plane crashes: how often does the CVR/FDR fail?

In commercial aircraft, the cockpit voice recorder and flight data recorder are designed to survive violent crashes and post-crash fires. But surely they are not designed to survive every conceivable crash.

Have their been crashes in which the data captured by the recorders was corrupted/lost because of damage sustained during/after the crash?

Curious because footage from the recent GermanWings crash site shows that the aircraft appears to have been obliterated by a very high-speed impact with terrain, and a picture in the media shows at least one of the recorders having sustained significant damage (although it sounds like they are able to recover the data in this particular case).

Here’s one instance:

I don’t know percentages of how often they survive or how often they don’t. I do know that the black boxes are rated for a certain temperature for a certain period of time, and some plane fires will exceed that.

One very well known example is American Airlines flight 77 from the 9/11 attacks. While the FDR managed to survive both the crash and the fire afterwards (at least enough to give useful data), the CVR tape was completely destroyed in the fire.

Modern units are all solid state. Some old tape-based CVRs are still in use.

Assuming the recorders are found, I’d WAG that good data comes out of 95+% of them. Some recorders are found to have failed some time before the mishap. Mishaps which include significant electrical system failures before impact may result in incomplete or totally absent recordings after the problems start.

And as engineer_comp_geek said, there is an upper limit to how big & hot a fire they can survive. With solid state recording media the physical impact forces are usually not the limiting factor. The circuit cards may be junk, but the chips can almost always be pulled and inserted into a fresh board for reading.