The THVD150 is one of the new transceivers that have a 50mV built-in fail-safe margin, so they see everything on the differential pair (A/B) that is higher than -50mV as a HIGH. Between -50mV and -200mV is undefined and bellow -200mV is a LOW. These devices see 0V on the pair as a HIGH. The old standard is undefined in the range 200mV to -200mV. Other chips that do this include ISL3172 and MAX3085.
If 0V between A/B is guaranteed to be seen as a HIGH, then why do I need to drive it? I do not, but let's be clear I am not using the RS485 standard.
When a UART TX line is at rest it is a digital HIGH. If I use the digital high to disable the transmitter and all the receivers see 0V as HIGH then communication will work. The possibility of software caused self-destructing cross conduction is eliminated (unfortunately the A/B hardware wires can still be swapped).
The RX line will receive everything placed on the media and it is enabled by default with a pull-up. This can be unexpected depending on previous experience. It is dealt with by software reading back the sent characters. The readback can be used to detect if the A/B pair has a short circuit, or if a collision is occurring (e.g. two or more drivers).
IOFF buffers are also added so any MCU voltage can be used.
This is how the out of band manager communicates between nodes on my projects (e.g. rpubus). Others may find it useful.
My notes (including the parts list) are on Github
The circuit board files also
I have not built (or ordered) a board yet, I need to review things first.