Highly Intermittant Production and Tantalum Caps

Sometimes the acid eats a hole in the side of the can. That takes a long time.

Yeah, wet tants are the old gold standard for reliability. Mil spec -> $$$$. It's been that way for 40 years that I know about.

The repeatable 1/f**3 dependence is interesting--maybe 1/f noise in _leakage_ rather than _voltage_. The capacitance would integrate that into a 1/f**3 dependence of the noise PSD.

Or it could be surface states, traps, program trading, or weakness in the Asian markets. ;)

Cheers

Phil "Friday stock market commentator wannabe" Hobbs

Dodgy, well, government-speak. I take it to mean the full parts design specs (e.g. your BOM, and the qualifications you need for each part, e.g., MIL-PRF-55365/8). The part would have been tested to measure the stresses (e.g. surge current) versus failures. I found this interesting paper (from 2001 though - might be out of date)

- RS

Thanks for your comments and the link. I found a statement that seems to provide the info I was looking for. "However, when the series resistance of a circuit approaches roughly 100-1000 Ω, the self-healing process begins to dominate." So maybe I should use a 1,000 ohm resistor for the current limiting initial power application.

When you talk about 1-2 ohms, I think you are talking about resistance to include in the UUT as part of the application circuit. That circuit is already designed and will not be changed. I am looking for info on the design and operation of a test fixture used in manufacturing to promote self healing after the boards are assembled and the soldering heat has caused more defects in the caps.

So I will recommend they use a 1,000 ohm resistor in the test fixture to limit the initial current and promote self healing using a jumper to bypass it during the functional test.