 MHz sample rate for square pixels as gospel, the CGA has a pixel aspect ratio of 6:7 instead of 5:6 and the display aspect ratio becomes 48:35 (rather less catchy than 4:3) which is ~1.372:1 (well within the tolerances of SMPTE 170M as we would expect). A 6:7 PAR is less convenient than 5:6 for drawing purposes, though, since the smallest square you can draw would then be an odd number of pixels horizontally, and odd numbers are less convenient than even ones (especially horizontally, where it's often more convenient to draw 2, 4, 8 or 16 pixels at once). Also it would be slightly off on VGA CRT (which really does have a 5:6 pixel aspect ratio when emulating CGA).

Scali
March 23rd, 2016, 04:50 AM
Well, yeah, the *display* aspect ratio is intended to be 4:3, but that doesn't necessarily say anything about *pixel* aspect ratio

Well, from one follows the other, does it not?
The NTSC aspect ratio is specified, so connecting a CGA card to an NTSC-compatible display gives you an exact aspect ratio.
Since CGA has a fixed resolution of 640x200 (or 320x200 in graphics mode), there is only one possible mapping of the CGA pixels to the NTSC display.
Ergo, there is a fixed display aspect ratio and a fixed pixel aspect ratio for CGA-on-NTSC.

vwestlife
March 23rd, 2016, 05:44 AM
A long, rambling, but somewhat amusing rant about how trying to make sense of aspect ratios from the published specs is hopeless. (http://www.lurkertech.com/lg/video-systems/)

That's almost as bad as the CD Red Book failing to adopt a uniform loudness standard, leading to the loudness war (https://en.wikipedia.org/wiki/Loudness_war).

Trixter
March 23rd, 2016, 07:42 AM
That is indeed a fascinating read - thanks! If we take the 12+3/11 MHz sample rate for square pixels as gospel, the CGA has a pixel aspect ratio of 6:7 instead of 5:6 and the display aspect ratio becomes 48:35 (rather less catchy than 4:3) which is ~1.372:1 (well within the tolerances of SMPTE 170M as we would expect). A 6:7 PAR is less convenient than 5:6 for drawing purposes, though, since the smallest square you can draw would then be an odd number of pixels horizontally, and odd numbers are less convenient than even ones (especially horizontally, where it's often more convenient to draw 2, 4, 8 or 16 pixels at once). Also it would be slightly off on VGA CRT (which really does have a 5:6 pixel aspect ratio when emulating CGA).

For now, I'm happy with this statement. I'm going to replace what is in cga_comp with a 4:3 aspect ratio pattern that looks perfect on NTSC monitors, and will include a summary of this conversation in the cga_comp documentation for those who ask why there is no corresponding pattern for RGB monitors.

I can personally tell the difference between 1:33:1 and 1.37:1; it is quite visible to me. This may because 1. I have been producing digital video for over 20 years, and 2. I am extraordinarily picky and OCD about these things. Unless I hear back from the IBM people, I'm going to abstain from making any judgments or statements about RGB/5153 aspect ratios.

If it turns out 1.372:1 is the correct ratio, then I hope I someday hear the explanation behind the 5153 bezel and the IBM documentation. Both may have been faulty assumptions or outright mistakes on the part of IBM.

Eudimorphodon
March 23rd, 2016, 09:25 AM
That is indeed a fascinating read - thanks!

:)

For me one of the most relevant aspect of the whole article to this discussion is this, which it repeats in somewhat different ways several times:

"The deep, dark secret of old-school video equipment is that it actually doesn't care exactly where the edges of the picture are located."

I think any discussion about the exact shape of the 5133's bezel, etc, needs to acknowledge that at the consumer level this was all *totally* an inexact science, and even at the professional level what they were mostly concerned with is having a standard for where the "center" of the viewable area is and setting a minimum guideline that essentially says "put stuff inside this rectangle and a reasonably tuned TV should have it visible on the screen, outside of that you're on your own.". When you look at old TVs you'll find they're all over the map in terms of bezel shape, so it's not as if the industry as a whole cared that much about making the bezel match a 4:3 viewable area.


Also it would be slightly off on VGA CRT (which really does have a 5:6 pixel aspect ratio when emulating CGA).

