so letsme guess, it set up the bus in output mode, screen 8, then it reads from the 9938 bus the BBBGGGRR value and transforms it to a bigger value selected from some palette pool.
Probably, because other screens supported by MSX2 can have their palette defined (also in MSX-BASIC), but it's not the case of screen 8. Maybe this extra chip comes with additionnal instructions, as it has its own register(s).
The MSX3 was scheduled for market in 1990. Delays in the development of its VDP—then named V9978 on the pre-release spec sheets—caused Yamaha to miss its time to market deadline.[13] In its place, an improved MSX2+ was released as the MSX Turbo-R; features of the new R800 processor such as DMA and 24-bit addressing were disabled. The VDP was eventually delivered two years after its planned deadline, by which time the market had moved on. In an attempt to reduce its financial loss, Yamaha stripped nearly all V9958 compatibility and marketed the resulting V9990 E-VDP III as a video-chipset for PC VGA graphic cards, with moderate success. Sony also employed the V7040 RGB encoder chip on many other products. MSX-FAN Magazine also mentions the then-impressive power of the V9990, being able to compete with much more expensive hardware such as the Sharp X68000.
There's nothing special about the V7040. It's just the output stage of the "MSX VIDEO" chipset (full superimpose+digitizer, as seen on the NMS-8280). Because it's a two input RGB to CVBS encoder it can be easily used on other devices. Think of it as a CXA1145 (RGB to composite encoder) which can mix two RGB sources into one (for the superimpose feature of the MSX VIDEO chipset).
