Two pictures of RADs, from their respective reference manuals. The 7212 is the high-speed swapping RAD, and the 7204 is the "normal" RAD. The pictures really don't show much; unless you tore one apart you couldn't see anything of the head mechanism.
Many people, at UW and elsewhere, cherished the ownership of a RAD platter for a momento, especially a crashed one. However the only person I actually know of who got one was Shel Klee (pronounced "clay") who worked on CP-V from the early days through the conversion to CP-6 by Honeywell. Upon leaving Honeywell, he was presented with a framed RAD platter. (per a Honeywell newsletter and photo)
When a RAD had a crash, you'd normally lose an entire surface at a minimum. The first clue was that you got lots of errors. The second clue was looking in through the plexiglass window in the bulkhead and noticing all the tiny bits of heads and wire on the floor of the sealed unit. There was only a 25% chance that the crashed surface was the one visible in the window, and I don't recall us ever being that "lucky".
Aside from SDS/XDS, there was a company in Israel that would rebuild crashed RADs. We did business with them on several occasions.