This is some ad-hoc reasoning in play here, but... going back to the OP, with the initial picture of the active area scaled to have roughly equal black borders all the way around: that's how I remember CGA monitors usually being adjusted, and if I was the one adjusting them that's what I'd set it to because I find that more aesthetically pleasing than the taller supposedly-more-accurate-for-4:3-taking-the-border/overscan *NTSC" "active area"-into-account shape. (I think human beings sort of naturally like nested rectangles; support for this could be derived from IBM's documentation for adjusting the EGA monitor which specifically says the correct adjustment is even size black areas all around; remember, it's different because on EGA there's no distinction between "active with pixels area" and "overscan/border active area" because EGA effectively has no overscan/border.) It's this scaling that makes 5:6 about the right PAR, correct? So.. I suppose it's at least remotely possible that IBM arrived on the optimal shape for the bezel as one that:

#1: Created an even-size border all around and

#2: Happened to make the circles drawn by the BASIC command (which used an easy mathematical shortcut for scaling) square.

The question really is did someone think super-hard about this, or is it all just post-hoc reasoning? The reason I think it *could* be the latter is, well, composite monitors usually have VSIZE controls too. I'd think that anyone who was anal about having their circles perfectly circular would mess with that to adjust the monitor to the CIRCLE command rather than vice-versa. (I know I did that myself even on my Tandy Color Computer, messing with the TV to try to make the circles round.)

I suppose another way to try to dig for evidence one way or another would be to try to find as much software as you can (spreadsheet graphing programs, programming languages, paint programs, whatever) that tried to draw perfect circles and compare the ratios the programmers used to see if there's a consistent trend distinguishing programs written before the 5153 came out and those after. I'd be really surprised if people really were resorting to more complex calculations to fix something a tiny tweak of a vsize dial can, but.. ?

KC9UDX
March 23rd, 2016, 10:37 AM
Unless you actually measure, it's hard to tell if a circle is ellipsoid until it gets pretty far out. The exception to that is if there's bad linearity.

cthulhu
March 23rd, 2016, 11:28 PM
That's almost as bad as the CD Red Book failing to adopt a uniform loudness standard, leading to the loudness war (https://en.wikipedia.org/wiki/Loudness_war).

A uniform loudness standard has no place in the Compact Disc specification. Even if it had included one it would've just been ignored anyway. Also, the loudness war predates the CD by decades.

Scali
March 24th, 2016, 12:40 AM
#1: Created an even-size border all around and

#2: Happened to make the circles drawn by the BASIC command (which used an easy mathematical shortcut for scaling) square.

From what I understood, the 5153 does only one of these at a time, not both.


I'd think that anyone who was anal about having their circles perfectly circular would mess with that to adjust the monitor to the CIRCLE command rather than vice-versa. (I know I did that myself even on my Tandy Color Computer, messing with the TV to try to make the circles round.)

The interesting question for me is: what did the person who wrote the CIRCLE command in BASIC use as a reference?
It clearly implements correction for a non-trivial pixel-aspect ratio (not just 1:1 or 2:1).
So what is the exact pixel-aspect ratio that the author used as a basis, and where did he get that number from?
Since BASIC was developed by MS/IBM during design of the 5150 PC, and BASIC was installed in the ROM of every PC, I think there would have been close contact between the software and hardware developers.

So in my opinion, BASIC is probably 'the absolute truth', regardless of how many programs do it differently.
Mind you, BASIC was written a few years before the 5153 was on the market.

reenigne
March 24th, 2016, 01:02 AM
From what I understood, the 5153 does only one of these at a time, not both.



The interesting question for me is: what did the person who wrote the CIRCLE command in BASIC use as a reference?
It clearly implements correction for a non-trivial pixel-aspect ratio (not just 1:1 or 2:1).
So what is the exact pixel-aspect ratio that the author used as a basis, and where did he get that number from?
Since BASIC was developed by MS/IBM during design of the 5150 PC, and BASIC was installed in the ROM of every PC, I think there would have been close contact between the software and hardware developers.

So in my opinion, BASIC is probably 'the absolute truth', regardless of how many programs do it differently.
Mind you, BASIC was written a few years before the 5153 was on the market.

Looks like spot on 5:6 to me: http://www.reenigne.org/misc/basic_circle_aspect.png

(The smaller circle, radius parameter 180, is 300 pixels tall and 360 wide when I measure it in Paint Shop Pro).

Combined with the VGA evidence, it's pretty compelling that IBM meant for the DAR of the CGA to be 4:3 at least for composite. If they meant 5153 to have a different DAR then I'm mystified as to why they would have done so.

