Over voltage - IC damage? 1

[stanayers]

We're developing a system that uses board level computers. After one computer failed an elevated temp test, we discovered that due to a missing feedback resistor, all the 3.3V Vcc rails have been operating near 5V since these were new. Several IC's on the CPU module are powered from the 3.3V bus; the Max Vcc ratings are typically ~3.6V.
We have tens of hours on these boards. I'm surprised they lasted as long as they did.
Our board vendor has offered to replace the main boards (with the regulator defect), but only to re-test the CPU modules. I'm shocked. To me, both boards have been compromised by the Vcc violation, and functional testing at this point can't establish whether their reliability is degraded. Therefore these CPU modules need to be replaced as well as the main boards. Am I on solid ground here? Are there any other issues I need to consider before I respond to them?

Thanks
Stan Ayers

 

[IRstuff]

Voltage overstress is a classic method for accelerating life on integrated circuits. Even though it's only been "tens" of hours, the equivalent life loss may be in the hundreds or thousands of hours. Potential issues:
> weakened junction contacts cause by excessive current, which may eventually short out
> damage to metallization through metal migration, caused by excessive current
> weakened transistor gate oxides caused by the excessive energy electrons getting trapped in the gate oxides

Any or all of which are potential latent defects that can significantly reduce the remaining reliability. Unless one spends lots of bucks to characterize the devices, there's no telling how much degradation in reliability has been incurred.

A story from the woolly days of early PCs. It was found that running processors like the 80286 at 10V instead of 5V could allow for significant overclocking, i.e., running the processors at double their rated specs. A coworker actually implemented that on his computer. Needless to say, after about one month, the coworker, found to his dismay, his processor was foobar.

TTFN
faq731-376

 

[Mrg1]

What's next? If you can, replace them. Heat stress is totally different than burning. If it were the fan they would just replace the fan. I think they should know better. Sounds like an inverter. 3.3 V rail voltage. Nailed all your octal converters. This is new to me, hope it helped. 3.3 v is awfully low for an amplifier, if that is what you are talking about. The processor I think is protected by a fuse or fusible link. The processor only works when needed. So tens of hours aren't bad for a little over heated processor,compared to the thousands of beautiful moments it would have had. Remember that extended warranty.