The main mistake Motorola made is the level of options and ambiguity in the first definition (I think v0.7 was the first one non-Motorola people got to see?), which made people feel they could just do what the fuck ever.
Actually don't know if that link is the standard, tbh, I just threw the first thing linked to Motorola original development at you without looking, because I'm tired
Anyway, the choices you have in SPI are: MSB first vs LSB first, Mode1 vs 2 vs 3 vs 4, Selected vs Always Listening
Later the single data-line got added, not sure if it's by Motorola itself
Yet, Semtech, for example, decided that when select is not asserted (not low) that the output pin would just repeat the input pin. By design. On purpose.
There's a few Motorola standards that are a bit iffy to find now, because when they became Freescale they forgot to update a few links and then when they became NXP the new mother company wanted everything in theme and only spent money on about half of that effort.
And I sometimes think it's because they already owned the Philips inherited I2C that SPI fell on the wrong side of that line
After 4 weeks of debugging code I discovered that my bus of 10 devices was always just effectively outputting the input back at me, because the one selected chip never won from the 9 not-selected.
So we contacted them with this
And they responded "Yeh, that's expected behaviour"
@ThePhoton Thats not an engineering diagram, it's art
Great realcomponents just bought out all of the world market for an EOL capacitor, and they want 10x the price for it. I love those parts pirates, I mean brokers.
@AdamUraynar You were talking about detents, so I figured you would be OK with mechanical. You can always spin your own with a 2 output photo-transistor gate, which runs about $2