Scali
March 24th, 2016, 01:57 AM
Looks like spot on 5:6 to me: http://www.reenigne.org/misc/basic_circle_aspect.png

Also looks like DOSBox gets it completely wrong.


If they meant 5153 to have a different DAR then I'm mystified as to why they would have done so.

Yes, but I guess there are too many open ends for that.
From what I understand, the width of the screen is set at the factory, whereas the user gets control over the height of the screen.
We can't rule out the possibility that the people responsible for adjusting the width at the factory used a wrong setting because of some kind of miscommunication.
This would preclude the user from dialing in the proper DAR by varying the height of the screen.

It could also be that the circuit design is somewhat flawed, and it cannot generate the required DAR within safe working parameters, so the factory decided to use safe settings rather than accurate settings.

reenigne
March 24th, 2016, 02:36 AM
Also looks like DOSBox gets it completely wrong.

Yes, DOSBox (at least in the default configuration, without any fancy scalers/shaders) uses a 2:1 (1:1 for low-res) PAR, so it's way off what it should be. That doesn't matter for figuring what PAR is programmed into the software, though.


It could also be that the circuit design is somewhat flawed, and it cannot generate the required DAR within safe working parameters, so the factory decided to use safe settings rather than accurate settings.

It can generate a DAR of 4:3 or 48:35 or even wider just by adjusting the image height...

Scali
March 24th, 2016, 03:22 AM
It can generate a DAR of 4:3 or 48:35 or even wider just by adjusting the image height...

Ah right, it would just lose nearly all of the vertical overscan, so it looks a bit disproportional to the horizontal overscan.

vwestlife
March 24th, 2016, 05:04 AM
A uniform loudness standard has no place in the Compact Disc specification.

Movie theaters have had an industry standard audio playback level for over 40 years. U.S. TV stations are required by law to limit the volume of commercials (the "CALM Act"). European FM radio stations have a mandatory audio loudness level (ITU-R BS.1770). iTunes and YouTube both use ReplayGain-based uniform loudness levels. Similar efforts could and should have been done with CDs.


Even if it had included one it would've just been ignored anyway.

There are always some rogues who smash their levels on purpose, but most recording studio engineers know what a VU meter is and how to keep an eye on it during recording. With analog recording you always needed to reserve some "headroom" due to the physical limitations of the media. The problem is that during the transition to digital, no one could agree on how much headroom to reserve, if any, because physical limitations were no longer a problem. A uniform standard headroom level for digital audio would've gone a long way to preventing the drastically different audio playback levels of early CD releases, which led to artists and consumers complaining that some CDs were too quiet and that future releases needed to be louder, exacerbating the loudness war.


Also, the loudness war predates the CD by decades.

But as I mentioned above, the transition to digital meant that loudness was no longer bound by physical limitations. A record will cause the needle to jump out of the groove if you cut it too hot, and a loud LP will have less playback time per side than one cut at a more conservative level. A loud, overmodulated CD will still play fine and for the same amount of time as a quiet CD; it'll just sound bad.

KC9UDX
March 24th, 2016, 05:29 AM
65535 isn't a physical limitation?

If loudness standards are good, everything should be mostly the same colour and most things should taste pretty much the same too.

The limited dynamic range of CDs stifles creativity enough as it is.

Scali
March 24th, 2016, 05:53 AM
The limited dynamic range of CDs stifles creativity enough as it is.

The dynamic range of 16-bit signed samples, as used on CDs, delivers far more dynamic range than any mainstream analog media that went before it (such as compact cassette, 8-track, vinyl, reel-to-reel etc).
So while it was a physical limitation, as stated, the amount of headroom to preserve was not commonly agreed on. There suddenly was so much more headroom than what went before it, and more importantly: there was basically no noise. Historically the point of loudness was to cram as much signal and quality into a very limited bandwidth, where signals that were too soft would disappear under the inherent noise of the media, and signals that were too loud, would clip/distort.
With CD, you suddenly had so much more room to play with, that it became quite arbitrary where you would draw the exact line. Aside from that, recording/mastering for digital media was still in its infancy, and it was not that easy to explore the limits of the media, so better safe than sorry.

I agree that 'loudness' should not be defined by the media (and standardization thereof) itself. No more than a specific level of 'brightness' or such should be defined for pictures/videos.
It is up to the artist to experiment with it, and find the levels that express his meaning best (not that this is actually what is happening in the loudness war, but that's another story).

cthulhu
March 24th, 2016, 06:41 AM
A uniform standard headroom level for digital audio would've gone a long way to preventing the drastically different audio playback levels of early CD releases, which led to artists and consumers complaining that some CDs were too quiet and that future releases needed to be louder, exacerbating the loudness war.

