RP2350 datasheet: HSTX_CTRL.EXPAND_TMDS register poorly documented regarding unused bits

[WolfWings]

The documentation doesn't specify if the unused data input bits are zero-filled or bit-repeated from the valid bits.

Experimenting I confirmed they are zero-filled; this prevents various otherwise useful configurations as lower-precision settings have reduced (sometimes greatly) value ranges.

This is both why 1bpp modes look quite dim because they're only displaying 0x80 not 0xFF with even 2bpp greyscale modes jumping to 0xC0 as a maximum value, and the source of the 'tinting' that RGB332 and RGB565 modes end up reported with because 'white' is not 0xFFFFFF but 0xE0E0C0 (RGB332) or 0xF8FCF8 (RGB565).

Documenting that the unused bits are zero-filled versus bit-repeated from the valid bits would be appreciated in the datasheet if possible.

[WolfWings]

As a minor addendum to this the TMDS_CTRL register documents this effect with good phrasing for the L0_NBITS/L1_NBITS/L2_NBITS:

Remaining LSBs are masked to 0 after the rotate.

Making the phrasing match between the TMDS_CTRL fields and the EXPAND_TMDS fields would be the best resolution for this I believe.