In a lot of EVs, when the car is "on" there is a DC/DC converter powering the 12V system from the HV battery. So if the car was "on" and experienced a loss of the 12V battery it could continue operating for some period of time.
I think the implication is that people could be driving with a dead 12V as if nothing is wrong. Then when they experience the HV failure in the recall, they would have no power whatsoever for safety systems.
That wasn't intended to be my implication. The replies to the original comment answered my question. Generally if you have a dead 12V you know it because you need a boost to get started, and you don't drive like that for long. I'm not worried about a cascade failure, I was worried about the 12V dying while somebody was driving and then having no brakes. That concern has been addressed.
When the 12v battery is dying, at least in a Tesla, it warns you, and starts disabling certain features, more likely to help preserve the battery. For example, heated seats run off 12v. It makes sense that a lot of commodity auto industry parts would run off 12v because the supply chain is there, and because you want low voltage in the cabin anyway.
Like another poster in this thread, my original model 3 battery went ~5 years (typical 12v failure age in a car), and I bought it for $89(!!) at Tesla. Autozone wanted $125 for the same group battery. I did a DIY replacement. For some reason, that one failed after a year and a half. Just bad luck I guess.
Yes, this drives a requirement for latent fault detection of LV battery faults. In general when a safety function is decomposed to provide statistical safety, there is a requirement to time-bound a single failure, since without that the decomposition doesn’t buy you anything. Latent fault detection is the standard option for time bounding for automotive — for aviation, you have a second escape hatch that issues that are reliably found during annual / 100 hour inspection and can be safely missed for that long can be caught by inspection instead.