If anything has exacerbated the loudness war it has been the incremental improvements in technology which have allowed the perceived volume of audio to be steadily increased year over year whilst keeping the distortion due to this processing from reaching a point where it became unacceptable to the masses. The desire to make recordings as loud as possible goes right back to the beginning of recording itself and this desire combined with the tools to actually achieve it are what have really driven the loudness war. I think the Compact Disc and digital audio in general have been made the scapegoat for this escalation. They didn't start the loudness war so why should they be expected to end it? And as I mentioned before any loudness standard included in the CD specification would've just been ignored anyway, since as you point out the playability of them is unaffected by exceeding any imposed volume limit.

Trixter
March 24th, 2016, 01:29 PM
it's pretty compelling that IBM meant for the DAR of the CGA to be 4:3 at least for composite.

I agree (the BASICA evidence is the strongest argument), which is why I completely revamped the aspect ratio calibration screen in The CGA Compatibility Tester (http://www.oldskool.org/pc/cgacomp) last night, and clarified it was to be used for NTSC composite monitors only.

bhtooefr
March 24th, 2016, 04:40 PM
There's also the possibility that Microsoft took a na´ve approach to circle drawing - "hey, it's 320x200, we're rendering to a 4:3 display, that means we've got a PAR of 1:1.2" (or for 640x200, a PAR of 1:0.6) - and didn't consider the hardware, figuring that they'd #1 actually be rendering to a 4:3 display, and #2 the hardware would fill the display or would tell users to adjust V size appropriately.

I will say that this means that for BASIC code specifically, this means that correct adjustment is for 4:3. But, that doesn't mean that other code, even from the pre-5153 era, makes that particular assumption.

vwestlife
March 24th, 2016, 05:56 PM
as I mentioned before any loudness standard included in the CD specification would've just been ignored anyway, since as you point out the playability of them is unaffected by exceeding any imposed volume limit.

But if an official loudness/headroom rule was written into the Red Book, then a CD would need to meet that standard in order to get the official "Compact Disc Digital Audio" logo stamped on it, and CD pressing plants could reserve the right to reject master discs that were not Red Book complaint. (Even without an official loudness standard, allegedly the Red Hot Chili Peppers had to sign a waiver to get their "Californication" album made, because it was so loud and distorted that it didn't meet the CD pressing plant's quality standards.)

Red Book compliance did become an issue in the 2000s when rootkit-virus-laden "copy protected" CDs and dual-sided CD/DVDs and CD/SACDs were pushed on the market -- none of which were Red Book compliant. These discs did not work in some CD players, resulting in the common disclaimer that "if you don't see the Compact Disc logo on the CD, then it's not guaranteed to play".

Scali
March 24th, 2016, 11:58 PM
There's also the possibility that Microsoft took a na´ve approach to circle drawing - "hey, it's 320x200, we're rendering to a 4:3 display, that means we've got a PAR of 1:1.2" (or for 640x200, a PAR of 1:0.6) - and didn't consider the hardware, figuring that they'd #1 actually be rendering to a 4:3 display, and #2 the hardware would fill the display or would tell users to adjust V size appropriately.

Well, with NTSC you *are* rendering to a 4:3 display, and most TVs won't allow the user to adjust H or V size (there's no point really, since it's all standardized. Set it up correctly at the factory and you're done).
You should be careful about deriving the PAR from only the resolution though, since especially these older systems had overscan areas (borders) of arbitrary width and size.
The use of these borders was mainly to align a particular grid of pixels to the colour clock (eg for CGA you have a resolution of 640 pixels horizontally, where every 4 adjacent pixels are used to generate a colour pattern).
So there's no reason why the horizontal border size should have any relation to the vertical border size, which means that the aspect ratio of your active display area (that which the resolution of 640x200 is based on) may not necessarily be the same as the total aspect ratio including horizontal and vertical overscan.

cthulhu
March 25th, 2016, 12:40 AM
But if an official loudness/headroom rule was written into the Red Book, then a CD would need to meet that standard in order to get the official "Compact Disc Digital Audio" logo stamped on it, and CD pressing plants could reserve the right to reject master discs that were not Red Book complaint.

