A little late to the party, but I’d like to point out that you will find very little documentation on 240p, because analog 240p isn’t actually a standard - it’s more of a happy accident caused by the way CRTs deal with non-standard input.
The NTSC standard defines an interlaced video signal wherein each field consists of 262.5 scanlines (including vertical blanking), with every other field starting with a half scanline (rather than a full one) which provides the required vertical offset to interlace alternate fields.
Standard-definition consoles don’t do any of that - they make certain adjustments to make video signal generation simpler and thus cheaper to implement (often with further adjustments in the interest of visual quality). In particular, each frame is only 262 scanlines (including vertical blanking), and they don’t bother with the half scanlines either. The lack of half scanlines means that there is no vertical offset every other field, resulting in progressive scan. The reduced scanline count means that they don’t actually display images at 60 Hz or 59.94 Hz, but in fact slightly over 60 Hz.
Then there’s the issue of vertical edge artifacts, which are caused by the interaction between the color data signal and the duration of scanlines and frames/fields. Color data is carried on a signal called the color subcarrier (I’ll call it Chroma for convenience), which runs at 3.579545 MHz - the amplitude of this signal represents the color’s saturation, and the phase shift represents the hue. According to the NTSC standard, one scanline is exactly 227.5 Chroma cycles in length, and therefore each field is 59718.75 Chroma cycles in length. Because neither of these values are integers, vertical edges between different hues aren’t perfect. The non-integral scanline/Chroma ratio means that the edge falls on a different part of a Chroma cycle from scanline to scanline, causing a pattern of hue artifacts. The non-integral field/Chroma ratio means that these artifacts appear to move up or down over time. Since consoles use non-standard timing, however, they have different vertical edge artifacts than standard 480i signals.
On the NES and SNES, for example, on scanline is actually 227.33 subcarrier cycles, and one frame (262 scanlines) is therefore 59561.33 subcarrier cycles. As a further complication, every other frame, one scanline of vertical blanking is shortened to 226.67 cycles - thus odd frames are 59561.33 cycles, even frames are 59560.67, and they average out to 59561 cycles per frame. Because of this, they don’t actually exhibit dot crawl per se - the visual artifacts around vertical edges appear to oscillate instead of moving up or down - and their video output actually runs at 60.10 FPS. The minor differences in frame timing parameters is one of the major reasons for the wide difference in composite video quality between different consoles (and on some consoles, like the TG-16, that allow software to adjust these parameters, the difference from one game to another).
You may also have noticed that the color subcarrier frequency (~227.5 cycles per scanline) is relatively low compared to the number of pixels drawn on a scan line (generally 256 to 512). In fact, of the NES (and other consoles with a 256-pixel horizontal resolution) each pixel is only 2/3 of a subcarrier cycle, on the Genesis (running in its standard 320x240) each pixel is only 8/15 of a cycle, and in the SNES and TG-16 hi-res modes (512 pixels horizontal) there are three pixels in each subcarrier cycle. This is why, when using composite out or RF out on these consoles, pixels tend to bleed into their horizontal neighbors, and also why the Genesis in particular tends to look like ass over composite/RF connectors.
RGB connectors, of course, don’t have these problems, because they don’t have to deal with cramming three analog signals into one, which is why RGB monitors were the standard in arcades and why consoles look best with RGB output - a notable exception being the NES. The NES video chip generates a composite video signal with no RGB at any step along the way - and actually generates some colors that are outside the RGB color space - and several developers, notably Sunsoft, got quite adept at making art tailored to the graphical quirks of the NES.
All of this, of course, is made possible by the fact that the NTSC standard requires that TV receivers accept signals that deviate by as much as 1% from the specified values. (Which was a very strict tolerance in those pre-solid-state days, but 2 decades later was oh-so-exploitable.)
