6

Many classic computers could optionally use a TV set as the display. At least in the early (pre-SCART) days, the TV would typically only have RF input, but if you were using a monitor, you could get a better color picture, not just because of the quality of the picture tube itself, but because you could use a better transmission format. Composite video would save an RF encoding step, but best of all was when computer and monitor could connect with an RGB cable, avoiding the distortion that comes with mixing the components together and separating them again.

Did some of this apply even with a monochrome system? That is, if you had a monochrome computer providing RF output to a black-and-white TV, and you wanted to instead connect it to a green-screen monitor, could the monitor produce a sharper image if it gets the signal in a different format than the TV could take?

3 Answers 3

13

RF encoding introduces significant noise to the video signal because it moves the signal up to an area of the spectrum that is much more susceptible to interference. A monochrome 240p video signal of the best quality will be defined by its dot clock frequency at ~14 MHz. This would create a good quality 80 column text image as seen on machines like IBM CGA or Amiga. For RF compatibility, such a signal would be promoted to the TV VHF spectrum, which resides in a large and noisy range of ~50 MHz to ~250 MHz. Here, the signal is subject to interference from the many other signals present in the VHF spectrum. Thus, even without any distortion from the frequency shift (not possible), the quality of the signal as is transmitted to the TV is significantly degraded by noisy spectrum interference. Naturally, the TV also has to shift the signal back down to the native frequency as well, and this too is imprecise.

So you have noisy artifacts introduced through both the up and down frequency conversion needed for RF transmissions to the TV, and you have the fact that you are moving the signal into a much noisier spectrum band where it is interfered by other signals.

Therefore, yes, any avoidance of RF encoding, even for a basic text-only monochrome display, will benefit the display quality. Examples of bypassing RF encoding to achieve better results using a monitor are readily seen amongst various early computers, including the 1977 Apple ][, Commodore PET, and TRS-80 Model 1, all of which relied on bypassing RF for more readable text displays on monitors.

In my opinion, TV sets as computer monitors was never meant as a practical alternative to a monitor if you were utilizing anything other than low-resolution color output, probably for games. The fact that early computers supported RF was more for keeping the costs/barriers to entry lower for the target home market, since virtually all homes already had one or more TV sets. It's the same reality that has always applied to home game consoles, where the output capabilities improved almost lock-step with the capabilities of low-cost consumer display devices (e.g. "TV's") to the present day.

6
  • 5
    I suspect the biggest problem with RF encoding is the 6MHz channel bandwidth (8MHz for PAL/SECAM) rather than the noise.
    – Ken Gober
    Jan 4, 2020 at 13:48
  • 1
    @KenGober You are right that there are multiple factors and I've expanded the answer.
    – Brian H
    Jan 4, 2020 at 13:51
  • 1
    Would good cabling minimize the interference? Jan 4, 2020 at 14:04
  • 2
    @ThorbjørnRavnAndersen I know from experience that using thick coaxial cable and barrel connectors works better than cheap antenna wire. But still no substitute for connecting to a monitor/non-RF input.
    – Brian H
    Jan 4, 2020 at 14:06
  • 2
    Pedant attack! SCART was compulsory on new French TV sets from 1980 and therefore increasingly common across Europe throughout the 1980s (economies of scale, etc). A reason the Oric was disproportionately popular in France was that it had a direct RGB output that was easy to connect to a SCART input. So a TV as a monitor output was much more feasible than the deficiencies of RF might imply. Things like dot pitch were more likely to be a hassle.
    – Tommy
    Jan 10, 2020 at 16:21
9

Not answering answering your question, but "black and white" TVs did produce a "better" picture as they didn't have the wire mask behind the glass, which was often very coarse on early TVs and even monitors. You could sometimes/often get a better grey image on a B&W TV than colour on a monitor.

1
  • 7
    Also, many cheap black and white TVs wouldn't try to remove the colour subcarrier, so wouldn't end up softening a genuine black and white signal.
    – Tommy
    Jan 10, 2020 at 16:24
3

A colour CRT display, whether a television or computer monitor, in the end always sends separate R, G and B channels to the picture tube as well as using a sync signal to control the scan. With a source that originates RGB, the quality differences depend on what other conversions the source applies that the display must undo in order to generate the RGB signal for the picture tube.

By far the largest improvement in colour display quality was gained by using RGB instead of a composite colour signal between the source and display. The difference is huge and easily visible if you compare the RGB and CVBS outputs from the same system on a monitor with both RGB and CVBS inputs.

