The colors on my rev 0 ][ are showing an interesting issue. Compared to the colors output by my ][+, they're a little off - greens look like yellow-greens, blues look like green-blues, purples look like light-purples. I've tried twirling the color trim control every which way but I can't get the colors to exactly match those of the ][+.
Is this a quirk of the rev 0 board or is there perhaps something else I should be looking at?
Can I assume that you are checking low res colors? Given the color generation design is the same on both boards for low res color, there should be no significant difference. Does the color trim cap actually change the color for both boards and is the level adjustment (200 ohm pot) set to the same levels for both boards.
There are some differences in vertical and horizontal refresh signals, between boards, but I don't think that these should affect the color pallet.
To verify, I just compared an old RFI board with a Rev 0 replica with the same monitor. Turns out the colors can be adjusted very close to each other. In particular light blue and yellow, seem to be most sensitive on both these boards, with light blue tending towards purple and yellow towards orange on this particular monitor. If anything the Rev 0 replica can be adjusted to a slightly more pleasing pallet, but the difference is minimal, at best.
Keep in mind how TV engineers have described NTSC: Never The Same Color, twice.
Regards,
Mike Willegal
Yeah, I was checking low-res. I can confirm that the color-trim control does actually change the color, and the level adjustment is the same (as far as I can tell).
Here are a couple of photos for comparison, using the same TV:
Apple ][ (rev 0)
Apple ][+ (RFI)
As you can see, the brown is much more reddish-brown on the ][, and the blue is a little more blue-green (I know it doesn't come across very well in the photo, but trust me, it's noticeable). This is as close as I was able to get them after fussing with the color trim.
The only "control" I have available is the AppleWin emulator, which agrees with the ][+ as far as brown goes, but is off in its own world with respect to blue:
Well, I "resolved" the issue without really figuring out root cause. For whatever reason, the ][ is putting out either too much yellow or not enough green. If I adjust the hue on the TV accordingly, I can get it to match the ][+'s output exactly.
It's an okay workaround, but ultimately unsatisfactory, since I'd really rather know why the heck supposedly identical color-generation hardware is showing this kind of discrepancy.
The RCA jack on the back of the II, II+, IIe, and IIc sends out an NTSC signal which your TV then decodes and displays. NTSC color calibration is, to put it mildly, not very reliable.
In the biz, engineers joke that NTSC stands for "Never Twice the Same Color."
The old Soviet system, SECAM, stood for "System Essentially Contrary to American Methods," according to the same engineers.
(For those who want to know, NTSC actually stands for "National Television System Committee." But I like the other explanation better.)
Correct me if I'm wrong, but doesn't NTSC's color issue have mainly to do with sensitivity to reception problems? I understand that the color signal can be altered by that, but this is just a very simple setup with a short RCA cable (I've tried switching cables as well). Wouldn't I be seeing issues from the ][+ as well if it were an NTSC signal problem?
I can't readily find a copy of the Apple ][ schematics online at the moment, but...
NTSC's method of encoding color is complicated. In essence, the "high resolution" pixel data of an NTSC picture is all black and white "luminance" data, which is "colorized" by information carried as a separate signal which has to be "filtered" in and out of the actual transmission. This compares to a computer attached to an RGB monitor, which instead essentially sends three separate equally-high-resolution frames of each color simultaneously. The process of encoding the color signal in NTSC is prone to errors, as small differences in input voltages and input frequency will produce relatively large difference in the colors produced at the other end. Decoding comes with its own issues, among which is the fact that information from the luminance signal can sometimes be erroneously interpreted by the color filter, creating phantom color changes. (The psychedelic colors you'll see when watching a television program featuring a person wearing a pinstriped suit or tie are the classic example of this.)
If I had to take a wild guess, knowing what little I do about how NTSC Color Encoding works, the Apple ][ probably uses a few resistors in front of the TTL logic as a simple D/A converter to fiddle with the phase and amplitude of the color signal. Inexpensive resistors tend to have tolerances of 5% or more, and said tolerances can change as they age. I'd be really surprised if any two *brand new* Apple ][s produced *exactly* the same colors. You could also potentially be seeing some aging differences in the amplification portions of the video circuits.
In short, I wouldn't worry about it. If it *really* bothers you could always rip out all the resistors and other analog components in both machines and replace them with new ones sourced from the same production run. That would probably even the colors out.
--Peace
works by "tricking" the color TV or monitor to output a color based on the interaction between the color carrier and the binary bitstream output by the Apple. It's the same interaction when the newscaster wears a striped black and white tie or suit, and you get a burst of color as he moves. Woz was clever in his design, and it seemed to work very well across TVs of the day.
Maybe Mike can hook you up with a newer color trim cap. Another thought - can you hook up to a high resolution monochrome monitor and compare the outputs? Any difference in noise or brightness?
Dave...
If you ask me, from your pictures the color looks better on the rev0 board, than the RFI board, which looks washed out.
The following assumes you have schematics and know how to read them.
There was a good point here about the variable values of the discrete components. You should be able to measure the resistors in circuit with an ohm meter to get an idea of relative values between the two boards. Compare the three placed in a row near the edge of the board. The are marked 2.7K, 2.0K and 1.5K. These set the relative levels of different components of the video signal. 2.7K for color burst: 1.5K for video signal itself: 2k for blanking. I don't think the other two resistors in the video circuit should have an effect on the intensity of the color, assuming that the 200 ohm pot changes intensity of the signal correctly. They might be worth checking anyway, though.
The hue is controlled by the variable cap, a 27uH inductor and a 47pF cap. The 47pF cap is a good quality Mica one and should be stable and pretty accurate. If the variable cap has an effect on the color, it is probably working ok. You could try measuring inductance of the color burst signal connected to the 2.7K resistor, if you had a meter with this capability. Be sure to tweak the variable cap and see if it has an effect on the inductance reading. I do have some extra replacements for the variable cap, and they are also available from Jameco (part #32855).
Regards,
Mike Willegal
I couldn't remember if the Apple II used "real" NTSC color encoding or what we who grew up mostly with the TRS-80 Color Computer and IBM's CGA adapter used to call "Artifact Colors" in the low-resolution modes or not. (The Apple II's *only* method of generating color works "accidentally" on other machines when they're displaying dot resolutions too high to be properly resolved by NTSC monitors. Which depending on the situation was either a blessing or a curse.)
So, yeah, I stand corrected on precisely what to look for.
--Peace
P.S. Interesting reading: The US Patent describing how the Apple II's color system works. This bit is notable:
"In many prior art microcomputer controlled displays, color information is stored as four digital bits which are used to designate green, red, blue, and high/low intensity. The color generation means generally includes a signal generator for
generating the pure color signals (CW). These pure color signals are then gated and mixed in accordance with the binary state of the four bits to provide a color signal compatible with standard television receivers. Generation of the video color signal in this manner is complex and requires a substantial amount of circuitry.
That statement about requiring a "substantial amount of circuitry" may of been true when the Apple II was being cobbled together, but when you jump forward only a couple years you'll find that most 8-bit computers were using a single-chip video display generator to do it more cheaply then the Apple's shortcut solution. It's scary just how quickly technology marched by back in the early days.