Posted by Sebastian: Tue Jan 24, 2006 4:47 pm    Post subject: Leakage

Guys,

Need some help...

We have SBGA packages more the 600 I/Os being tested. Usually an O/S failure which is considered Bin5 are caused either by wire shorting or lifted ball/stitch.

Lifted ball/stitch is a clear category and can easily be detected. Lately though we have some reject units failing on B5. No lifted stictch/balls seen on xray. No shorting wires either. The closest we can get is two wires that are very close to each other even less than the diameter of the wires itself. Some units were decapped and no physical shorting of wires seen.

This phenomenon is very puzzling. How can units with no opens or shorts fall on B5? Are wires close to each other (most probably with mold compound in between) create some leakage effect enough to be considered shorting thus fall to B5?

Any input will be highly appreciated.

regards...

Postedby Beentheretoo: Wed Jan 25, 2006 2:57 pm    Post subject:

Hi Sebastian,

Check with the test guys what failure mode the datalog of the failing units is indicating. They should be able to tell you exactly if the failure is due to an open pin, a short, or leakage between pins.

Once you know what the exact failure mode is and which pins are affected, use a curve tracer to confirm the failure. Chances are you won't see the failures at room temp, so you need to do the curve tracing at low and high temp too. Reason for this is that the application of heat or pressure to the package can be enough to shift the features of the parts and make wires touch each other, or bonds lift off their pads.

Only after you've confirmed the affected pins electrically should you focus on visual inspection of the areas surrounding it.

Lastly, some failures are really mysterious, and may remain unsolved despite all the efforts.

Beentheretoo

Posted by Polgi-Wan: Wed Jan 25, 2006 5:33 pm    Post subject:

try to check first with your test (or product) engineer what kind of continuity test is being applied. is it a dynamic opens/shorts test? (wherein you use a test pattern and dynamic loads to force current on I/O pins and pins act as floats or Zs or tristate) or a parametric continuity test approach (wherein each pin is "individually" tested by PMU method or more known as force current/measure voltage per pin).

"ball" pins are more prone to contact resistance since physically not 100% of the "ball area" hits correctly. simply put, like in basketball when the ball appears to have hit the line (lineball) and appeared to be out-of-bounds may not have exactly touched the line. i dont know if that was clear.

basically, BGAs have a tendency to have a high ohmic resistance because of the nature of the contact point. generally, they are simple contact or setup problems.

other tests can be used to reinforce debugging like leakage test, output tests (vol/voh) or some other static tests.

curve tracing is also helpful as mentioned here. by curve trace, you can characterize your pins but oh this is painful and excruciating (600 pins) haha. .

other things to look at:

are the failure (or failures) shown on certain pins? or various?
are the failures consistent? repeatable? intermittent?
are these hard shorts? marginal readings?