Sony defined a maximum allowable peak modulation with the Scarlet Book specification for SACD. A pressing plant will refuse an SACD master if out of any consecutive 28 samples more than 24 or less than 4 are “1”.

bhtooefr
March 25th, 2016, 12:49 AM
You should be careful about deriving the PAR from only the resolution though, since especially these older systems had overscan areas (borders) of arbitrary width and size.

Which is why I suggested that Microsoft's approach was na´ve, NOT considering any of that, just saying "we're on a 4:3 display, we're outputting 320/640x200, that gives us how much we have to compensate".

Scali
March 25th, 2016, 01:12 AM
Which is why I suggested that Microsoft's approach was na´ve, NOT considering any of that, just saying "we're on a 4:3 display, we're outputting 320/640x200, that gives us how much we have to compensate".

Well, comparing a BASIC circle against an NTSC test card should answer that one :)

Eudimorphodon
March 25th, 2016, 10:47 AM
From what I understood, the 5153 does only one of these at a time, not both.

Doh, you're absolutely right. I looked again at the pictures in the OP, this time pulling them into an image editor and tracing out the rectangles created by the "Even all around" border and the "it looks too tall" border and the one that creates a 1.33 rectangle is the latter (which is the shape necessary for the 5:6 PAR to be correct.). Aspect ratio of the first one is even worse than 1.37, it's more like 1.43. I guess I was just sucked in by how nice it looks that way into thinking that's "correct". (And also I was under the impression that the rounder circle in the test pattern was being displayed with the tighter border, and, well, it looked pretty round to me. Maybe I just can't tell the difference between exactly right and 3% off, while the 15%-ish off 1:1 pixels is obvious.)

(It is notable that you can actually see a *little* bit of the inactive area in the upper left-hand corner of the tighter picture so maybe it should have the dial spun just a tiny bit more, making the "even all around" height more like 1.35-1.40-ish?)


Since BASIC was developed by MS/IBM during design of the 5150 PC, and BASIC was installed in the ROM of every PC, I think there would have been close contact between the software and hardware developers.

So in my opinion, BASIC is probably 'the absolute truth', regardless of how many programs do it differently.
Mind you, BASIC was written a few years before the 5153 was on the market.

Therefore... the 5153 is doing it wrong, or at least its bezel is the wrong shape for looking quite right when it actually doing it right, for the chosen-in-its-ROM PAR corrective ratio of 5:6. This could mean one of two things:

#1: IBM put a great deal of thought into it, decided that there was some compelling reason to make the 5153's aspect ratio different than the 5:6 they chose for BASIC, and then never really told anybody about it (and changed their mind again when they settled on the 5:6 ratio for displaying CGA doublescanned on VGA), or:

#2: When designing CGA IBM just put the number of pixels on the screen that would easily fit into 16k of memory and was an even multiple of their 8x8 character dimensions without really thinking about aspect ratios at all, and the 5:6 number was arrived at by Microsoft as an easy-to-compute approximation that looked fine on your average composite monitor. Later, when the 5153's plastics were being designed, the guy doing that job just made the hole a shape he thought looked good and called it a day. Since WYSIWYG wasn't really a "thing" back then nobody really cared that much that a CGA monitor adjusted to the letter of IBM's instructions (turn the knob until the black outside the borders goes away) made circles a wee bit egg shaped.

Still thinking #2 is substantially (overwhelmingly) more likely, but maybe there was some method to their madness of making that border a little squat for the PAR they settled on later. But I do think I've convinced myself that I ever have a 5150+5153 again I'll go for the stretched borders.

Trixter
March 25th, 2016, 11:31 AM
Still thinking #2 is substantially (overwhelmingly) more likely, but maybe there was some method to their madness of making that border a little squat for the PAR they settled on later.

This is why I've been trying to track down people involved with the project. So far no luck, although I may have a few more people to try. My guess is that the guy responsible for the bezel dimensions messed up.

Eudimorphodon
March 25th, 2016, 12:02 PM
Out of curiosity I tried measuring the bezel aspect ratio for the 5151 from pictures (which is of course subject to some error because a few degrees off from vertical from the plane of the screen could affect the results significantly) and averaging across a couple photos it seems to come out between 1.39 and 1.42, vs. the 1.37 of the 5153 in your photos. (In each case I adjusted the rectangle so it intersected with the furthest point of each edge.) That's close enough that it might be a reasonable theory that the designer just scaled it from the hole in the earlier monitor?