Apple II timing vs. NTSC (was Re: Good Composite->VGA converters for classic computers (& video games...)

I wrote: Jim wrote: NTSC requires that the horizontal rate be 227.5 times the color subcarrier frequency, and phase-locked to it. The color carrier thus alternates in phase 180 degrees on consecutive lines within a field. The Apple II horizontal rate is 288.0 times the color subcarrier frequency, so that the color phase is the same on every line. This is also why every 65th cycle of the processor clock is longer than the other 64. The Apple timing is derived from a 14.31818 MHz oscillator, four times the color burst frequency. For NTSC there should be 910 cycles per scan line (227.5 * 4). The CPU clock is normally the oscillator frequency divided by 14, to get 1.023 MHz. However, if that timing was used consistently, the color phase would alternate as per the NTSC spec. Instead, every 65th processor cycle is divided by 16, rather than 14, so that there are a total of 912 oscillator cycles per scan line, to get the constant color phase. This makes the average CPU clock slightly lower, at 1.020 MHz. (14.31318 MHz * 65) / (64 * 14 + 16) This caused terrible problems for anyone that wanted to use an Apple II as a source for video broadcast; the signal couldn't be used even with most genlocks. I worked in a studio that used an Apple II with a VB3 Microkeyer from Video Associate Labs. This was a two-board set that installed in the Apple II to provide a genlock and proc amp. It gave a 100% compliant video signal by replacing a fair bit of the Apple II circuitry. You had to pull about a dozen TTL chips out of the Apple and replace them with ribbon cables to the two cards. One card went into a slot, and the other sat above the power supply. Naturally this wouldn't work on an Apple IIe. Many years later Apple introduced their own video overlay card. Rather than modifying the circuitry of the base Apple, it completely replicated the video circuitry on the card, but provided standard NTSC timing. Eric