If you read the very end of that document, it was a sales pitch from the NuCore guys to Stern for developing a new system. With Stern's history of system development it's not surprising they took the idea and ran with it on their own.

I'd be interested to know what the failure points are on the dead node board. I wonder if the boards are failing for vibration issues or electro-migration. Those low profile LCC's are notorious for both of those problems and under a playfield is going to expose any marginal solder joints. Boards are too clean, so I doubt they're using a high reliability no-clean based paste, and there's definitely no under-fill or potting to mitigate vibration issues. If they're using organic flux on a board this dense I'm sure there's still some corrosive residues under the low profile parts. Be interesting to see what you'd find if you sent one out for an ion chromatography test.

Looking at the design, most board houses can kick them out the door. But there's a lot going on, bargain and overseas suppliers would struggle building these to be reliable. Particularly if Stern isn't asking for certain build and test requirements, either out of budget or experience reasons. Given the failure rates I'd be running these through either a thermal chamber or hot-room burn-in at least until the failure point are found and addressed.

-Hans

I don't know exact or even general numbers. Just going by the anecdotal evidence here. But let me put it another way, one failure per thousand boards is considered unacceptably high by the automotive customers at my full time job.