RGB sends three separate analogue¹ luminance signals over three wires. The sync might also have its own one (composite) or two (separate horizontal and vertical) wires or be overlaid on the green signal; this makes no visible difference because the sync is very easy to separate and occurred only in non-display areas (off the edges of the screen) anyway.

Using a CVBS colour signal² caused a huge reduction in quality because the colour information was transformed drastically (to NTSC colour or PAL colour) in order to be able to use a single channel; this was also destructive to the colour information because they had to reduce the overall bandwidth of the signal. (Essentially, the resolution of individual colours is lower than the overall resolution of luminance ("brightness") information.)

(As mentioned in a comment below, there is also a similar form of signal, often called S-video, that uses the same luma and chroma signals but transfers them over two separate wires, rather than mixing the luma and chroma into a single signal down one wire. This increases the resolution of the luma signal back to its original monochrome resolution—luma resolution is otherwise reduced because the chroma signal is using up some of its bandwidth—but the colour resolution remains poor. This produces a small improvement in image quality, but it's nowhere near the improvement you see with separate R, G and B signals.)

A monochrome CVBS signal, on the other hand, contains only luminance information for a single channel, and so is electronically exactly the same as a single channel of the RGB connection with sync, such as the sync-on-green situation I described above. So with a monitor, so long as it is not trying to filter out parts of the signal that would be used for colour in a CVBS colour system, will display a monochrome signal just clearly as an RGB colour signal. In fact, you can plug a monochrome CVBS signal into the green channel input of an RGB monitor and it will display just fine. (This assumes that the monitor supports sync-on-green; if not, sending the monochrome CVBS signal to any or all of the RGB input channels and the composite sync input will usually work.)

The transformation from a baseband CVBS signal to that signal modulated onto another signal in the a radio frequency (RF) range will, and its reversal at the display end, can be the source of a slight degradation in quality, but except in pathological circumstances this is nowhere near the level of degradation introduced by RGB/CVBS/RGB conversion. In good circumstances it will be barely noticeable. In less-than-great circumstances you might see more degradation of a CVBS colour signal than of a monochrome signal, which is rather the opposite of what you're thinking of.

The signal quality differences I'm describing above ones that I've seen and A/B tested myself on various computers that produce both RGB and monochrome CVBS (e.g., the Fujitu FM-7), RGB and colour CVBS (e.g., the National/Panasonic JR-200), and CVBS and separate chroma/luma (e.g., the Commodore 64). The tests were done on a professional video monitor with inputs for all of the above, a Sony PVM-9045Q. This is a late-'90s PVM which cost about $2000 when first introduced and has very high quality analogue decoding circuitry for CVBS/NTSC/PAL/etc.


¹ Digital RGB signals were also frequently used, but from the picture quality point of view these were essentially the same as analogue signals; the only thing that changed was that the monitor would do a trivial 1-bit digital to analog conversion. Essentially, this just converted a 0-1.8 V digital input to 0 V analogue, and 2.4-5.2 V digital signal input to ~1.0 V analogue.)

² This was also sometimes called a "composite colour" signal; I avoid using the word "composite" here because it may refer to combining all three colour signals together into a single signal or, even in a monochrome system, combining luminance information with sync into a single signal.

3
  • 1
    Separation of luma and chroma (as was done on e.g. the Commodore 1702 monitor) offered a huge improvement. Elimination of chroma modulation was a definite further improvement, though I don't know that I'd say it was bigger than the change going from single-lead composite to two-lead luma+chroma. Another huge change is the use of a good quality RF modulator with a TV, as compared with the crummy ones usually bundled with computers.
    – supercat
    Jul 29 at 17:44
  • @supercat I've A/B tested CVBS versus S-video, too, and it it's not hugely better; the improvement is nowhere near that of using RGB. Most of the degradation comes from the conversion of three separate luma signals to a luma and a chroma signal, and that's the same whether your then transfer the luma and chroma separately (S-video) or further mix those together (CVBS colour). But thanks for mentioning this, I've updated my answer.
    – cjs
    Aug 12 at 1:29
  • Many computers have a pixel clock which is either near to being 2x chroma (e.g. the Commodore 64) or exactly 2x chroma (e.g. Atari 8-bit machines and the Apple II). An alternating pattern of light and dark pixels will be perceived as color if sent through a composite cable, but not if sent via S-video. For an Apple II, such modulation is the only way of achieving color, so but on the Commodore 64 it's a distracting and annoying artifact, especially on older versions of the VIC-II chip.
    – supercat
    Aug 12 at 6:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.